Research article

A robust identification method for stochastic nonlinear parameter varying systems

  • Received: 24 February 2021 Accepted: 09 March 2021 Published: 17 March 2021
  • Successful identification procedures are undoubtedly important for accurate model description and the consequent implementation of control strategies. Linear Parameter Varying (LPV) models are nowadays standard for control design purposes and powerful identification techniques accordingly available. Anyhow, recent advances have brought to focus the class of Nonlinear Parameter Varying (NLPV) models, which keep some nonlinearities embedded to the formulation. Identification tools for this latter class are still not available. Therefore, this paper proposes a novel method for the robust identification of stochastic NLPV systems, considering that the nonlinear parameter part is a priori known and obeys a Lipschitz condition. The method is based on a modified extended Masreliez-Martin filter and yields the joint estimation of both NLPV systems states and model parameters. The method manages the stochasticity of the system by considering the presence of measurement outliers with non-Gaussian distributions. Results considering real data from a vehicle suspension system are presented in order to demonstrate the consistency of the proposed method.

    Citation: Marcelo Menezes Morato, Vladimir Stojanovic. A robust identification method for stochastic nonlinear parameter varying systems[J]. Mathematical Modelling and Control, 2021, 1(1): 35-51. doi: 10.3934/mmc.2021004

    Related Papers:

    [1] Yaxin Zhao, Xiuli Wang . Multiple robust estimation of parameters in varying-coefficient partially linear model with response missing at random. Mathematical Modelling and Control, 2022, 2(1): 24-33. doi: 10.3934/mmc.2022004
    [2] Biresh Kumar Dakua, Bibhuti Bhusan Pati . A frequency domain-based loop shaping procedure for the parameter estimation of the fractional-order tilt integral derivative controller. Mathematical Modelling and Control, 2024, 4(4): 374-389. doi: 10.3934/mmc.2024030
    [3] Xingwen Liu, Shouming Zhong . Stability analysis of delayed switched cascade nonlinear systems with uniform switching signals. Mathematical Modelling and Control, 2021, 1(2): 90-101. doi: 10.3934/mmc.2021007
    [4] Hongyu Ma, Dadong Tian, Mei Li, Chao Zhang . Reachable set estimation for 2-D switched nonlinear positive systems with impulsive effects and bounded disturbances described by the Roesser model. Mathematical Modelling and Control, 2024, 4(2): 152-162. doi: 10.3934/mmc.2024014
    [5] Hongli Lyu, Yanan Lyu, Yongchao Gao, Heng Qian, Shan Du . MIMO fuzzy adaptive control systems based on fuzzy semi-tensor product. Mathematical Modelling and Control, 2023, 3(4): 316-330. doi: 10.3934/mmc.2023026
    [6] Zhaoxia Duan, Jinling Liang, Zhengrong Xiang . Filter design for continuous-discrete Takagi-Sugeno fuzzy system with finite frequency specifications. Mathematical Modelling and Control, 2023, 3(4): 387-399. doi: 10.3934/mmc.2023031
    [7] Xiaoyu Ren, Ting Hou . Pareto optimal filter design with hybrid $ H_{2} /H_{\infty} $ optimization. Mathematical Modelling and Control, 2023, 3(2): 80-87. doi: 10.3934/mmc.2023008
    [8] Lusong Ding, Weiwei Sun . Neuro-adaptive finite-time control of fractional-order nonlinear systems with multiple objective constraints. Mathematical Modelling and Control, 2023, 3(4): 355-369. doi: 10.3934/mmc.2023029
    [9] Bharatkumar K. Manvi, Shravankumar B. Kerur, Jagadish V Tawade, Juan J. Nieto, Sagar Ningonda Sankeshwari, Hijaz Ahmad, Vediyappan Govindan . MHD Casson nanofluid boundary layer flow in presence of radiation and non-uniform heat source/sink. Mathematical Modelling and Control, 2023, 3(3): 152-167. doi: 10.3934/mmc.2023014
    [10] Gani Stamov, Ekaterina Gospodinova, Ivanka Stamova . Practical exponential stability with respect to $ h- $manifolds of discontinuous delayed Cohen–Grossberg neural networks with variable impulsive perturbations. Mathematical Modelling and Control, 2021, 1(1): 26-34. doi: 10.3934/mmc.2021003
  • Successful identification procedures are undoubtedly important for accurate model description and the consequent implementation of control strategies. Linear Parameter Varying (LPV) models are nowadays standard for control design purposes and powerful identification techniques accordingly available. Anyhow, recent advances have brought to focus the class of Nonlinear Parameter Varying (NLPV) models, which keep some nonlinearities embedded to the formulation. Identification tools for this latter class are still not available. Therefore, this paper proposes a novel method for the robust identification of stochastic NLPV systems, considering that the nonlinear parameter part is a priori known and obeys a Lipschitz condition. The method is based on a modified extended Masreliez-Martin filter and yields the joint estimation of both NLPV systems states and model parameters. The method manages the stochasticity of the system by considering the presence of measurement outliers with non-Gaussian distributions. Results considering real data from a vehicle suspension system are presented in order to demonstrate the consistency of the proposed method.



    Linear Parameter Varying (LPV) models have become progressively popular over the last two decades [20,28,43]. LPV models retain the design synthesis advantages of the Linear Time Invariant (LTI) setting, while being able to represent nonlinear dynamics with truthfulness. The LPV framework has been applied for control [37], observation [29] and identification [3,5] purposes. In this paper, we address the two latter topics.

    Recent works have shown the particular interest of the class of Nonlinear Parameter Varying (NLPV) models [6,33]. In contrast to the "regular" LPV paradigm, which encapsulates the nonlinearities into time-varying scheduling parameters ρ, NLPV models retain some explicit nonlinearities that are "easily handled" outside of the scheduling term ρ, such as local Lipschitz nonlinear terms. On one hand, the LPV class exhibits linearity w.r.t. the state space, on the other hand, the NLPV class is nonlinear on both parameter and state spaces. The main interest in using the explicit nonlinearities coupled to the LPV structure is that they can be taken into account by the control design method, reducing the over-bounding of the nonlinearities by the scheduling parameter and, thus, enabling less conservative control performances. Anyhow, the development of identification and state estimation methods specifically conceived for NLPV models is yet unseen.

    Over the past few decades, particular attention has been devoted to nonlinear system identification under non-Gaussian measurement noises[8,9,36,40]. These stochastic tools are generic and could be applied to the case of NLPV models, yet they could benefit from the availability of the scheduling parameter data if properly designed.

    Motivated by this research gap, the aim of this paper is to extend and generalise the Masreliez-Martin filter [40] joint state and parameter estimation method for the class of NLPV models with explicit Lipchitz nonlinearities. We stress that the main interest regarding the considered method is that it offers a simple recursive implementation and convergence properties verifiable through an LMI problem. The main novelty is that, by exploiting the local Lipschitz condition of the NLPV model, better state and parameter estimates are obtained. Furthermore, the use of the scheduling parameter data also enhances the results, since more information regarding the system dynamics is available. Accordingly, the major contributions are:

    ● The proposition of a robust identification procedure for joint parameter and state estimation for the class of NLPV models. Assuming that the explicit nonlinear term is structurally known and Lipschitz, the algorithm is is based on an adaptation of the modified Masreliez-Martin filter. The method considers coloured noise upon the measurement data and a random walk behaviour for the parameter variations.

    ● A sufficient LMI condition is provided for the verification of the local convergence of the proposed parameter and state estimation tool. This LMI remedy is derived from the direct Lyapunov method applied over an Ordinary Differential Equation (ODE) associated to the estimation error. The solution of the LMI also serves as an estimate for the zone of attraction of the algorithm. We also discuss how the the Lipschitz constant directly affects the size of this basin of attraction.

    ● The application of the proposed NLPV identification procedure to a mechatronic test-rig of an automotive semi-active suspension system. The effectiveness of the proposed method is discussed in terms of state estimation error and parameter identification precision, measured through standard indexes.

