Estimation of most optimal azimuthal angles for maximum PV solar efficiency using multivariate adaptive regression splines (MARS)

Gokhan Sahin; W.G.J.H.M. Van Wilfried Sark; Gokhan Sahin; W.G.J.H.M. Van Wilfried Sark

doi:10.3934/energy.2023060

AIMS Energy

2023, Volume 11, Issue 6: 1328-1353. doi: 10.3934/energy.2023060

Previous Article Next Article

Research article

Estimation of most optimal azimuthal angles for maximum PV solar efficiency using multivariate adaptive regression splines (MARS)

Copernicus Institute of Sustainable Development, Utrecht University, Princetonlaan 8A, 3584 CB Utrecht/The Netherlands

Received: 07 July 2023 Revised: 29 August 2023 Accepted: 11 October 2023 Published: 01 December 2023

The aim of this study was to build a regression model of solar irradiation in the Kulluk region of Turkey by using the multivariate adaptive regression splines (MARS) technique. Using the well-known data mining algorithm, MARS, this study has explored a convenient prediction model for continuous response variables, i.e., average daily energy production from the given system (Ed), average monthly energy production from given system (Em), average daily sum of global irradiation per square meter (Hd) and average annual sum of global irradiation per square meter (Hm). Four continuous estimators are included to estimate Ed, Em, Hd and Hm: Estimated losses due to temperature and low irradiance (ESLOTEM), estimated loss due to angular reflection effect (ESLOANGREF), combined photovoltaic system loss (COMPVLOSS) and rated power of the photovoltaic system (PPVS). Four prediction models as constructed by implementing the MARS algorithm, have been obtained by applying the smallest generalized cross-validation (GCV) criterion where the means of penalty are defined as 1 and the backward pruning method for the package "earth" of R software is used. As a result, it can be suggested that the procedure of the MARS algorithm, which achieves the greatest predictive accuracy of 100% or nearly 100%, permits researchers to obtain some remarkable hints for ascertaining predictors that affect solar irradiation parameters. The coefficient of determination denoted as R² was estimated at the highest predictive accuracy to be nearly 1 for Ed, Em, Hd and Hm while the GCV values were found to be 0.000009, 0.018908, 0.000013 and 0.019021, respectively. The estimated results indicate that four MARS models with the first degree interaction effect have the best predictive performances for verification with the lowest GCV value.

Keywords:

Citation: Gokhan Sahin, W.G.J.H.M. Van Wilfried Sark. Estimation of most optimal azimuthal angles for maximum PV solar efficiency using multivariate adaptive regression splines (MARS)[J]. AIMS Energy, 2023, 11(6): 1328-1353. doi: 10.3934/energy.2023060

Related Papers:

[1]	Carlo Bianca . Differential equations frameworks and models for the physics of biological systems. AIMS Biophysics, 2024, 11(2): 234-238. doi: 10.3934/biophy.2024013
[2]	Carlo Bianca . Interplay and multiscale modeling of complex biological systems. AIMS Biophysics, 2022, 9(1): 56-60. doi: 10.3934/biophy.2022005
[3]	Carlo Bianca . Theoretical frameworks and models for biological systems. AIMS Biophysics, 2020, 7(3): 167-168. doi: 10.3934/biophy.2020013
[4]	Carlo Bianca . Mathematical and computational modeling of biological systems: advances and perspectives. AIMS Biophysics, 2021, 8(4): 318-321. doi: 10.3934/biophy.2021025
[5]	Bradley Vis, Jonathan J. Powell, Rachel E. Hewitt . Imaging flow cytometry methods for quantitative analysis of label-free crystalline silica particle interactions with immune cells. AIMS Biophysics, 2020, 7(3): 144-166. doi: 10.3934/biophy.2020012
[6]	Mario P. Carante, Francesca Ballarini . Modelling cell death for cancer hadrontherapy. AIMS Biophysics, 2017, 4(3): 465-490. doi: 10.3934/biophy.2017.3.465
[7]	Mehmet Yavuz, Fuat Usta . Importance of modelling and simulation in biophysical applications. AIMS Biophysics, 2023, 10(3): 258-262. doi: 10.3934/biophy.2023017
[8]	Domenico Lombardo, Pietro Calandra, Maria Teresa Caccamo, Salvatore Magazù, Luigi Pasqua, Mikhail A. Kiselev . Interdisciplinary approaches to the study of biological membranes. AIMS Biophysics, 2020, 7(4): 267-290. doi: 10.3934/biophy.2020020
[9]	Nor Afiqah Mohd Aris, Siti Suhana Jamaian . Dynamical analysis of fractional-order chemostat model. AIMS Biophysics, 2021, 8(2): 182-197. doi: 10.3934/biophy.2021014
[10]	Raghvendra P Singh, Guillaume Brysbaert, Marc F Lensink, Fabrizio Cleri, Ralf Blossey . Kinetic proofreading of chromatin remodeling: from gene activation to gene repression and back. AIMS Biophysics, 2015, 2(4): 398-411. doi: 10.3934/biophy.2015.4.398

Abstract

1. Introduction

In the last decades the interest towards the complex systems for the modeling and studying of biological systems has been ever growing (see among the others [1]–[11], and references therein).

More generally, the study of complex systems [12]–[15] is one of the main resaerch topics of the last years. However the description of such systems, with related quantities and parameters, depends on the particular approach that is chosen. Different mathematical frameworks have been proposed depending on the representation scale of the system. In the present paper, we turn our attention on the so-called “kinetic approach” which is a generalization of the model proposed by Boltzmann [16] to describe the statistical dynamics of gases, and is based on a suitable version of his equation. The interacting entities of the system are called active particles. The evolution of the system is described by a distribution function which depends on the time, on the mechanical variables (i.e. space and velocity) and on a scalar variable called “activity”, whose meaning depends on the current application [17]. In the kinetic theoretical description of a complex system, the evolution depends on an integral term that defines the interactions between the particles.

A new modeling framework has been recently proposed for the description of a complex system under the action of an external force field: the thermostatted kinetic theory [18], [19]. The action of an external force field moves the systems out of the equilibrium. In this framework, the complex system under investigation is divided into n functional subsystems such that particles belonging to the same functional subsystem share the same strategy (in a suitable sense, the same aim) [20]. The macroscopic state is described by specific pth-order moments (see Section 2). The introduction of a dissipative term, called thermostat, constrains the system to keep the 2nd-order moment of the system, which can be regarded as the physical global activation energy, costant in time. The evolution of the system is then described by a system of nonlinear integro-differential equations with quadratic nonlinearity. The thermostatted kinetic theory has been developed, for instance, in the study of Kac equation too [21].

A biological system is constituted by a large number of interacting entities (the active particles) whose microscopic state can be described by a real variable (activity) which represents the individual ability to express a specific strategy. The active particles of a biological systems have the ability to develop behaviour that cannot only be explained by the classical mechanics laws, and, in some cases, can generate proliferative and/or destructive processes. By using the functional subsystems, the complexity can be reduced by decomposing the biological systems into several interacting subsystems [22]. In fact, this approach can possibly be considered the first fundamental contribution to biological studies [23]. The decomposition method can be regarded as a tool to reduce complexity. In fact, the active particles in each module, which is described as a functional subsystem, are not of the same type, while they express the same strategy collectively. Thus the system can be studied regarding the evolution of each functional subsystem rather than the single active particle.

A biological system can be described at different representation scales, depending on the particular application. For instance, the microscopic scale in biology corresponds to cells, while the dynamics of cells depends on the dynamics at the lower molecular scale, namely the scale of molecules. In some cases, the complexity of the system induces the use of even a larger scale, the macroscopic scale, which corresponds to population dynamics [24]–[26]. By using the framework of kinetic theory, the microscopic scale models the interactive dynamics among active particles, and it is described by using the activity variable, while the macroscopic state of the whole system is described by the pth-order moments related to the distribution function. The description of the biological systems depends on the pairwise interactions among the active particles (microscopic scale) and the state of the overall system (macroscopic scale).

In the last years, a widely interesting biological application is the modeling of the dynamics of epidemics with virus variations (see [3]–[6], [27]–[30], and references therein). One of the most used model is SIR (Susceptible, Infectious, Recovered), with its variants (see [6], [31]–[36], and references therein). In this contest, the system may be decomposed into 3 functional subsystem of interacting individual. The first functional subsystems denotes the healthy individuals whose microscopic state models the susceptibility to contract the pathological state. The second functional subsystems represents denotes individuals healthy carriers of the virus, whose microscopic state models the infectivity of the virus. The thirs functional subsystem denotes the individuals affected by the virus at the first and subsequent stages, whose microscopic state models the progression of both the infectivity and the pathological state.

Beyond biological systems, the thermostatted framework is used to study, among other topics: socio-economic systems [37], [38], pedestrian dynamics [39], vehicular traffic [40], crowd dynamics [41], human feelings [42], [43] and opinion formation [44].

The present paper deals with the discrete thermostatted framework: the activity variable attains its value in a discrete subset of ℝ, so that in this case the evolution of the system is modeled by a system of nonlinear ordinary differential equations with quadratic nonlinearity. The existence and uniqueness of the solutions is assured for both the related Cauchy and nonequilibrium stationary problems [45], [46]. The analysis of the discrete framework is important in order to define the methods for the numerical simulations.

The use of a discrete variable is not only a technical choice. In some application, this is related to the description of the current biological systems [47], [48]. For instance in order to describe the state of a cell, three values are assigned to the activity variable: normal, infected, dead. Moreover, in socio-economic systems [37], [38], the discretization of the activity variable has conceptual reason. As a matter of fact, a continuous activity variable for the socio-economic quantities may not make any sense: e.g. the wealth-state.

The main aim of this paper is the study of stability in the Hadamard sense of the discrete themostatted kinetic framework. A model is stable in the Hadamard sense if the solution exists and is unique, and it depends continuously on the initial data. In the current paper, two Cauchy problems, related to the discrete thermostatted framework that differ for their initial data, are considered. The distance between the initial data is estimated by a δ > 0. Using typical analytical techniques, this paper shows that the distance between the corresponding solutions, in a suitable norm, is estimated continuously in function of δ (see [49] for the continuous case), i.e. if δ goes to 0, the two solutions collapse.

The importance of continuous dependence of the solutions on the initial data is related, among other things, to the numerical simulations that can be performed [7], [50]–[52]. As a matter of fact, since the initial data is in general derived from a statistical analysis, its value is affected by an error, then the stability of the model may assure a “slow and small propagation” of such an error during the evolution, at least for small time intervals. Among others, the application of the discrete thermostatted framework to the epidemic dynamics may be considered. In this case the initial data is obtained from a statistical study, then it may be affected by an error. The continuous dependence with respect to the initial data is an important issue such that the numerical simulations can be performed such that the solution represents the evolution of epidemic with sufficient accuracy.

This paper presents a first step towards the study of stability and dependence on the initial data for the discrete kinetic thermostatted framework.

The contents of this paper are divided into 5 more sections which follow this brief Introduction. In Section 2 the discrete thermostatted kinetic framework is presented, and the related state of art about existence and uniqueness of the solutions. In Section 3 the main Theorem about the dependence on the initial data is presented, after introducing suitable norms. Section 4 deals with the proof of Theorem, proving the stability and giving an explicit form to the constant in the final inequality which is obtained by performing some typical analytic arguments. Finally Section 5 deals with the future research perspective that may follow this paper, which is meant as a first step towards the dependence on the initial data for the discrete thermostatted kinetic framework.

2. The discrete thermostatted framework

Let us consider a complex system 𝒞 homogeneous with respect to the mechanical variables, i.e. space and velocity, which is divided into n (where n ∈ ℕ is fixed) functional subsystems such that particles belonging to the same functional subsystem share the same strategy [20]. Since some physical and life sciences models are characterized by discrete variables, in this paper the microscopic state of the system is described by a scalar variable u, called activity, which attains its values in a fixed discrete subset of ℝ, i.e.

$u \in I_{u} : = {u_{1}, u_{2}, \dots, u_{n}} \subseteq ℝ .$

The ith functional subsystem, for i ∈ {1, 2, …, n}, is described by:

$f_{i} (t) : [0, + \infty [\to ℝ^{+},$

which is the discrete distribution function, at time t > 0, of active particles with activity u_i. The distribution function of the whole system reads:

$f (t, u) = \sum_{i = 1}^{n} f_{i} (t) δ (u - u_{i}),$

where δ is the Delta function. The quantity f(t,u)du represents the number of active particles, at time t > 0, in the elementary volume [u, u + du].

The related vector distribution function of the system 𝒞 reads:

$f = f (t) : = (f_{1} (t), f_{2} (t), \dots, f_{n} (t)) .$

The macroscopic state of the system, at a time t > 0, is described by the pth-order moment, which is defined, for p ∈ ℕ, as:

$𝔼_{p} [f] (t) = \sum_{i = 1}^{n} u_{i}^{p} f_{i} (t) .$

The physical meaning of the moment depends on the value of p, e.g. the local density, the linear activity momentum and the global activation energy are obtained, respectively, for p = 0, p = 1 and p = 2.

The interaction among the active particles is described by the following parameters:

$\sum_{i = 1}^{n} B_{h k}^{i} = 1.$ (2.1)

Let F_i(t) : [0; +∞[→ ℝ⁺, for i ∈ {1,2, …, n}, be an external force acting on the ith functional subsystem, such that the external force field writes: $F (t) = (F_{1} (t), F_{2} (t), \dots, F_{n} (t))$ .

A discrete thermostat is included in order to keep constant the 2nd-order moment 𝔼₂[f](t). Let 𝔼₂ be the fixed value of the 2nd-order moment, then the evolution of the ith functional subsystem, for i ∈ {1,2, …, n}, is:

$\frac{d f_{i}}{d t} (t) = J_{i} [f] (t) + F_{i} (t) - \sum_{j = 1}^{n} (\frac{u_{j}^{2} (J_{j} [f] + F_{j})}{𝔼_{2}}) f_{i} (t),$ (2.2)

where the interaction operator J_i[f](t) can be split as follows:

$J_{i} [f] (t) = G_{i} [f] (t) - L_{i} [f] (t),$

where

$G_{i} [f] (t) = \sum_{h = 1}^{n} \sum_{k = 1}^{n} η_{h k} B_{h k}^{i} f_{h} (t) f_{k} (t)$

and

$L_{i} [f] (t) = f_{i} (t) \sum_{k = 1}^{n} η_{i k} f_{k} (t)$

are the gain term operator and the loss term operator (whose expression is related to the (2.1)), respectively. The equation (2.2) is obtained by balancing the time-derivative of the distribution function related to the ith functional subsystem, i.e f_i(t), and the net flux, J_i[f](t), related to the same functional subsystem, the external force field and the thermostat. Therefore the evolution of the system (2.2) is a system of nonlinear ordinary differential equations with quadratic nonlinearity.