    Remark 1. We stress that in the previous works by the Authors [40], the joint parameter-state estimation of a nonlinear stochastic system is considered and verified through simulations on a mathematical model. In this paper, the joint estimation algorithm of a stochastic NLPV system is considered, for which the nonlinear part is explicit, a priori known, and obeys a local Lipschitz condition. The proposed algorithm is verified through experimental results on a real vehicle system. Due to LPV parameter-dependent model matrices and to the nonlinear Lipschitz condition of the explicit nonlinear term, the proposed algorithm is completely different from the prior [40] in terms of matrix structures, gains and recursive estimation laws, but also conceptually, exploiting the availability of scheduling parameters to refine the joint estimation results.

    The rest of this paper is organised as follows. The background state-of-the-art regarding nonlinear and LPV identification is discussed is Section 2, where the formal problem setup is also given. The proposed NLPV state and parameter estimation algorithm is presented in Section 3. The LMI problem used to verify the local asymptotic convergence of the method is also presented therein. The experimental validation results are given in Section 4. This paper ends with conclusions in Section 5.

    In the sequel, the set of nonnegative real number is denoted by R+, whist the set of nonnegative integers including zero is denoted by N. Rq represents a q-dimensional real space. The index set N[a,b] represents {iN|aib}, with 0ab. The identity matrix of size j is denoted as Ij; col{a,b,c} denotes the vectorization (collection) of the entries and diag{v} denotes the diagonal matrix generated with the line vector v.

    System identification [2] is a topic of great importance for coherent control synthesis and implementation. The vast majority of control strategies are model-driven, which means that their success relies on the accurate knowledge of the model parameters. Accordingly, the issue of system identification under stochastic process signals deserves special attention [10,27,39], since real processes are not purely deterministic and stochastic noises are always present.

    Besides the identification problem regarding model parameters, state estimation is a crucial aspect for the control of processes modelled through state-space representations, for which only output data is available. The classic state-feedback design requires online information of the state variables, which can be provided by observers/estimators. Due to these reasons, we note that joint state and parameter estimation is very useful and much facilitates the operation of adaptive and fault-tolerant state-feedback methods, since state observation and parameter variation estimates are provided by a single operator.

    The Extended Kalman Filter (EKF) is a very well established method [25] for the joint estimation of model parameters and system states, as seen in different applications [17,47,49,50]. The EKF is popular for many reasons, but especially because it is not a complex algorithm, it has a recursive implementation law, and ensures asymptotic convergence. Its main drawback is that it is based on the assumption that the measurement noise has a Gaussian distribution. Many empirical and practical essays have shown that this assumption is rather weak [40]. It is a fact that measurement data from real processes may present inconsistent values at some points, i.e. "outliers". These corruptions are typically non-Gaussian and significantly reduce the effectiveness of algorithms such as the EKF. Research has progressively focused on overcoming the EKF restrictions; some of the recent approaches are: maximum likelihood methods [24,45], prediction error techniques [42] and robust identification algorithms [40]. These are all computationally less expensive than the EKF and require less restrictive assumptions on the measurement data.

    The known fact is that the measurements have inconsistent observations with the largest part of the observation population. Justification of the proposed approach, which considers the presence of outliers, was confirmed in practice [21,32]. The presence of outliers can destroy the good features of linearly recursive algorithms, which are typically designed for the estimation in the presence of Gaussian noises. Therefore, it is very important to design a robust algorithm which is insensitive to outliers, since they exist in all real instrumented systems.

    We note that many of the popular identification methods, such as the KF and the EKF, have been extended for the case of nonlinear [11] and LPV models [4]. For the latter class of models, the known parameter-varying data is used to schedule the estimation problem, which has input-output/input-state linearity. The state-of-the-art regarding LPV identification includes subspace techniques [12,15], successive LTI approximations [13], robust filtering [5], multi-step canonical variate analysis [22,35], prediction error methods [51] and nonlinear programs [23]. The review book [14] thoroughly discusses these recent developments and the future trends on LPV identification. A clear investigation gap is to consider NLPV models, for which there exists an explicit nonlinear function which must be taken into account by the estimation procedure, since input-output/input-state linearity no longer holds. Up to the Authors' knowledge, there has been no technique specifically devoted for the class of NLPV models, albeit the fact that many of NL identification methods can be applied to the NLPV case. The focus of this paper is given w.r.t. to this matter, pursuing the generalisation of a nonlinear joint parameter and state estimation method to the NLPV class, by exploiting the availability of the scheduling map data and the properties of the explicit nonlinearity. The purpose of exploiting the explicit nonlinearity is to seek better estimation results than just applying the nonlinear method itself.

    Nonlinearity is a key feature present in almost all physical system controlled over large operating conditions. The extended Masreliez-Martin filter (EMMF) [26] is a an interesting option for the robust identification of nonlinear systems under stochastic, non-Gaussian corruptions. This robust method is based on Huber's statistics foundations [21]. Instead of using a Gaussian distribution assumption as the EKF, the EMMF is synthesised considering that the noise is coloured and belongs to some known distribution class. Furthermore, this algorithm ensures robustness and insensitivity w.r.t. the differences between this distribution class and the real outlier distribution. This paper considers the adaptation of the EMMF to the NLPV case due to its rather simple implementation and easily verifiable convergence property, as previously discussed [40,41]. Furthermore, we stress that the EMMF framework offers offset-free joint parameter and state estimation, with a zone of attraction that can be estimated from an LMI problem.

    Remark 2. System identification theory has been set up since the 70's as the problem of generating parameter estimates in a stochastic framework [25,38], since input data samples are consider as deterministic signals corrupted by noise. As demonstrated by then [16,18,38], the input signal needs to be independent from the process noise for the estimation of the parameter to be unbiased, in the majority of cases. This property remains true in the context of this work. We note that in the experimental validation results (Section 4), we consider a PRBS input signal which is independent from this process noise.

    In this paper, we generalise and apply the EMMF filter filter for the joint state and parameter estimation problem of NLPV models with strcuturally known Lipschitz nonlinearities. The task of estimation both states and parameters is done by using a joint formulation. For this goal, the following first-order affine Lipschitz NLPV system model is considered:

    x(k+1)=A(ρk,θk)x(k)+B(ρk,θk)Φ(x(k))u(k)+wx(k)y(k)=C(ρk,θk)x(k)+wy(k) (2.1)

    with model matrices given as follows:

    A(ρk,θk)=A0θk+ρkA1θk (2.2)
    B(ρk,θk)=B0θk+ρkB1θk (2.3)
    C(ρk,θk)=C0θk+ρkC1θk (2.4)

    being x(k)Rnx the state vector (to be estimated), θ(k)Rnp the parameter vector (also to be estimated), y(k)Rny the output measurement vector, u(k)Rnu the input variable and ρkRnρ the scheduling vector. The explicit nonlinear term Φ:RnxR is structurally known and Lipschitz, as details Assumption 1. We note that u and ρ are, a priori, deterministic and known signals. Furthermore, we stress that x(k) and θ(k) are independent from ρ(k) and u(k).

    Assumption 1. The nonlinear term Φ(x(k)) in the considered NLPV model of Eq (2.1) obeys a local Lipschitz condition around x, this is:

    ||Φ(x)Φ(ˆx)||||Γ(xˆx)||,x,ˆx. (2.5)

    Remark 3. Assuming that the structure of the explicit nonlinearity of the NLPV is known is not at all absurd. Note that, in the majority of LPV cases, these nonlinearities are embedded to the scheduling parameters ρ. On the contrary, for the NLPV case, they are willingly made explicit (outside of ρ) because the designer knows how they are physically derived. In many processes, when two physical systems are cascaded (actuator and the actual process, for instance), and one of them has a known nonlinearity, this nonlinearity re-appears in the input/output relationship for the complete cascaded block. As an example, take a vehicle suspension system: the dynamics of the suspended vehicle chassis depend on the suspension control system (damper and spring), which inherently have known nonlinear behaviours; for the case of electro-rheological suspensions, this behaviour is a hyperbolic tangent function [33].