Let $f^{0} : = (f_{1}^{0}, f_{2}^{0}, \dots, f_{n}^{0})$ be a suitable initial data. Then the Cauchy problem related to the discrete thermostatted framework (2.2) is defined as:

$(\begin{array}{l} (2.2) & i \in {1,2, \dots, n}, t \in [0, + \infty [ \\ f (0) = f^{0} . \end{array}$ (2.3)

Definition 2.1. Fixed 𝔼₂ > 0, the space function $ℛ_{f}^{2} = ℛ_{f}^{2} (ℝ^{+}, 𝔼_{2})$ is defined as:

${f (t) \in C ([0, + \infty [; {(ℝ^{+})}^{n}) : 𝔼_{2} [f] (t) = 𝔼_{2}, \forall t} .$

Henceforth, the following assumptions are taken into account:

$u_{i} \geq 1, \forall i \in {1,2, \dots, n};$

$𝔼_{2} [f^{0}] = 𝔼_{2};,$

H3 there exists η > 0 such that η_hk = η, for all h, k ∈ {1,2, …, n};

H4 there exists F > 0 such that F_i(t) = F, for all i ∈ {1,2, …, n} and for all t > 0.

Remark 2.2. The H1 is a technical assumption since the generalized framework has been treated in [53], where u_i ≥ 1 is not required.

In this framework, under the assumptions H1-H4, there exists one only function $f (t) \in C ([0, + \infty [; {(ℝ^{+})}^{n}) \cap ℛ_{f}^{2}$ which is solution of the Cauchy problem (2.3) [45].

Complex systems usually operate far from equilibrium since their evolution is related to an internal dynamics, due to the interaction between particles, and to an external dyanmics, due to the external force field. Then nonequilibrium stationary states are reached during the evolution.

The stationary problem related to (2.2) is, for i ∈ {1,2, …, n}:

$J_{i} [f] (t) + F_{i} - \sum_{j = 1}^{n} (\frac{u_{j}^{2} (J_{j} [f] + F_{j})}{𝔼_{2}}) f_{i} (t) = 0.$ (2.4)

In [46] the existence of a solution of (2.4), called non-equilibrium stationary solution, is gained, and the uniqueness is proved under some restrictions on the value of the external force F. Henceforth, without leading of generality, 𝔼₂[f] = 𝔼₀[f] = 1 is assumed.

If these further assumptions hold true:

$\sum_{i = 1}^{n} u_{i} B_{h k}^{i} = 0, \forall h, k \in {1,2, \dots, n},$

$\sum_{i = 1}^{n} u_{i}^{2} B_{h k}^{i} = u_{h}^{2}, \forall h, k \in {1,2, \dots, n},$

then the evolution equation of 𝔼₁[f](t) reads [46]:

$𝔼_{1} [f] (t) = 𝔼_{1} [f^{0}] e^{- ϕ (t)} + k,$

where

$ϕ (s) = \int_{0}^{s} (η + \sum_{j = 1}^{n} u_{j}^{2} F_{j}) d τ,$

and

$k = \frac{\sum_{i = 1}^{n} u_{i} F_{i}}{η + \sum_{i = 1}^{n} u_{i}^{2} F_{i}} .$

Moreover:

$𝔼_{1} [f] (t) \overset{t \to \infty}{\to} k .$

Remark 2.3. The assumptions H3-H4 are not restricted. Indeed the results of existence and, if possible, uniqueness can be proved if there exist F,η > 0 such that:

η_hk ≤ η, for all h, k ∈ {1,2, …, n};
F_i(t) ≤ F, for all i ∈ {1,2, …, n} and for all t > 0.

3. Dependence on the initial data

Under suitable assumptions the Cauchy problem (2.3) has a unique solution $f (t) \in C ([0, + \infty [; {(ℝ^{+})}^{n}) \cap ℛ_{f}^{2}$ [45].

Let us consider now two Cauchy problems related to the discrete thermostatted framework (2.2):

$(\begin{array}{l} (2.2) & t \in [0, + \infty [ \\ f (0) = f_{1}^{0} \end{array} (\begin{array}{l} (2.2) & t \in [0, + \infty [ \\ f (0) = f_{2}^{0} \end{array}$ (3.1)

where

$f_{1}^{0} = (f_{11}^{0}, f_{12}^{0}, \dots, f_{1 n}^{0})$ and

$f_{2}^{0} = (f_{21}^{0}, f_{22}^{0}, \dots, f_{2 n}^{0})$ are two different initial data.

If the assumptions H1-H6 hold true, there exist $f_{1} (t), f_{2} (t) \in C ([0, + \infty [; {(ℝ^{+})}^{n}) \cap ℛ_{f}^{2}$ solutions of the Cauchy problems (3.1) with initial data $f_{1}^{0}$ and $f_{2}^{0}$ , respectively. The main result of this paper estimates the distance between the two solutions for T > 0, i.e.

$\begin{array}{l} {‖ f_{1} (t) - f_{2} (t) ‖}_{C ([0, T]; {(ℝ^{+})}^{n})} & : = \underset{t \in [0, T]}{m a x} {‖ f_{1} (t) - f_{2} (t) ‖}_{1} \\ = \underset{t \in [0, T]}{m a x} \sum_{i = 1}^{n} | f_{1 i} (t) - f_{2 i} (t) |, \end{array}$

when the distance between the initial data is estimated, i.e. there exists δ > 0 such that

$‖ f_{1}^{0} - f_{2}^{0} ‖ : = \sum_{i = 1}^{n} | f_{1 i}^{0} - f_{2 i}^{0} | \leq δ .$

Theorem 3.1. Consider the Cauchy problems (3.1). Suppose that the assumptions H1-H6 hold true. If there exists δ > 0 such that:

${‖ f_{1}^{0} - f_{2}^{0} ‖}_{1} \leq δ,$ (3.2)

then for T > 0

${‖ f_{1} (t) - f_{2} (t) ‖}_{C ([0, T]; {(ℝ^{+})}^{n})} \leq δ e^{C T},$ (3.3)

where the constant C = C(η,F,I_u) depends on the parameters of the system.

Remark 3.2. As consequence of Theorem 3.1, the problem (2.3) is well-posed in the Hadamard sense.

4. Proof of main result

Proof of Theorem 3.1. The thermostatted equation (2.2), for i ∈ {1, 2, …, n}, writes:

$\begin{array}{l} \frac{d f_{i}}{d t} (t) & = \sum_{h = 1}^{n} \sum_{k = 1}^{n} η B_{h k}^{i} f_{h} (t) f_{k} (t) - f_{i} (t) \sum_{k = 1}^{n} η f_{k} (t) + F \\ - \sum_{j = 1}^{n} (u_{j}^{2} (J_{j} [f] + F_{j})) f_{i} (t) . \end{array}$ (4.1)

Integrating between 0 and t, the (4.1) reads:

$\begin{array}{l} \int_{0}^{t} \frac{d f_{i}}{d τ} (τ) d τ & = η \int_{0}^{t} \sum_{h = 1}^{n} \sum_{k = 1}^{n} B_{h k}^{i} f_{h} (τ) f_{k} (τ) d τ - η \int_{0}^{t} f_{i} (τ) d τ \\ + \int_{0}^{t} F d τ - \sum_{j = 1}^{n} (u_{j}^{2} (J_{j} [f] + F_{j})) \int_{0}^{t} f_{i} (τ) d τ . \end{array}$ (4.2)

By assumptions:

$\sum_{j = 1}^{n} (u_{j}^{2} (J_{j} [f] + F_{j})) = \sum_{j = 1}^{n} u_{j}^{2} F_{j},$

and then the (4.2) rewrites:

$\begin{array}{l} f_{i} (t) - f_{i} (0) & = η \int_{0}^{t} \sum_{h = 1}^{n} \sum_{k = 1}^{n} B_{h k}^{i} f_{h} (τ) f_{k} (τ) d τ - η \int_{0}^{t} f_{i} (τ) d τ \\ + F t - (\sum_{j = 1}^{n} u_{j}^{2} F_{j}) \int_{0}^{t} f_{i} (τ) d τ . \end{array}$ (4.3)

Finally, by (4.3), for i ∈ {1, 2, …, n}:

$\begin{array}{l} f_{i} (t) & = f_{i} (0) + η \int_{0}^{t} \sum_{h = 1}^{n} \sum_{k = 1}^{n} B_{h k}^{i} f_{h} (τ) f_{k} (τ) d τ - η \int_{0}^{t} f_{i} (τ) d τ \\ + F t - (\sum_{j = 1}^{n} u_{j}^{2} F_{j}) \int_{0}^{t} f_{i} (τ) d τ . \end{array}$ (4.4)

Using the (4.4) for $f_{1} (t), f_{2} (t) \in C ([0, + \infty [; {(ℝ^{+})}^{n}) \cap ℛ_{f}^{2}$ solutions of the Cauchy problems (3.1) with initial data $f_{1}^{0}$ and $f_{2}^{0}$ , respectively, one has, for i ∈ {1, 2, …, n}:

$\begin{array}{l} f_{1 i} (t) & = f_{1 i}^{0} + η \int_{0}^{t} \sum_{h = 1}^{n} \sum_{k = 1}^{n} B_{h k}^{i} f_{1 h} (τ) f_{1 k} (τ) d τ - η \int_{0}^{t} f_{1 i} (τ) d τ \\ + F t - (\sum_{j = 1}^{n} u_{j}^{2} F_{j}) \int_{0}^{t} f_{1 i} (τ) d τ \end{array}$ (4.5)

and

$\begin{array}{l} f_{2 i} (t) & = f_{2 i}^{0} + η \int_{0}^{t} \sum_{h = 1}^{n} \sum_{k = 1}^{n} B_{h k}^{i} f_{2 h} (τ) f_{2 k} (τ) d τ - η \int_{0}^{t} f_{2 i} (τ) d τ \\ + F t - (\sum_{j = 1}^{n} u_{j}^{2} F_{j}) \int_{0}^{t} f_{2 i} (τ) d τ . \end{array}$ (4.6)

Subtracting the (4.5) and (4.6), for i ∈ {1, 2, …, n}:

$\begin{array}{l} f_{1 i} (t) & - f_{2 i} (t) = (f_{1 i}^{0} - f_{2 i}^{0}) \\ + η \int_{0}^{t} \sum_{h = 1}^{n} \sum_{k = 1}^{n} B_{h k}^{i} (f_{1 h} (τ) f_{1 k} (τ) - f_{2 h} (τ) f_{2 k} (τ)) d τ \\ - η \int_{0}^{t} (f_{1 i} (τ) - f_{2 i} (τ)) d τ - (\sum_{j = 1}^{n} u_{j}^{2} F_{j}) \int_{0}^{t} (f_{1 i} (τ) - f_{2 i} (τ)) d τ . \end{array}$ (4.7)

By (4.7), for i ∈ {1, 2, …, n}:

$\begin{array}{l} | f_{1 i} (t) & - f_{2 i} (t) | \leq | f_{1 i}^{0} - f_{2 i}^{0} | \\ + η \int_{0}^{t} \sum_{h = 1}^{n} \sum_{k = 1}^{n} B_{h k}^{i} | f_{1 h} (τ) f_{1 k} (τ) - f_{2 h} (τ) f_{2 k} (τ) | d τ \\ + η \int_{0}^{t} | f_{1 i} (τ) - f_{2 i} (τ) | d τ + (\sum_{j = 1}^{n} u_{j}^{2} F_{j}) \int_{0}^{t} | f_{1 i} (τ) - f_{2 i} (τ) | d τ . \end{array}$ (4.8)

Moreover, for i ∈ {1, 2, …, n}:

$\begin{array}{l} | f_{1 i} (t) & - f_{2 i} (t) | \leq | f_{1 i}^{0} - f_{2 i}^{0} | \\ + η \int_{0}^{t} \sum_{h = 1}^{n} \sum_{k = 1}^{n} B_{h k}^{i} | f_{1 h} (τ) f_{1 k} (τ) - f_{2 h} (τ) f_{2 k} (τ) | d τ \\ + (η + \sum_{j = 1}^{n} u_{j}^{2} F_{j}) \int_{0}^{t} | f_{1 i} (τ) - f_{2 i} (τ) | d τ . \end{array}$ (4.9)

Summing the (4.9) on i:

$\begin{array}{l} \sum_{i = 1}^{n} | f_{1 i} (t) & - f_{2 i} (t) | \leq \sum_{i = 1}^{n} | f_{1 i}^{0} - f_{2 i}^{0} | \\ + η \int_{0}^{t} \sum_{i = 1}^{n} \sum_{h = 1}^{n} \sum_{k = 1}^{n} B_{h k}^{i} | f_{1 h} (τ) f_{1 k} (τ) - f_{2 h} (τ) f_{2 k} (τ) | d τ \\ + (η + \sum_{j = 1}^{n} u_{j}^{2} F_{j}) \int_{0}^{t} \sum_{i = 1}^{n} | f_{1 i} (τ) - f_{2 i} (τ) | d τ . \end{array}$ (4.10)

Then by (4.10):

$\begin{array}{l} ‖ f_{1} (t) & {- f_{2} (t) ‖}_{1} \leq {‖ f_{1}^{0} - f_{2}^{0} ‖}_{1} \\ + η \int_{0}^{t} \sum_{h = 1}^{n} \sum_{k = 1}^{n} | f_{1 h} (τ) f_{1 k} (τ) - f_{2 h} (τ) f_{2 k} (τ) | d τ \\ + (η + \sum_{j = 1}^{n} u_{j}^{2} F_{j}) \int_{0}^{t} {‖ f_{1} (τ) - f_{2} (τ) ‖}_{1} d τ . \end{array}$ (4.11)

Since, by straightforward calculations:

$\begin{array}{l} | f_{1 h} (τ) f_{1 k} (τ) - f_{2 h} (τ) f_{2 k} (τ) | \\ = | f_{1 h} (τ) f_{1 k} (τ) - f_{2 h} (τ) f_{1 k} (τ) + f_{2 h} (τ) f_{1 k} (τ) - f_{2 h} (τ) f_{2 k} (τ) | \\ \leq | f_{1 h} (τ) - f_{2 h} (τ) | | f_{1 k} (τ) | + | f_{1 k} (τ) - f_{2 k} (τ) | | f_{2 h} (τ) |, \end{array}$

then the second term of the right hand side of the (4.11) is estimated as follows:

$\begin{array}{l} \int_{0}^{t} \sum_{h = 1}^{n} \sum_{k = 1}^{n} | f_{1 h} (τ) f_{1 k} (τ) - f_{2 h} (τ) f_{2 k} (τ) | d τ \\ \leq \int_{0}^{t} \sum_{h = 1}^{n} \sum_{k = 1}^{n} (| f_{1 h} (τ) - f_{2 h} (τ) | | f_{1 k} (τ) | + | f_{1 k} (τ) - f_{2 k} (τ) | | f_{2 h} (τ) |) d τ \\ \leq \int_{0}^{t} \sum_{h = 1}^{n} | f_{1 h} (τ) - f_{2 h} (τ) | \sum_{k = 1}^{n} | f_{1 k} (τ) | d τ \\ + \int_{0}^{t} \sum_{k = 1}^{n} | f_{1 k} (τ) - f_{2 k} (τ) | \sum_{h = 1}^{n} | f_{2 h} (τ) | d τ \\ \leq 2 \int_{0}^{t} {‖ f_{1} (τ) - f_{2} (τ) ‖}_{1} d τ . \end{array}$ (4.12)

Using the (4.12) and (4.11) follows:

$\begin{array}{l} {‖ f_{1} (t) - f_{2} (t) ‖}_{1} & \leq {‖ f_{1}^{0} - f_{2}^{0} ‖}_{1} \\ + 2 η \int_{0}^{t} {‖ f_{1} (τ) - f_{2} (τ) ‖}_{1} d τ \\ + (η + \sum_{j = 1}^{n} u_{j}^{2} F_{j}) \int_{0}^{t} {‖ f_{1} (τ) - f_{2} (τ) ‖}_{1} d τ \\ \leq {‖ f_{1}^{0} - f_{2}^{0} ‖}_{1} + (3 η + \sum_{j = 1}^{n} u_{j}^{2} F_{j}) \int_{0}^{t} {‖ f_{1} (τ) - f_{2} (τ) ‖}_{1} d τ . \end{array}$ (4.13)

Let introduce the constant

$C (η, F, I_{u}) = C : = 3 η + \sum_{j = 1}^{n} u_{j}^{2} F_{j} = 3 η + F \sum_{j = 1}^{n} u_{j}^{2},$

that depends on the parameters of the system. Then, by the (3.2), the (4.13) rewrites:

${‖ f_{1} (t) - f_{2} (t) ‖}_{1} \leq δ + C \int_{0}^{t} {‖ f_{1} (τ) - f_{2} (τ) ‖}_{1} d τ .$ (4.14)

Applying the Gronwall Lemma [54] to the (4.14), one has:

${‖ f_{1} (t) - f_{2} (t) ‖}_{1} \leq δ e^{C t},$ (4.15)

for t ∈ (0, T).

Finally, by (4.15), the claim (3.3) is gained, since:

${‖ f_{1} (t) - f_{2} (t) ‖}_{C ([0, T]; {(ℝ^{+})}^{n})} = \underset{t \in [0, T]}{m a x} {‖ f_{1} (t) - f_{2} (t) ‖}_{1} \leq δ e^{C T} .$

5. Conclusion and research perspectives

The mathematical analysis performed in this paper has been addressed to the dependence on the initial data of the discrete thermostatted kinetic framework. Theorem 3.1 ensures the stability in the Hadarmd sense of the framework (2.2), for all T > 0. The constant C > 0, obtained after some technical computations, is explicit and is an important issue for future numerical simulations that may be performed for several applications (see among others [55], [56] and references therein).

Numerical simulations towards the framework (2.2) are a first future research prespective. Specifically, the parameters of the system, i.e. interaction rates η_hk and transition probability densities $B_{h k}^{i}$ , will have an explicit form and meaning related to the particular application taken into account. For instance, the modeling of the tumor-immune system competition may be considere. In this case the functional subsystems are: cells of immune system, normal cells and infected cells (see [57]–[59]). The initial distribution of cells, related to the three different functional subsystems, is gained by statistical analysis; then it may be affected by error. The solutions, related to the real and affected value of the initial data, differs of δ at time t = 0. By using the estimate (3.3), the maximal time time interval where the two solutions are as “close” as possible to each other is estimated.

For Theorem 3.1 the norm ${‖ f ‖}_{C ([0, T]; {(ℝ^{+})}^{n})}$ has been used in according to [18]. A research perspective is the estimate of the distance between the two solutions, during the time evolution, by using another suitable norm which may provide a sharper result. In this perspective, the estimate of distance between the time-derivatives of the two functions may be a next step.

Theorem 3.1 provides stability for the discrete framework when the 2nd-order moment is preserved. The general framework [45], when the generic pth-order moment is preserved, is not investigated in this paper. An interesting future research perspective is the analysis of the general case by using, in a first approach, the same norm of Theorem 3.1.