    Assumption 2. The NLPV system in Eq (2.1) is corrupted by process noise wx(k)Rnx and measurement noise wyRny. We consider that the process noise is Gaussian, whereas the measurement noise is a Gaussian mixture. This is: the process noise is a zero-mean white noise, i.e. p(wx):N(0,Qx(k)), where p() stands for the probability density function and Qx(k) for the co-variance matrix related to wx, whilst the measurement noise wy is non-Gaussian, including the presence of outliers. This noise is distributed as follows:

    Pϵ={p(wy):(1ϵ)p1(wy)+ϵp2(wy)} (2.6)

    which is a mixture of a nominal normal distribution p1(wy):N(0,Q1y(k)) and a contaminating probability density p2(wy):N(0,Q2y(k)), being Q1y(k) and Q2y(k) the co-variance matrices for the nominal and the contamination term. The degree of contamination parameter abides to: 0<ϵ<1.

    Remark 4. The output measurement noise wy, for ϵ=0, would have a Gaussian distribution p1(wy), whereas, for ϵ=1, it would also present a normal distribution, of p2(wy). In both cases, there would be a complete absence of outliers, which is not envied if the stochasticity of the process is to be taken into account.

    Assumption 3. The change in the model parameters has a random walk behaviour, this is:

    θk+1=θk+wθ(k) (2.7)

    or, more generally,

    θk+1=Cθθk+wθ(k) (2.8)

    where p(wθ):N(0,Qθ(k)), being Qθ(k) the co-variance matrix for the random walk distribution. Matrix Cθ is a prior known and non-singular; it can be used to plug information on the model parameter change phenomenon that occurs.

    Despite the fact that many system parameters can be available with some reasonable accuracy, given within a certain range, some model parameters may be entirely unknown because manufacturers consider these data as proprietary information. For example, the precise determination of leakage and friction coefficients causes great difficulty in the control of electro-rheological actuators [29]; accordingly loss of effectiveness faults must be included to reflect the variation of the physical system parameters. Due to the obvious presence of non-Gaussian noises, as well as the impossibility of determining the exact values of the system parameters, the observed system is considered as a stochastic system whose parameters are successfully determined by the proposed identification procedure.

    Bearing in mind the previous Assumptions, the goal of this paper is reiterated: propose an algorithm to jointly estimate the states x(k) and parameters θk of the NLPV system described by Eq (2.1), for all k. Moreover, asymptotic convergence of the estimation is required, despite the presence of non-Gaussian measurement noise (outliers) wy. Output data y(k), input data u(k) and scheduling data ρ(k) is considered available. Formally, this Problem is described:

    Problem 1. For the NLPV system in Eq (2.1), consider an augmented states vector given by:

    z(k)=[x(k)θk]Rnx+np (2.9)

    which compiles not only the states, but also the varying process model parameters. Consider qk() as a nonlinear map at instant k. Then, find an estimation law of the following form:

    ˆz(k+1)=qk(ˆz(k),y(k),u(k),ρk,θk,ξk)=[fk(ˆx(k),y(k),u(k),ρk,ˆθk,ξk)gk(ˆx(k),y(k),u(k),ρk,ˆθk,ξk)] (2.10)
    y(k)=hk(z(k),θk,wy(k)) (2.11)

    with ξk:N(0,Qξ(k)), with Qξ(k)=diag{Qx(k),(1ϵ)Q1y(k)+ϵQ2y(k),Qθ(k)}, encompassing all disturbances:

    ξk=[wTx(k)wTy(k)wTθ(k)]T. (2.12)

    This estimation should minimise the sensitivity of ˆz to wy and ensure that the expectation of estimation error E((e(k))=E(z(k)ˆz(k)) has local asymptotic convergence for any state error starting condition ex(0)=x(0)ˆx(0) given within a basin of attraction Ω, which is equivalent to:

    limke(k)0,ex(0)Ω. (2.13)

    In the sequel, Problem 1 is solved by the use of a modified EMMF. More specifically, the Fischer information of the regular EMMF in the a posterior co-variance matrix is replaced by an approximation given by the derivative of Huber's function, which increases the flexibility of the algorithm and speeds up its convergence rate. The benefits of this modified version of the EMMF is discussed and illustrated in [40,41].

    Remark 5. We note that the EMMF has been shown to possess offset-free convergent robust filtering qualities [41], which can be checked through a set of linear inequalities. If such inequalities are verified, the estimates from the EMMF, given in an augmented state form alike of z(k), retain low sensitivity to measurement outliers.

    Remark 6. The considered NLPV model presented in Eqs (2.1)-(2.4) has affine parameter-dependence w.r.t. the scheduling variable ρk. Anyhow, the method could directly consider polynomial, polytopic and fractional parameter dependencies, for which only the form of matrices A(), B() and C() have to be altered. Moreover, we note that the proposed joint estimation problem, given in the recursive form of Eq (2.10), is set to benefit from the availability of the scheduling parameter data ρk, which coordinates the estimation function qk(), as shown in the sequel. In contrast to the nonlinear EMMF, the proposed NLPV EMMF uses an explicit ρk-dependent time-varying function qk(). As done in other LPV identification methods [14], the scheduling variable ρk data is used to provide more data to the filter, which is potentially better than using a nonlinear EMMF that disregards the knowledge of ρk.

    Lemma 1. Consider a joint state and parameters estimation law of the form given by Eq (2.10), in Problem 1. The convergence of the estimation error e(k) can be analysed with an associated ODE system. If this associated system is locally asymptotically stable, which can be verified through the direct Lyapunov method, then, the expectation of the estimation error E(e(k)) converges.

    Proof. Provided in the original paper by L. Ljung [25].

    In order to develop the proposed robust method for joint estimation of states and parameters of affine NLPV systems with explicit Lipchitz nonlinearities, given in the form of Eq (2.1), the full expression for z(k), in terms of the nonlinear maps qk() and hk() are detailed:

    qk1()=[(A0+A1ρk1)θk10]z(k1)+(B0+B1ρk1)Ψ([Inx0np]z(k1))+([Inx0ny0np]ξk1) (3.1)
    hk()=[(C0+C1ρk)θk0]z(k)+([0nxIny0np]ξk). (3.2)

    Then, based on these expressions, the extended robust filter (EMMF) [40] is proposed for the NLPV case. The estimation of z(k), namely ˆz(k), addressing Problem 1, has the following form:

    P(k)=F(k1)P(k1)FT(k1)+L(k1)Qξ(k1)LT(k1)=P(k)K(k)Ψ(v(k))TKT(k) (3.3)
    K(k)=PT(k)HT(k)TT(k) (3.4)
    v(k)=T(k)[y(k)hk(ˆz(k),ˆθk,0)] (3.5)
    ˆz(k)=ˆz(k1)+K(k)Ψ(v(k)) (3.6)
    Ψ(v(k))=col{ψ(v1(k))ψ(vny(k))}T (3.7)
    T(k)=[H(k)P(k)HT(k)+V(k)Q1y(k)VT(k)]12, (3.8)

    being implied that:

    F(k)=qkz|ˆz(k)=[fkxfkθ0nxInp]|ˆx(k),ˆθ(k) (3.9)
    L(k)=qkξk|ˆz(k)=Inx+np (3.10)
    H(k)=hkz|ˆz(k)=[hkxhkθ]|ˆx(k),ˆθ(k) (3.11)
    V(k)=hkwy|ˆz(k)=Iny. (3.12)

    For the outputs contaminated with outliers, the nonlinear function ψ() issues a transformation upon v(k), with is associated to the output prediction error. This nonlinearity is a Huber's influence function, as gives:

    ψ(vj(k))=min{||vj(k)||,kjϵ}sign{vj(k)} (3.13)

    with its derivative with respect to vj(k) given by:

    ˙ψ(vj(k))={1,||vj||<kjϵ,0,||vj||kjϵ. (3.14)

    Remark 7. The sign function is defined as usual, being equal to identity for positive entries, null for null entries and minus identity for negative entries.

    Remark 8. The use of the nonlinear Huber's influence function, given in Eq (3.13), is set to assign less weight to small quantities of larger residuals, such that the presence of outliers does not have significant influence on the final estimate ˆz(k), while giving unit weight to the main part of the data population with moderate residuals.

    The following block structured form of the EMMF estimation gain and co-variance matrices are given:

    K(k)=[N(k)TM(k)T]T (3.15)
    P(k)=[P1(k)P2(k)P3(k)]. (3.16)

    Finally, the estimation law from Eq (3.6) is given in a compact, recursive form, as follows:

    ˆz(k)=ˆz(k1)+[N(k)M(k)]Ψ(v(k)) (3.17)

    which is directly equivalent to:

    ˆx(k)=fk1(ˆx(k1),u(k1),ˆθk1) (3.18)
    +N(k)Ψ(v(k))=A(ρk1,ˆθk1)ˆx(k1)+B(ρk1,ˆθk1)Φ(ˆx(k1))u(k1)+N(k)Ψ(v(k))ˆθk=Cθˆθk1+M(k)Ψ(v(k)). (3.19)

    We note how Eq (3.18) explicitly expresses the state and parameter estimates w.r.t. previous scheduling parameters data, benefiting from its availability. In a full nonlinear formulation, as in the original EMMF, with unknown scheduling parameter data ρk, the nonlinear map fk1() is only written in terms of ˆx(k1), u(k1), and ˆθk1.

    Remark 9. The initial conditions for the estimation algorithm are:

    {ˆz(0)=0P(0)=[P(x0)0np0nxP(θ0)]. (3.20)

    These values (ˆx(0) and θ0) represent some a priori knowledge about the NLPV system variables. For the case when these are not known, they can be chosen, for simplicity, as null entries.

    Remark 10. The main difference from the proposed joint state and parameter estimation algorithm is the use of Huber's nonlinearity Ψ(). Its insertion suitably transforms the output prediction errors to penalize the presence of outliers and, thus, eliminate their influence upon the resulting estimation.

    In order verify that the proposed joint state and parameter estimation algorithm converges, i.e. limke(k)0, an associated ODE system is presented and its stability is verified using a direct Lyapunov method over the associated ODE from Lemma 1.

    Lemma 2. The expectation of the joint parameter and state estimation error derived from proposed method in Eqs (3.18)-(3.19) is associated to an ODE system*, which is given by:

    *With ex(τ)=x(τ)ˆx(τ) and eθ(τ)=θ(τ)ˆθ(τ).

    de(τ)dτ=[dex(τ)dτ)deθ(τ)dτ] (3.21)
    dex(τ)dτ=fex(τ) (3.22)
    deθ(τ)dτ=feθ(τ). (3.23)

    The nonlinear functions fex() and feθ() derive from the difference between the Lipschitz affine NLPV system dynamics and the estimation from the algorithm, as follows:

    fex(τ)=AAex(τ)+BA(Φ(x(τ))Φ(ˆx(τ))¯uN(τ)Ψ(v(τ)) (3.24)
    feθ(τ)=CAeθ(τ)M(τ)Ψ(v(τ)). (3.25)

    Proof. The proof follows from [25]. Take a continuous version of the discrete time system where τ=kTs, where Ts is the sampling period of the NLPV system in Eq (2.1). Then, taking k=kTs, the following continuous-time approximation is valid: dr(τ)dτ=r(k)r(k1).

    Firstly, take the dynamics of the parameter estimation error, for which the parameter estimates are given by Eq (3.19), considering wθ null:

    eθ(k)=(Cθθk1)(Cθˆθk1+M(k)Ψ(v(k))). (3.26)

    It directly follows that:

    deθ(τ)dτ=CA(CθInp)eθ(τ)M(τ)Ψ(v(τ)) (3.27)

    which asymptotically converges to some constant ¯eθ due to the construction of M(τ).

    Then, take the dynamics of the state estimation error, for which the state estimates are given by Eq (3.18):

    ex(k)=x(k)ˆx(k)=A(ρk1,θk1)x(k)+B(ρk1,θk1)Φ(x(k1))u(k1)A(ρk1,ˆθk1)ˆx(k)B(ρk1,ˆθk1)Φ(ˆx(k1))u(k1)N(k)Ψ(v(k))

    which, by exploiting the affine nature of A and B, leads to:

    ex(k)=(A0+A1ρk1)eθ(k1)ex(k1)+(B0+B1ρk1)eθ(k1)(Φ(x(k1)))u(k1)(B0+B1ρk1)eθ(k1)(Φ(ˆx(k1)))u(k1)N(k)Ψ(v(k)). (3.28)

    Finally, since eθ(τ)=¯eθ and taking u(k1)=¯u as constant, one obtains:

    dex(τ)dτ=AA[A0+A1ρk1Inx]¯eθex(τ)+BA[B0+B1ρk1Inx]¯eθ(Φ(x(τ))Φ(ˆx((τ)))¯uN(τ)Ψ(v(τ)). (3.29)

    This concludes the proof.

    Lemma 3. The ODE system of Eq (3.21) is locally asymptotically stable for any real Lipschitz constant Γ if there exists a positive definite matrix PL=PTL0 and a scalar ϵγR such that the following LMI holds:

    [ϵγΓTΓBTAPLATATA(τ)TPL0ϵγInx0PLTA(τ)AAPLBA(N(τ)TPL+PLN(τ))]<0T(τ)[(C0+C1ρk)¯eθ]=TA(τ).

    Then, for any starting condition x(0)Ω, which is the sub-level ellipsoid associated to PL, the proposed algorithm in Eq (2.10) yields local asymptotically stability.

    Proof. In order verify the stability property of the ODE system of Eq (3.21), the direct Lyapunov method is used. Assuming there exists some positive definite matrix PL, consider the following positive definite candidate Lyapunov function:

    V(τ)=E{Ψ(v(τ))TPLΨ(v(τ))}. (3.30)

    The local basin of attraction is given w.r.t. the state error estimation ex(k), since the parameter error eθ(k) asymptotically converges to some constant ¯eθ by construction (refer to the proof of Lemma 2). This basin of attraction can be estimated as sub-level set ellipsoid related to PL, i.e.:

    Ω:={xRnx|xTPLx0}. (3.31)

    Differentiating V w.r.t. τ along the solution of the ODE system, we get:

    dV(τ)dτ=E{Ψ(v(τ)TPLΨ(v(τ))τ}. (3.32)

    Note that the latter term Ψ(v(τ))τ can be expanded as follows:

    Ψ(v(τ))τ=dΨ(v(τ))dvv(τ)τ. (3.33)

    Moreover, note that v(τ)=T(τ)[C(ρk,θ(τ))x(τ)C(ρk,ˆθ(τ))ˆx(τ)]. Then, it holds for dT(τ)dτ=0 that:

    v(τ)=T(τ)[(C0+C1ρk)¯eθ]=TA(τ)ex(τ) (3.34)
    dv(τ)dτ=TA(τ)dex(τ)dτ. (3.35)

    Recall from previous development that Ψ(v(τ))v is naturally bounded, i.e.:

    0<Ψ(v(τ))vInx (3.36)

    which means that (dΨ(v(τ))dτdv(τ)dτ) holds. Thus:

    dV(τ)dτE{dv(τ)dτTPLΨ(v(τ))+Ψ(v(τ))TPLdv(τ)dτ}dV(τ)dτE{[TA(τ)AAex(τ)+BA(Φ(x(τ))Φ(ˆx(τ)))¯u]TPLΨ(v(τ))Ψ(v(τ))TPL[TA(τ)AAex(τ)+BA(Φ(x(τ))Φ(ˆx(τ)))¯u]}

    Defining λ(τ)=(ex(τ)T[(Φ(x(τ))Φ(ˆx(τ)))¯u]TΨ(v(τ))T)T, it follows that:

    dV(τ)dτE{λ(τ)TMλλ(τ)} (3.37)

    with:

    Mλ=[0BTAPLATATA(τ)TPL000PLTB(τ)AAPLBA(N(τ)TPL+PLN(τ))].