References

[1]	REN 21 (2019) Renewables 2019 global status report. Available from: https://www.ren21.net/gsr.
[2]	Despotovic M, Nedic V, Despotovic D, et al. (2016) Evaluation of empirical models for predicting monthly mean horizontal diffuse solar radiation. Renewable Sustainable Energy Rev 56: 246–260. https://doi.org/10.1016/j.rser.2015.11.058 doi: 10.1016/j.rser.2015.11.058
[3]	Jamil B, Akhtar N (2017) Comparison of empirical models to estimate monthly mean diffuse solar radiation from measured data: Case study for humid-subtropical climatic region of India. Renewable Sustainable Energy Rev 77: 1326–1342. https://doi.org/10.1016/j.rser.2017.02.057 doi: 10.1016/j.rser.2017.02.057
[4]	Filho EPM, Oliveira AP, Vita WA, et al. (2016) Global, diffuse and direct solar radiation at the surface in the city of Rio de Janeiro: Observational characterization and empirical modeling. Renewable Energy 91: 64–74. https://doi.org/10.1016/j.renene.2016.01.040 doi: 10.1016/j.renene.2016.01.040
[5]	Liao W, Wang X, Fan Q, et al. (2015) Long-term atmospheric visibility, sunshine duration and precipitation trends in South China. Atm Environ 107: 204–216. https://doi.org/10.1016/j.atmosenv.2015.02.015 doi: 10.1016/j.atmosenv.2015.02.015
[6]	Chelbi M, Gagnon Y, Waewsak J (2015) Solar radiation mapping using sunshine duration-based models and interpolation techniques: Application to Tunisia. Energy Conv Manage 101: 203–215. https://doi.org/10.1016/j.enconman.2015.04.052 doi: 10.1016/j.enconman.2015.04.052
[7]	Zhu W, Lu A, Jia S, et al. (2017) Retrievals of all-weather daytime air temperature from MODIS products. Remote Sens Environ 189: 152–163. https://doi.org/10.1016/j.rse.2016.11.011 doi: 10.1016/j.rse.2016.11.011
[8]	Ho HC, Knudby A, Xu Y, et al. (2016) A comparison of urban heat islands mapped using skin temperature, air temperature, and apparent temperature (Humidex), for the greater Vancouver area. Sci Total Environ 544: 929–938. https://doi.org/10.1016/j.scitotenv.2015.12.021 doi: 10.1016/j.scitotenv.2015.12.021
[9]	Wenbin Z, Aifeng L, Shaofeng J (2013) Estimation of daily maximum and minimum air temperature using MODIS land surface temperature products. Remote Sens Environ 130: 62–73. https://doi.org/10.1016/j.rse.2012.10.034 doi: 10.1016/j.rse.2012.10.034
[10]	Chang K, Zhang Q (2019) Improvement of the hourly global solar model and solar radiation for air-conditioning design in China. Renewable Energy 138: 1232–1238. https://doi.org/10.1016/j.renene.2019.02.069 doi: 10.1016/j.renene.2019.02.069
[11]	Jiang H, Lu N, Qin J, et al. (2019) A deep learning algorithm to estimate hourly global solar radiation from geostationary satellite data. Renewable Sustainable Energy Rev 114: 1–13. https://doi.org/10.1016/j.rser.2019.109327 doi: 10.1016/j.rser.2019.109327
[12]	Li DHW, Chen W, Li S, et al. (2019) Estimation of hourly global solar radiation using Multivariate Adaptive Regression Spline (MARS)—A case study of Hong Kong. Energy 186: 1–14. https://doi.org/10.1016/j.energy.2019.115857 doi: 10.1016/j.energy.2019.115857
[13]	Cornejo-Bueno L, Casanova-Mateo C, Sanz-Justo J, et al. (2019) Machine learning regressors for solar radiation estimation from satellite data. Sol Energy 183: 768–775. https://doi.org/10.1016/j.solener.2019.03.079 doi: 10.1016/j.solener.2019.03.079
[14]	Manju S, Sandeep M (2019) Prediction and performance assessment of global solar radiation in Indian cities: A comparison of satellite and surface measured data. J Clean Prod 230: 116–128. https://doi.org/10.1016/j.jclepro.2019.05.108 doi: 10.1016/j.jclepro.2019.05.108
[15]	Gouda SG, Hussein Z, Luo S, et al. (2019) Model selection for accurate daily global solar radiation prediction in China. J Clean Prod 221: 132–144. https://doi.org/10.1016/j.jclepro.2019.02.211 doi: 10.1016/j.jclepro.2019.02.211
[16]	Guermoui M, Melgani F, Danilo C (2018) Multi-step ahead forecasting of daily global and direct solar radiation: A review and case study of Ghardaia region. J Clean Prod 201: 716–734. https://doi.org/10.1016/j.jclepro.2018.08.006 doi: 10.1016/j.jclepro.2018.08.006
[17]	Feng Y, Gong D, Zhang Q, et al. (2019) Evaluation of temperature-based machine learning and empirical models for predicting daily global solar radiation. Energy Conver Manag 198: 111780. https://doi.org/10.1016/j.enconman.2019.111780 doi: 10.1016/j.enconman.2019.111780
[18]	Makade RG, Chakrabarti S, Jamil B, et al. (2020) Estimation of global solar radiation for the tropical wet climatic region of India: A theory of experimentation approach. Renewable Energy 146: 2044–2059. https://doi.org/10.1016/j.renene.2019.08.054 doi: 10.1016/j.renene.2019.08.054
[19]	Kisi O, Heddam S, Yaseen ZM (2019) The implementation of univariable scheme-based air temperature for solar radiation prediction: New development of dynamic evolving neural-fuzzy inference system model. Appl Energy 241: 184–195. https://doi.org/10.1016/j.apenergy.2019.03.089 doi: 10.1016/j.apenergy.2019.03.089
[20]	Kaplan AG, Kaplan YA (2020) Developing of the new models in solar radiation estimation with curve fitting based on moving least-squares approximation. Renewable Energy 146: 2462–2471. https://doi.org/10.1016/j.renene.2019.08.095 doi: 10.1016/j.renene.2019.08.095
[21]	Anis MS, Jamil B, Ansari MA, et al. (2019) Generalized models for estimation of global solar radiation based on sunshine duration and detailed comparison with the existing: A case study for India. Sustainable Energy Technol Assess 31: 179–198. https://doi.org/10.1016/j.seta.2018.12.009 doi: 10.1016/j.seta.2018.12.009
[22]	Gürel AE, Ağbulut U, Biçen Y (2020) Assessment of machine learning, time series, response surface methodology and empirical models in prediction of global solar radiation. J Clean Prod 277: 122353. https://doi.org/10.1016/j.jclepro.2020.122353 doi: 10.1016/j.jclepro.2020.122353
[23]	El-Baz W, Tzscheutschler P, Wagner U (2018) Day-ahead probabilistic PV generation forecast for buildings energy management systems. Sol Energy 171: 478–490. https://doi.org/10.1016/j.solener.2018.06.100 doi: 10.1016/j.solener.2018.06.100
[24]	Hu K, Cao S, Wang L, et al. (2018) A new ultra-short-term photovoltaic power prediction model based on ground-based cloud images. J Clean Prod 200: 731–745. https://doi.org/10.1016/j.jclepro.2018.07.311 doi: 10.1016/j.jclepro.2018.07.311
[25]	Dong J, Olama MM, Kuruganti T, et al. (2020) Novel stochastic methods to predict short-term solar radiation and photovoltaic power. Renewable Energy 145: 333–346. https://doi.org/10.1016/j.renene.2019.05.073 doi: 10.1016/j.renene.2019.05.073
[26]	Gao M, Li J, Hong F, et al. (2019) Day-ahead power forecasting in a large-scale photovoltaic plant based on weather classification using LSTM. Energy 187: 115838. https://doi.org10.1016/j.energy.2019.07.168 doi: 10.1016/j.energy.2019.07.168
[27]	Gulin M, Pavlovic T, Vašak M (2017) A one-day-ahead photovoltaic array power production prediction with combined static and dynamic on-line correction. Sol Energy 142: 49–60. https://doi.org/10.1016/j.solener.2016.12.008 doi: 10.1016/j.solener.2016.12.008
[28]	Wang K, Qi X, Liu H (2019) A comparison of day-ahead photovoltaic power forecasting models based on deep learning neural network. Appl Energy 251: 1–14. https://doi.org/10.1016/j.apenergy.2019.113315 doi: 10.1016/j.apenergy.2019.113315
[29]	Wang G, Su Y, Shu L (2016) One-day-ahead daily power forecasting of photovoltaic systems based on partial functional linear regression models. Renewable Energy 96: 469–478. https://doi.org/10.1016/j.renene.2016.04.089 doi: 10.1016/j.renene.2016.04.089
[30]	Han S, Qiaoa YH, Yan J, et al. (2019) Mid-to-long term wind and photovoltaic power generation prediction based on copula function and long short term memory networke. Appl Energy 239: 181–191. https://doi.org/10.1016/j.apenergy.2019.01.193 doi: 10.1016/j.apenergy.2019.01.193
[31]	Yang XS, Deb S (2009) Cuckoo search via Lévy flights. 2009 World Congress on Nature & Biologically Inspired Computing (NaBIC), Coimbatore, India, 210–214. https://doi.org/10.1109/NABIC.2009.5393690
[32]	Wang H, Yi H, Peng J, et al. (2017) Deterministic and probabilistic forecasting of photovoltaic power based on deep convolutional neural network. Energy Conver Manag 153: 409–422. https://doi.org/10.1016/j.enconman.2017.10.008 doi: 10.1016/j.enconman.2017.10.008
[33]	Sharifzadeh M, Sikinioti-Locka A, Shah N (2019) Machine-learning methods for integrated renewable power generation: A comparative study of artificial neural networks, support vector regression, and gaussian process regression. Renewable Sustainable Energy Rev 108: 513–538. https://doi.org/10.1016/j.rser.2019.03.040 doi: 10.1016/j.rser.2019.03.040
[34]	Wood DA (2019) German solar power generation data mining and prediction with transparent open box learning network integrating weather, environmental and market variables. Energy Conver Manage 196: 354–369. https://doi.org/10.1016/j.enconman.2019.05.114 doi: 10.1016/j.enconman.2019.05.114
[35]	Heydari A, Garcia DA, Keynia F, et al. (2019) A novel composite neural network based method for wind and solar power forecasting in microgrids. Appl Energy 251: 1–17. https://doi.org/10.1016/j.apenergy.2019.113353 doi: 10.1016/j.apenergy.2019.113353