    The Lipschitz condition from Eq (2.5) holds as:

    [(Φ(x(τ))Φ(ˆx(τ)))¯u]T[(Φ(x(τ))Φ(ˆx(τ)))¯u]ex(τ)TΓTΓex(τ) (3.38)

    which is analogous to:

    λ(τ)T[ΓTΓ000Inx0000]MΓλ(τ)0. (3.39)

    Since the direct Lyapunov method requires dV(τ)dτ to be nonpositive, the explicit nonlinearities can be overlapped with the aid of the Lipschitz condition, taking:

    dV(τ)dτ<0 (3.40)
    ϵγ(λ(τ)TMΓλ(τ))<0 (3.41)
    dV(τ)dτϵγ(λ(τ)TMΓλ(τ))<0 (3.42)

    for ϵγ being a (scalar) slack variable.

    Finally, the S-procedure [7] can be applied to Eq (3.42), considering λ(τ)TMΓλ(τ)=E{λ(τ)TMΓλ(τ)}. By doing so, condition:

    E{λ(τ)T(MλϵγMΓ)λ(τ)}<0 (3.43)

    is equivalently expressed in the form of the LMI in Lemma 3. Therefore, if the positive definite PL and a scalar ϵγ indeed exists, the Lyapunov condition is verified and the proposed algorithm guarantees offset-free joint state and parameter estimation. This concludes the proof.

    Remark 11. We note that inequality (3.43) is a potentially conservative sufficient condition that ensures the convergence of the NLPV joint estimation algorithm. Therefore, if there exists some positive definite PL which verifies LMI in Lemma 3, the local zone of attraction within which the ODE from Eq (3.21) displays asymptotic convergence may be small. The size of this zone depends on a series of matters, such as the model of the process and the Lipschitz constant Γ. In Section 4, we show an illustration of this attraction region for a suspension system example.

    In this Section, we present experimental validation results concerning the proposed joint parameter and state estimation, considering Lipschitz NLPV in form of Eq (2.1).

    The identified process is that of a semi-active suspension system. The used data is provided by the sensors of a 1/5-scaled vehicle mechatronic test-bench, equipped with 4 semi-active Electro-rheological (ER) dampers with the force range of ±50N. This vehicle test-rig, shown in Figure 1, allows testing under different configurations and use-cases (for full details, refer to www.gipsa-lab.fr/projet/inove). Each damper is controlled through a PWM signal given within the [00.35] range. Below each wheel of the vehicle lies an OMRONTM linear servomotor which is able to mimic various road type conditions, with a maximal velocity threshold of 1.5m/s.

    These PWM modules operate at 25kHz rate duty-cycles and vary the electric field that is applied over each ER damper chambers and, thus, change the ER fluid flows viscosity and the delivered damping force.

    Figure 1.  INOVE 1/5-sized vehicle.

    In order to describe the vertical dynamics of this experimental platform, a decoupled, quarter-car modelling framework is followed: each corner of the vehicle is represented by an individual set of equations. Each damper is controlled seeking to minimise the effects of the road upon the safety and comfort of the passengers on-board. The dynamic coupling effects between the corners are neglected. Through the sequel, the presented results are those from the front-left corner test-bench.

    Figure 2 shows a schematic diagram of the quarter-car representation of a vehicle with four suspension units. Each semi-active suspension system comprises a spring with a stiffness parameter ks and a controlled damper of variable damping coefficient c(). The chassis body at each side is represented by a sprung mass ms, which is connected to the wheel-link, represented by the unsprung mass mus. The wheel is represented by a spring with stiffness kt. The vertical dynamics are described regarding the displacements of the sprung and unsprung masses: zs(t) and zus(t), respectively. The road profile is zr(t) a disturbance signal to this system.

    Figure 2.  Vehicle corner with semi-active suspension system.

    The force provided by each ER damper is described through a dynamic version [33] of the widely-used nonlinear hyperbolic tangent model [19], as follows:

    Fd(t)=k0zdef(t)+c0˙zdef(t)+FER(t)dFERdt(t)=1τFER(t)+fcτtanh(k1zdef(t)+c1˙zdef(t))u(t) (4.1)

    where zdef(t)=(zs(t)zus(t)) represents the suspension deflection and u(t) stands for duty cycle of a PWM signal that regulates the ER damper. Regarding control purposes, this PWM variable u(t) is the control input to the suspension system.

    The spring and tire forces are given by:

    Fs(t)=kszdef(t) (4.2)
    Ft(t)=kt(zus(t)zr(t)). (4.3)

    The dynamics of the sprung and unsprung masses are obtained using regular laws of motion around an origin equilibrium:

    ms¨zs(t)=Fs(t)Fd(t) (4.4)
    mus¨zus(t)=Fs(t)+Fd(t)Ft(t). (4.5)

    Table 1 presents the model parameter values and their descriptions. The nominal values for spring stiffness, passive damper stiffness and viscous damping coefficient are provided by the manufacturers of these models. The remaining values were previously identified using regular state-of-the-art procedures [48].

    Table 1.  Vehicle model parameters.
    Parameter Description Value Unit
    ms Sprung mass 2.27 kg
    mus Unsprung mass 0.32 kg
    ks Spring stiff. 1396 N/m
    kt Tire stiff. 12270 N/m
    k0 Passive damp. stiff. 170.4 N/m
    c0 Viscous damp. coef. 68.83 N.s/m
    k1 Hysteresis displ. coef. 218.16 N/m
    c1 Hysteresis vel. ceof. 21 N.s/m
    fc Dynamic yield force 28.07 N
    τ Time constant 43 s

     | Show Table
    DownLoad: CSV

    We must stress that the damping characteristics of the ER damper vary over time. According to the lifespan of these components, the passive stiffness k0 and viscous damping coefficient c0 have variations, for multiple reasons [29,31], such as small oil leakages, air pressure inside the damper chamber, influence of external (high) temperatures, etc. Therefore, we assume that these parameters should be identified online and are subject to a random walk behaviour and additive noise, in the fashion of Eq (2.8). Accordingly, it is implied that:

    θk=[k0kc0k]T (4.6)

    where θ0 comprises the values for k0 and c0 given in Table 1 (nominal values).

    The two major control objectives [34] of semi-active suspensions systems are: vehicle body isolation and passenger comfort enhancement. These two goals are physically conflicting: while stiff/high damping enhances passenger comfort, smooth/low damping enables easier road holding. An accurate knowledge of the actuator dynamics (damping force delivered by the ER damper) is necessary to correctly design control strategies regarding these two objectives. Accordingly, the online estimation of parameters θ is essential.

    We note that the active hysteresis coefficients k1, c1 and the dynamic yield force of the ER damper fc appear only in the controlled part of the damper force FER(t), see Eq (4.1). This means that possible variations upon these parameters (due to external factors, for instance) can be treated directly as actuator loss of effectiveness, as done in previous papers [29]. Therefore, their online estimation is not as important as the one regarding the passive terms, present in Fd(t)FER(t), which directly appear in the dynamics and cannot be indirectly accounted for.

    Furthermore, as illustrates Table 1, we note that the tire stiffness and spring stiffness values are much larger that those for the damping coefficient, which means that the model is not so sensible to their variations. Changes upon the sprung mass and unsprung mass are usually accounted for by robustness, since any suspension control requires the tolerance of these parameters within a nominal interval [30], since these parameters vary according to the amount of vehicle passengers.

    In order to cast this quarter-car suspension system dynamics into a Lipschitz NLPV state-space model, the system states are selected as follows:

    x(t)=[zdef(t)˙zs(t)(zus(t)zr(t))˙zusFER(t)]T.

    The available measurements of this system are the vertical acceleration variables, given through on-boards sensors (inertial units/accelerometers), as follows:

    y(t)=[¨zs(t)¨zus(t)]T.

    Then, considering wy as coloured measurement noise and ˙zr(t) as the load disturbance variable, an NLPV state-space formulation is found with ρ=u(t) as the scheduling variable:

    ˙x(t)=Acx(t)+Bc(ρ)Φ(x(t))+Dc1ω(t)y(t)=Ccx(t)+Dc2ω(t) (4.7)

    where ω(t)=[˙zr(t),wy(t)]T is a concise disturbance vector. The nonlinearity is:

    Φ(x(t))=tanh(Γinx(t)) (4.8)

    with Γin=[k1,c1,0,c1,0]. Notice that Eq. (4.8) verifies a local Lipschitz condition in x, as expected by Assumption 1:

    ||Φ(xˆx)||Γ||(xˆx)||x,ˆx (4.9)

    with Γ=||Γin|| is the smallest Lipchitz constant that verifies Eq (4.9).

    The state-space matrices Ac, Bc, Cc, Dc1 and Dc2 are given by:

    Ac=[01010(ks+k0k)msc0ms0c0kms1ms00010(ks+k0k)musc0musktmusc0kmus1mus00001τ]Bc(ρ)=[0000fcτρ]TDc1=[0010000000]TCc=[(ks+k0k)msc0kms0c0ms1ms(ks+k0k)musc0kmusktmusc0mus1mus]Dc2=[0000.01].

    This model is Euler-discretized with a sampling period of Ts=5ms, which is an operational constraint of the considered test-bench. This leads to discrete-time Lipschitz NLPV system of the form in Eq (2.1). W.r.t. to the original problem setup model, the discretization implies A()=Inx+TsAc(), B()=TsBc() and C()=TsCc().

    Regarding the previous notation, the road profile derivative disturbance ˙zr(t) stands for wx(t). We note that in many modern cars, cameras (and adaptive estimation algorithms) are used to preview the future road profiles, e.g. [30,44,46]. This is trivial and widely seen in the automotive suspension literature [30]. Thus, from the viewpoint of the joint estimation algorithm, wx is a known variable.

    It must be remarked that the considered experimental test-bench has physical limits to the motion variables (zs, ˙zs, etc.). Therefore, it is implied that the states x have upper and lower limits; mathematically, this is expressed as a box-type set constraint xXt, where:

    X:={x=col{xj}Rnx|xj_xj¯xjjN[1,nx]}.

    The numerical values for these upper and lower (physical) limits of the states are given below, in Table 2.

    Table 2.  State admissible limits.
    Description xj_ State xj ¯xj
    Suspension Deflection 0.6m x1 0.6m
    Chassis Body Velocity 20m/s x2 20m/s
    Tire Deflection 0.6m x3 0.6m
    Wheel-Link Velocity 60m/s x4 60m/s
    Controlled Damper Force 50N x5 50N

     | Show Table
    DownLoad: CSV

    Before presenting the actual estimation results obtained with the proposed algorithm, its convergence property regarding state estimation must be analysed. In fact, state estimation convergence can be tested with the LMI provided in Lemma 3. For the considered system, the corresponding LMI is indeed verified, which ensures that the Lyapunov convergence condition holds. The symmetric positive definite matrix PL for which this LMI holds is the following, with ϵγ =8.02108:

    PL=[0.05150.00170.01863.1e111.58e95.6e56.14e42.9e115.72e110.01065.06e1151.36e96.31e51.29e52.48e4].

    The sub-level ellipsoid set of PL determines the zone of attraction estimate for the proposed NLPV EMMF. This is, for any starting condition x0Ω, convergence is verified. We note that the LMI problem from Lemma 3 is evaluated offline within 2.10s, using Matlab software and SDPT3 solver on an i5 CPU@2.4 GHz Macintosh with 8 GB of RAM.

    Regarding this matter, Figure 3 shows this zone of attraction in three different 2D cuts: on the left-side, the region of attraction for states x1 and x2 is given; in the center, the region for states x3 and x4 is shown; while on the right-side, the region for states x4 and x5 is exhibited. In this Figure, the (minimal and maximal) physical limits of each state are also displayed (those in Table 2). Clearly, for any physically possible x0, the algorithm will ensure convergence, since all possible x0 are indeed contained inside Ω: XΩ, which means that x0(x0Xx0Ω) and, thus, the proposed solution is recursively feasible and ensures convergence when applied to the considered suspension system process.

    Figure 3.  Zone of attraction of the proposed algorithm.

    For the application of the proposed algorithm, a real input set comprising t=5s of vertical vehicle motions is considered (i.e. k=1000 iterations). The road profile zr(t) represents a vehicle running in a straight line on a dry road when it encounters a sequence of 13mm bumps along its four wheels. This real road profile zr(t) applied to the test-bench through the controlled servomotors under the wheels and its derivative (input of the NLPV model) wx(t)=˙zr(t) are presented in Figure 4.

    Figure 4.  Road profile scenario (Known Disturbances).

    The test-bed model is then fed with a PRBS signal bounded within [0,0.3], which is given in Figure 5. This input signal is of PRBS type in order to provide frequency-rich output signals y (chassis body and wheel-link accelerations). These outputs, which are corrupted with coloured noise, are presented in Figure 6. As shown in this Figure, we note that the measurement noise is quite significant: the average noise-to-signal ratios for the chassis body and wheel-link acceleration outputs are of roughly 10% and 5%, respectively.

    Figure 5.  PRBS control input / scheduling variable.
    Figure 6.  Measured outputs, ϵ=0.05.

    The achieved estimation results are now presented. In order to show the advantages of the proposed method conceived for Lipschitz NLPV systems, it is compared against: i) an extended Kalman filter [47], which is tuned based on a tangent linerised model of the process around the origin equilibrium; and ii) a robust EMMF [41], which considers nonlinear process dynamics and does not make use of the scheduling variable and does not take into account the Lipschitz behaviour along the input trajectory. Through the sequel, NLPV-MEMMF denotes the results achieved with the proposed algorithm, EKF those obtained with a the extended Kalman filter and RMMF those for the robust nonlinear filter.

    We note that the results comprise the application of these methods for different degrees of contamination ϵ, as gives Eq (2.6). For presentation simplicity, the estimation results are shown only for ϵ=0.05. Anyhow, we assess the results considering other contamination levels through performance indexes.

    The following results, considering the joint estimation of the NLPV system states x and the variations of the damper force passive parameters θk, were elaborated on a i5 CPU@2.4 GHz (2 Cores) Macintosh with 8 GB of RAM, with the aid of Matlab. The average execution time required to evaluate the recursive law given in Eq (2.10) is of 3.43ms per iteration, which means that the proposed method can be used in online vehicle suspension applications, running on embedded on-board micro-controllers.

    Considering a ϵ=0.05, Figure 7 shows the identification of the varying parameters over the 5s of data, while Figure 8 shows the estimation of the five system states. Clearly, convergent state estimation results and adequate parameter identification curves are obtained with the proposed method. These results could certainly serve for the computation of online state-feedback control laws, such as robust MPCs, for instance.

    Figure 7.  Parameters estimation, ϵ=0.05.
    Figure 8.  State estimation, ϵ=0.05.

    In order to quantify the performance of the proposed tool against the other methods, Figures 9, 10 and 11 show, respectively the Mean Square Error (MSE) index (in logarithmic scale), computed as the squared mean 2-norm of the deviance between real variables z and their estimations ˆz (along the dataset), for different leves of contamination: ϵ=0.05, 0.1 and 0.2, respectively. Note that z comprises states and parameters. Clearly, the smallest MSE results are obtained with the proposed NLPV-EMMF method, for all degrees of contamination, indicating the effectiveness of the proposed tool. Recall that, in the logarithmic scale, results closer to indicate smaller estimation errors.

    Figure 9.  Joint parameter and state estimation: mean square error for ϵ=0.05.
    Figure 10.  Joint parameter and state estimation: mean square error for ϵ=0.1.
    Figure 11.  Joint parameter and state estimation: mean square error for ϵ=0.2.

    Complementary, to quantify the identification performance of the three methods regarding state and parameter estimations separately, the Robust Akaike criterion (RAC) [1] is considered. This criterion measures the inconsistency between the probability density of the estimation results against the exact probability density of the parameter variations (random walk). This criterion has a dynamic solution, as follows: RAC(k)=ln(1k1(RAC(k1)+||θkˆθk||2)).

    Table 3 shows the means values obtained for the parameter estimation considering the RAC index (RAC) and the logarithmic Mean Square Error (MSE) for the state estimation; both indexes are computed for the ϵ=0.05 case. Evidently, the proposed NLPV-MEMMF achieves the best results. This is very interesting and means that the proposed solution can indeed serve for an online parameter and state estimation tool for controlled Lipschitz NLPV processes subject to inherent stochasticity, benefiting from the availability of the scheduling parameter data and the Lipschitz characteristic of Φ(x).