[36]	Zhang WG, Goh ATC (2012) Multivariate adaptive regression splines for analysis of geotechnical engineering systems. Comput Geotech 48: 82–95. https://doi.org/10.1016/j.compgeo.2012.09.016 doi: 10.1016/j.compgeo.2012.09.016
[37]	Zhang G, Bateni SM, Jun C, et al. (2022) Feasibility of random forest and multivariate adaptive regression splines for predicting long-term mean monthly dew point temperature. Front Environ Sci 10: 826165. https://doi.org/10.3389/fenvs.2022.826165 doi: 10.3389/fenvs.2022.826165
[38]	Raj N, Gharineiat Z (2021) Evaluation of multivariate adaptive regression splines and artificial neural network for prediction of mean sea level trend around northern Australian coastlines. Mathematics 9: 2696. https://doi.org/10.3390/math9212696 doi: 10.3390/math9212696
[39]	Zhang WG, Goh ATC (2013) Multivariate adaptive regression splines for analysis of geotechnical engineering systems. Comput Geotech 48: 82–95. https://doi.org/10.1016/j.compgeo.2012.09.016 doi: 10.1016/j.compgeo.2012.09.016
[40]	Dhimish M, Silvestre S (2019) Estimating the impact of azimuth-angle variations on photovoltaic annual energy production. Clean Energy 3: 47–58. https://doi.org/10.1093/ce/zky022 doi: 10.1093/ce/zky022
[41]	Li DHW, Chen W, Li S, et al. (2019) Estimation of hourly global solar radiation using Multivariate Adaptive Regression Spline (MARS)—A case study of Hong Kong. Energy 186: 115857. https://doi.org/10.1016/j.energy.2019.115857 doi: 10.1016/j.energy.2019.115857
[42]	Srivastava R, Tiwari AN, Giri VK (2019) Solar radiation forecasting using MARS, CART, M5, and random forest model: A case study for India. Heliyon 5: e02692. https://doi.org/10.1016/j.heliyon.2019.e02692 doi: 10.1016/j.heliyon.2019.e02692
[43]	Turk S, Sahin G (2020) Multi-criteria decision-making in the location selection for a solar PV power plant using AHP. Measurement 153: 107384. https://doi.org/10.1016/j.measurement.2019.107384 doi: 10.1016/j.measurement.2019.107384
[44]	Sahin G, Isik G, van Sark W (2023) Predictive modeling of PV solar power plant efficiency considering weather conditions: A comparative analysis of artificial neural networks and multiple linear regression. Energy Rep 10: 2837–2849. https://doi.org/10.1016/j.egyr.2023.09.097 doi: 10.1016/j.egyr.2023.09.097
[45]	Friedman JH (2011) Multivariate adaptive regression splines. The Ann Stat 19: 14122.
[46]	Duzen H, Aydin H (2012) Sunshine-based estimation of global solar radiation on horizontal surface at Lake Van region (Turkey). Energy Convers Manage 58: 35e46. https://doi.org/10.1016/j.enconman.2011.11.028 doi: 10.1016/j.enconman.2011.11.028
[47]	Bou-Rabeea M, Sulaimanb SA, Salehc MS, et al. (2017) Using artificial neural networks to estimate solar radiation in Kuwait. Renewable Sustainable Energy Rev 72: 434–438. https://doi.org/10.1016/j.rser.2017.01.013 doi: 10.1016/j.rser.2017.01.013
[48]	Li DHW, Chen W, Li S, et al. (2019) Estimation of hourly global solar radiation using Multivariate Adaptive Regression Spline (MARS)—A case study of Hong Kong. Energy 186: 115857. https://doi.org/10.1016/j.energy.2019.115857 doi: 10.1016/j.energy.2019.115857
[49]	Mohammad Rezaie-Balf M, Maleki N, Kim S, et al. (2019) Forecasting daily solar radiation using CEEMDAN decomposition-based mars model trained by crow search algorithm. Energies 12: 1416. https://doi.org/10.3390/en12081416 doi: 10.3390/en12081416
[50]	Rehman S, Mohandes M (2008) Artificial neural network estimation of global solar radiation using air temperature and relative humidity. Energy Policy 36: 571–576. https://doi.org/10.1016/j.enpol.2007.09.033 doi: 10.1016/j.enpol.2007.09.033
[51]	Olatomiwa L, Mekhilef S, Shamshirband S, et al. (2015) Adaptive neuro-fuzzy approach for solar radiation prediction in Nigeria. Renewable Sustainable Energy Rev 51: 1784–1791. https://doi.org/10.1016/j.rser.2015.05.068 doi: 10.1016/j.rser.2015.05.068
[52]	Zeng J, Qiao W (2013) Short-term solar power prediction using a support vector machine. Renewable Energy 52: 118–127. https://doi.org/10.1016/j.renene.2012.10.009 doi: 10.1016/j.renene.2012.10.009
[53]	Wang L, Kisi O, Zounemat-Kermani M, et al. (2017) Prediction of solar radiation in China using different adaptive neuro-fuzzy methods and M5 model tree. Int J Climatol 37: 1141–1155. https://doi.org/10.1002/joc.4762 doi: 10.1002/joc.4762
[54]	Kim S, Seo Y, Rezaie-Balf M, et al. (2018) Evaluation of daily solar radiation flux using soft computing approaches based on different meteorological information: peninsula vs continent. Theor Appl Climatol 137: 693–712. https://doi.org/10.1007/s00704-018-2627-x doi: 10.1007/s00704-018-2627-x
[55]	Sahin G, Eyduran E, Turkoglu M, et al. (2018) Estimation of global irradiation parameters at location of migratory birds in Igdir, Turkey by means of MARS algorithm. Pakistan J Zool 50: 2317–2324. http://dx.doi.org/10.17582/journal.pjz/2018.50.6.2317.2324 doi: 10.17582/journal.pjz/2018.50.6.2317.2324
[56]	Sahin F, Isik G, Sahin G, et al. (2020) Estimation of PM10 levels using feed forward neural networks in Igdir, Turkey. Urban Clim 34: 100721. https://doi.org/10.1016/j.uclim.2020.100721 doi: 10.1016/j.uclim.2020.100721
[57]	Kaya F, Sahin G, Alma MH (2021) Investigation effects of environmental and operating factors on PV panel efficiency using by multivariate linear regression. Int J Energy Res 45: 554–567. https://doi.org/10.1002/er.5717 doi: 10.1002/er.5717
[58]	Friedman JH (2011) Multivariate daptive regression spline. Ann Stat 19: 1–141. https://doi.org/10.1214/aos/1176347963 doi: 10.1214/aos/1176347963
[59]	Sahin F, Kara MK, Koc A, et al. (2020) Multi-criteria decision-making using GIS-AHP for air pollution problem in Igdir Province/Turkey. Environ Sci Poll Rsch 27: 36215–36230. https://doi.org/10.1007/s11356-020-09710-3 doi: 10.1007/s11356-020-09710-3
[60]	Sahraei MA, Duman H, Çodur MY, et al. (2021) Prediction of transportation energy demand: Multivariate adaptive regression splines. Energy 224: 120090. https://doi.org/10.1016/j.energy.2021.120090 doi: 10.1016/j.energy.2021.120090
[61]	Yadav P, Chandel SS (2014) Comparative analysis of diffused solar radiation models for optimum tilt angle determination for Indian locations. Appl Sol Energy 50: 53–59. https://doi.org/10.3103/S0003701X14010137 doi: 10.3103/S0003701X14010137
[62]	Samui P (2013) Multivariate adaptive regression spline (Mars) for prediction of elastic modulus of jointed rock mass. Geotech Geol Eng 31: 249–253. https://doi.org/10.1007/s10706-012-9584-4 doi: 10.1007/s10706-012-9584-4
[63]	Rustemli S, İlcihan Z, Sahin G, et al. (2023) A novel design and simulation of a mechanical coordinate based photovoltaic solar tracking system. AIMS Energy 11: 753–773. https://doi:10.3934/energy.2023037 doi: 10.3934/energy.2023037
[64]	Hastie T, Tibshirani R, Friedman JH, et al. (2004) The elements of statistical learning: Data mining, inference, and prediction. Math Intel 27: 83–85. Available from: https://hastie.su.domains/Papers/ESLII.pdf.
[65]	Craven P, Wahba G (1979) Smoothing noisy data with spline functions: Estimating the correct degree of smoothing by the method of generalized cross-validation. Numer Math 31: 377–403. https://doi.org/10.1007/BF01404567 doi: 10.1007/BF01404567
[66]	Kornacki J, Ćwik J (2005) Statistical learning systems (in Polish). WNT Warsaw, 16.
[67]	Hastie T, Tibshirani R, Friedman J (2011) The elements of statistical learning: Data mining, inference and prediction. Stanford, California: Springer-Verlag, 337–343.
[68]	Put R, Xu Q, Massart D, et al. (2004) Multivariate adaptive regression splines (MARS) in chromatographic quantitative structure-retention relationship studies. J Chromatog 1055: 11–19. https://doi.org/10.1016/j.chroma.2004.07.112 doi: 10.1016/j.chroma.2004.07.112
[69]	Khan MA, Tariq MM, Eyduran E, et al. (2014) Estimating body weight from several body measurements in Harnai sheep without multicollinearity problem. J Anim Pl Sc 24: 120–126.
[70]	Mohammad MT, Rafeeq M, Bajwa MA, et al. (2012) Prediction of body weight from body measurements using regression tree (RT) method for indigenous sheep breeds in Balochistan. Pakistan J Anim Pl Sci 22: 20–24.
[71]	Roy SS, Roy R, Balas VE (2018) Estimating heating load in buildings using multivariate adaptive regression splines, extreme learning machine, a hybrid model of MARS and ELM. Renewable Sustainable Energy Rev 82: 4256e68. https://doi.org/10.1016/j.rser.2017.05.249 doi: 10.1016/j.rser.2017.05.249
[72]	Huang H, Ji X, Xia F, et al. (2020) Multivariate adaptive regression splines for estimating riverine constituent concentrations. Process 34: 1213e27. https://doi.org/10.1002/hyp.13669 doi: 10.1002/hyp.13669
[73]	Gross F, Persaud B, Lyon C, et al. (2010) A guide to developing quality crash modification factors. Publication, FHWA-SA-10-032, FHWA. U.S. Department of Transportation. Available from: https://www.cmfclearinghouse.org/collateral/CMF_Guide.pdf.
[74]	Tunay KB (2001) Estimated income velocity of money method MARS in Turkey. METU Develop J 28: 431–454.
[75]	Li Y, He Y, Su Y, et al. (2016), Forecasting the daily power output of a grid-connected photovoltaic system based on multivariate adaptive regression splines. Appl Energy 180: 392e401. https://doi.org/10.1016/j.apenergy.2016.07.052 doi: 10.1016/j.apenergy.2016.07.052