    Table 3.  Performance indexes: RAC (Parameter Estimation) and MSE (State Estimation).
    Method θ: (mean RAC) x: ln(MSE)
    EKF 1.248 1.2211
    RMMF 1.559 1.2764
    NLPV-MEMMF 3.402 3.8948

     | Show Table
    DownLoad: CSV

    The presented results have shown that the widely-used extended Kalman filter (EKF) is very sensitive to the presence of non-Gaussian noises, as opposed to the proposed robust joint estimation algorithm (NLPV-MEMMF). Also, it is clearly shown that the included modifications to extended Masreliez-Martin filer (RMMF) for joint estimation increase the practical usability and convergence rate of the algorithm, which is now specially tendered for Lipschitz NLPV systems. In order to show robustness of the proposed robust joint estimation algorithm for systems with parameter faults with respect to these other conventional (and widely-used) joint estimation algorithms, the algorithms are tested over 1000 random and independent simulations, for different contamination degrees. Regarding these tests, it is particularly important to note that NLPV-MEMMF maintains its high performances regarding the other methods, for all contamination degrees.

    To conclude this Section, Table 4 provides the statistical data based on these random independent tests. Regarding this Table, for the cases of levels of high contamination degree (ϵ=0.2), it can be seen that the worst results obtained by NLPV-MEMMF are even better than the best result obtained by other methods. It can be clearly seen that the superiority of the proposed robust algorithm is greater in higher degrees of contamination, since the proposed method yields the least amount of variance in the considered performance index. Also, it is important to notice that, in presence of non-Gaussian noises, the proposed NLPV-MEMMF is an attractive alternative solution which outperforms other algorithms and, at the same time, has reduced complexity in comparison with them.

    Table 4.  Joint parameter and state MSE for different degrees of contamination, 1000 random tests.
    Method Mean Best Worst Var.
    ϵ=0.05
    EKF 1.325 0.118 2.915 0.815
    RMMF 2.555 3.769 0.890 0.217
    NLPV-MEMMF 3.411 3.916 2.667 0.081
    ϵ=0.1
    EKF 2.633 0.397 4.954 0.912
    RMMF 2.201 3.501 0.935 0.214
    NLPV-MEMMF 2.922 3.647 1.928 0.094
    ϵ=0.2
    EKF 4.172 0.820 7.251 1.732
    RMMF 1.147 0.071 1.359 0.544
    NLPV-MEMMF 2.454 3.066 1.741 0.111

     | Show Table
    DownLoad: CSV

    This paper proposed a robust system identification algorithm specific conceived for Nonlinear Parameter-Varying models subject to stochasticity. The explicit nonlinear term obeys a Lipschitz condition and the matrices are affine on the scheduling parameters. The proposed approach, based on a modified extended Masreliez-Martin filter, jointly estimates the NLPV system states together with the physical parameters with precision. The estimation error asymptotically converges, despite the non-Gaussian disturbances. The effectiveness of the proposed algorithm is verified using real data from vehicle suspension mechatronic test-bed. As illustrated, the method can effectively estimate estates and stochastic parameter variations, becoming an interesting option for online state-feedback control policies, since it only uses output data. The major drawback of the proposed method is that it requires the explicit time-varying nonlinearity to be structurally known and Lipschitz, as well as needing a linear parametrisation of a priori evaluated basis functions. For future works, the Authors plan on comparing the method against artificial neural networks and genetic algorithms set for the structural discovery of the model dynamics, which do not require such assumptions.

    The Authors thank Prof. Olivier Sename and his team at gipsa-lab for the application example (dataset and models). M. M. Morato is partially supported by ITEA3 European project 15016 EMPHYSIS (Embedded Systems With Physical Models in the Production Code Software) and CNPq project 304032/20190. V. Stojanovic has been supported by the Serbian Ministry of Education, Science and Technological Development (4510368/202014/200108).

    The authors declare no potential conflict of interests.



    [1] H. Akaike, A new look at the statistical model identification, IEEE T. Automat. Contr.,, 19 (1974), 716–723.
    [2] K. J. Åström, P. Eykhoff, System identification: A survey, Automatica, 7 (1971), 123–162.
    [3] A. Bachnas, R. Tóth, J. Ludlage, A. Mesbah, A review on data-driven linear parameter-varying modeling approaches: A high-purity distillation column case study, J. Process Contr., 24 (2014), 272–285.
    [4] B. Bamieh, L. Giarre, Identification of linear parameter varying models, Int. J. Robust Nonlin., 12 (2002), 841–853.
    [5] F. D. Bianchi, R. S. Sánchez-Peña, Robust identification/invalidation in an LPV framework, Int. J. Robust Nonlin., 20 (2010), 301–312.
    [6] J. Blesa, P. Jiménez, D. Rotondo, F. Nejjari, V. Puig, An interval NLPV parity equations approach for fault detection and isolation of a wind farm, IEEE T. Ind. Electron., 62 (2014), 3794–3805.
    [7] S. Boyd, L. El Ghaoui, E. Feron, V. Balakrishnan, Linear matrix inequalities in system and control theory, vol. 15, Siam, 1994.
    [8] S. L. Brunton, J. L. Proctor, J. N. Kutz, Discovering governing equations from data by sparse identification of nonlinear dynamical systems, P. Natl Acad. Sci., 113 (2016), 3932–3937.
    [9] H.-F. Chen, L. Guo, Identification and stochastic adaptive control, Springer Science & Business Media, 2012.
    [10] P. Cui, H. Zhang, J. Lam, L. Ma, Real-time kalman filtering based on distributed measurements, Int. J. Robust Nonlin., 23 (2013), 1597–1608.
    [11] F. Daum, Nonlinear filters: beyond the kalman filter, IEEE Aero. El. Sys. Mag., 20 (2005), 57–69.
    [12] P. L. Dos Santos, J. Ramos, J. M. de Carvalho, Identification of linear parameter varying systems using an iterative deterministic-stochastic subspace approach, in Proceedings of the European Control Conference, (2007), 4867–4873. IEEE.
    [13] P. L. Dos Santos, J. A. Ramos, J. M. De Carvalho, Identification of LPV systems using successive approximations, in 47th IEEE Conference on Decision and Control, (2008), 4509–4515. IEEE.
    [14] P. L. Dos Santos, Linear parameter-varying system identification: New developments and trends, vol. 14, World Scientific, 2012.
    [15] P. L. Dos Santos, T. P. Azevedo-Perdicoúlis, J. A. Ramos, J. L. M. De Carvalho, G. Jank, J. Milhinhos, An LPV modeling and identification approach to leakage detection in high pressure natural gas transportation networks, IEEE T. Contr. Syst. T., 19 (2010), 77–92.
    [16] E. Gassiat, E. Gautherat, Identification of noisy linear systems with discrete random input, IEEE T. Inform. Theory, 44 (1998), 1941–1952.
    [17] M. Gharamti, B. Ait-El-Fquih, I. Hoteit, An iterative ensemble kalman filter with one-step-ahead smoothing for state-parameters estimation of contaminant transport models, J. Hydrol.,, 527 (2015), 442–457.
    [18] G. Giordano, J. Sjöberg, Consistency aspects of wiener-hammerstein model identification in presence of process noise, in 55th Conference on Decision and Control, (2016), 3042–3047. IEEE.
    [19] S. Guo, S. Yang, C. Pan, Dynamic modeling of magnetorheological damper behaviors, J. Intel. Mat. Syst. Str., 17 (2006), 3–14.
    [20] C. Hoffmann, H. Werner, A survey of linear parameter-varying control applications validated by experiments or high-fidelity simulations, IEEE T. Contr. Syst. T., 23 (2014), 416–433.
    [21] P. J. Huber, Robust statistics, Springer, 2011.
    [22] W. E. Larimore, P. B. Cox, R. Tóth, CVA identification of nonlinear systems with LPV state-space models of affine dependence, in Proceedings of the 2015 American Control Conference, (2015), 831–837. IEEE.
    [23] L. H. Lee, K. Poolla, Identification of Linear Parameter-Varying Systems Using Nonlinear Programming, J. Dyn. Syst-T ASME, 121 (1999), 71–78.
    [24] J. Li, W. X. Zheng, J. Gu, L. Hua, A recursive identification algorithm for wiener nonlinear systems with linear state-space subsystem, Circ. Syst. Signal Pr., 37 (2018), 2374–2393.
    [25] L. Ljung, Asymptotic behavior of the extended kalman filter as a parameter estimator for linear systems, IEEE T. Automat. Contr., 24 (1979), 36–50.
    [26] C. Masreliez, R. Martin, Robust bayesian estimation for the linear model and robustifying the kalman filter, IEEE T. Automat. Contr., 22 (1977), 361–371.
    [27] W. Mei, G. Shan, C. Wang, Practical development of the second-order extended kalman filter for very long range radar tracking, Signal Process., 91 (2011), 1240–1248.
    [28] J. Mohammadpour, C. W. Scherer, Control of linear parameter varying systems with applications, Springer Science & Business Media, 2012.
    [29] M. M. Morato, O. Sename, L. Dugard, M. Q. Nguyen, Fault estimation for automotive electro-rheological dampers: LPV-based observer approach, Control Eng. Pract., 85 (2019), 11–22.
    [30] M. M. Morato, M. Q. Nguyen, O. Sename, L. Dugard, Design of a fast real-time LPV model predictive control system for semi-active suspension control of a full vehicle, J. Franklin I., 356 (2019), 1196–1224.
    [31] M. M. Morato, O. Sename, L. Dugard, LPV-MPC fault tolerant control of automotive suspension dampers, IFAC-PapersOnLine, 51 (2018), 31–36.
    [32] A. S. Morris, R. Langari, Measurement and instrumentation: theory and application, Academic Press, 2012.
    [33] T.-P. Pham, O. Sename, L. Dugard, Real-time damper force estimation of vehicle electrorheological suspension: A nonlinear parameter varying approach, IFAC-PapersOnLine, 52 (2019), 94–99.
    [34] C. Poussot-Vassal, S. M. Savaresi, C. Spelta, O. Sename, L. Dugard, A methodology for optimal semi-active suspension systems performance evaluation, in 49th IEEE Conference on Decision and Control, (2010), 2892–2897. IEEE.
    [35] S. Z. Rizvi, J. M. Velni, F. Abbasi, R. Tóth, N. Meskin, State-space lpv model identification using kernelized machine learning, Automatica, 88 (2018), 38–47.
    [36] T. B. Schön, A. Wills, B. Ninness, System identification of nonlinear state-space models, Automatica, 47 (2011), 39–49.
    [37] G. Scorletti, V. Fromion, S. De Hillerin, Toward nonlinear tracking and rejection using LPV control, IFAC-PapersOnLine, 48 (2015), 13–18.
    [38] T. Söderström, Identification of stochastic linear systems in presence of input noise, Automatica, 17 (1981), 713–725.
    [39] V. Stojanovic, V. Filipovic, Adaptive input design for identification of output error model with constrained output, Circ. Syst. Signal Pr., 33 (2014), 97–113.
    [40] V. Stojanovic, N. Nedic, Joint state and parameter robust estimation of stochastic nonlinear systems, Int. J. Robust Nonlin., 26 (2016), 3058–3074.
    [41] V. Stojanovic, N. Nedic, Robust kalman filtering for nonlinear multivariable stochastic systems in the presence of non-gaussian noise, Int. J. Robust Nonlin., 26 (2016), 445–460.
    [42] S. Tayamon, T. Wigren, Recursive prediction error identification and scaling of non-linear systems with midpoint numerical integration, in Proceedings of the 2010 American Control Conference, (2010), 4510–4515. IEEE.
    [43] R. Tóth, Modeling and identification of linear parameter-varying systems, vol. 403, Springer, 2010.
    [44] J. C. Tudón-Martínez, S. Fergani, O. Sename, J. J. Martinez, R. Morales-Menendez, L. Dugard, Adaptive road profile estimation in semiactive car suspensions, IEEE T. Contr. Syst. T., 23 (2015), 2293–2305.
    [45] J. Umenberger, J. Wågberg, I. R. Manchester, T. B. Schön, Maximum likelihood identification of stable linear dynamical systems, Automatica, 96 (2018), 280–292.
    [46] A. Unger, F. Schimmack, B. Lohmann, R. Schwarz, Application of LQ-based semi-active suspension control in a vehicle, Control Eng. Pract., 21 (2013), 1841–1850.
    [47] T. Utz, C. Fleck, J. Frauhammer, D. Seiler-Thull, A. Kugi, Extended kalman filter and adaptive backstepping for mean temperature control of a three-way catalytic converter, Int. J. Robust Nonlin., 24 (2014), 3437–3453.
    [48] C. Vivas-Lopez, D. H. Alcántara, M. Q. Nguyen, S. Fergani, G. Buche, O. Sename, L. Dugard, R. Morales-Menéndez, INOVE: a testbench for the analysis and control of automotive vertical dynamics, in 14th International Conference on Vehicle System Dynamics, Identification and Anomalies, VSDIA, 2014, pp–403.
    [49] Z. Wang, X. Liu, Y. Liu, J. Liang, V. Vinciotti, An extended kalman filtering approach to modeling nonlinear dynamic gene regulatory networks via short gene expression time series, IEEE/ACM T. Comput. Bi., 6 (2009), 410–419.
    [50] C.-Y. Wu, J.-H. Tsai, S.-M. Guo, L.-S. Shieh, J. I. Canelon, F. Ebrahimzadeh, L. Wang, A novel on-line observer/kalman filter identification method and its application to input-constrained active fault-tolerant tracker design for unknown stochastic systems, J. Franklin I., 352 (2015), 1119–1151.
    [51] Y. Zhao, B. Huang, H. Su, J. Chu, Prediction error method for identification of lpv models, J. Process Contr., 22 (2012), 180–193.
  • This article has been cited by:

    1. Zaineb Aloui, Kamel Brahim, Fractional Fourier Transform, Signal Processing and Uncertainty Principles, 2023, 42, 0278-081X, 892, 10.1007/s00034-022-02138-9
    2. Kader Ali Ibrahim, Feng Ding, Parameter estimation methods for time‐invariant continuous‐time systems from dynamical discrete output responses based on the Laplace transforms, 2024, 38, 0890-6327, 3213, 10.1002/acs.3871
    3. Ashish Kumar Jain, Input delay compensation of nonlinear systems with uncertain time-varying parameters, 2024, 80, 09473580, 101100, 10.1016/j.ejcon.2024.101100
    4. Jin Liu, Zan Li, Yue Gao, Qiguang Miao, Li Yang, Danyang Wang, Peihan Qi, Adaptive stochastic resonance enhanced weak linear frequency modulated signal perception in low signal-to-noise ratio environments, 2025, 36, 0957-0233, 016102, 10.1088/1361-6501/ad8255
    5. V. Soni Ishwarya, Mohanaprasad Kothandaraman, A Novel Feature-Fusion-Based Sparse Masked Attention Network for Acoustic Echo Cancellation Using Wavelet and STFT Synergies, 2024, 0278-081X, 10.1007/s00034-024-02955-0
    6. Muhammad Jawad, Husna Zafar, Zhuojia Fu, Sultan Alshehery, Muhammad Junaid, Ilyas Khan, Physics-informed neural networks for rotating EMHD flow of Jeffrey hybrid nanofluid with arrhenius activation energy and mass convections, 2025, 09574174, 126517, 10.1016/j.eswa.2025.126517
  • Reader Comments
  • © 2021 the Author(s), licensee AIMS Press. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0)
通讯作者: 陈斌, bchen63@163.com
  • 1. 

    沈阳化工大学材料科学与工程学院 沈阳 110142

  1. 本站搜索
  2. 百度学术搜索
  3. 万方数据库搜索
  4. CNKI搜索

Metrics

Article views(2886) PDF downloads(161) Cited by(6)

Figures and Tables

Figures(11)  /  Tables(4)

Other Articles By Authors

/

DownLoad:  Full-Size Img  PowerPoint
Return
Return

Catalog