This article has been cited by:

1.	Sana Abdulkream Alharbi, Azmin Sham Rambely, A New ODE-Based Model for Tumor Cells and Immune System Competition, 2020, 8, 2227-7390, 1285, 10.3390/math8081285
2.	Bruno Carbonaro, Marco Menale, Towards the Dependence on Parameters for the Solution of the Thermostatted Kinetic Framework, 2021, 10, 2075-1680, 59, 10.3390/axioms10020059
3.	Carlo Bianca, On the Modeling of Energy-Multisource Networks by the Thermostatted Kinetic Theory Approach: A Review with Research Perspectives, 2022, 15, 1996-1073, 7825, 10.3390/en15217825
4.	Carlo Bianca, Thermostatted Kinetic Theory Structures in Biophysics: Generalizations and Perspectives, 2024, 4, 2673-9909, 1278, 10.3390/appliedmath4040069
5.	Carlo Bianca, A decade of thermostatted kinetic theory models for complex active matter living systems, 2024, 50, 15710645, 72, 10.1016/j.plrev.2024.06.015
6.	Carlo Bianca, The Role of the Table of Games in the Discrete Thermostatted Kinetic Theory, 2024, 12, 2227-7390, 2356, 10.3390/math12152356
7.	Bruno Carbonaro, Marco Menale, A nonconservative kinetic framework under the action of an external force field: Theoretical results with application inspired to ecology, 2023, 34, 0956-7925, 1170, 10.1017/S0956792523000232
8.	Fengling Jia, Peiyan He, Lixin Yang, A Novel Coupled Memristive Izhikevich Neuron Model and Its Complex Dynamics, 2024, 12, 2227-7390, 2244, 10.3390/math12142244
9.	Bruno Carbonaro, Marco Menale, Markov Chains and Kinetic Theory: A Possible Application to Socio-Economic Problems, 2024, 12, 2227-7390, 1571, 10.3390/math12101571
10.	Marco Menale, Ezio Venturino, A kinetic theory approach to modeling prey–predator ecosystems with expertise levels: analysis, simulations and stability considerations, 2024, 43, 2238-3603, 10.1007/s40314-024-02726-2
11.	Muhammad Naeem Aslam, Nadeem Shaukat, Muhammad Sarmad Arshad, Muhammad Waheed Aslam, Javed Hussain, An intelligent approach for analyzing the effects of normal tumor immune unhealthy diet model through unsupervised physics informed neural-networks integrated with meta-heuristic algorithms, 2025, 18, 1864-5909, 10.1007/s12065-024-01007-5

Reader Comments

Your name:*

Email:*
© 2023 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)