Research article Special Issues

Enhanced fault detection models with real-life applications

  • Nonconforming events are rare in high-quality processes, and the time between events (TBE) may follow a skewed distribution, such as the gamma distribution. This study proposes one- and two-sided triple homogeneously weighted moving average charts for monitoring TBE data modeled by the gamma distribution. These charts are labeled as the THWMA TBE charts. Monte Carlo simulations are performed to approximate the run length distribution of the one- and two-sided THWMA TBE charts. The THWMA TBE charts are compared to competing charts like the DHWMA TBE, HWMA TBE, THWMA TBE, DEWMA TBE, and EWMA TBE charts at a single shift and over a range of shifts. For the single shift comparison, the average run length (ARL) and standard deviation run length (SDRL) measures are used, whereas the extra quadratic loss (EQL), relative average run length (RARL) and performance comparison index (PCI) measures are employed for a range of shifts comparison. The comparison reveals that the THWMA TBE charts outperform the competing charts at a single shift as well as at a certain range of shifts. Finally, two real-life data applications are presented to evaluate the applicability of the THWMA TBE charts in practical situations, one with boring machine failure data and the other with hospital stay time for traumatic brain injury patients.

    Citation: Showkat Ahmad Lone, Zahid Rasheed, Sadia Anwar, Majid Khan, Syed Masroor Anwar, Sana Shahab. Enhanced fault detection models with real-life applications[J]. AIMS Mathematics, 2023, 8(8): 19595-19636. doi: 10.3934/math.20231000

    Related Papers:

    [1] Xueyong Wang, Gang Wang, Ping Yang . Novel Pareto Z-eigenvalue inclusion intervals for tensor eigenvalue complementarity problems and its applications. AIMS Mathematics, 2024, 9(11): 30214-30229. doi: 10.3934/math.20241459
    [2] Panpan Liu, Hongbin Lv . An algorithm for calculating spectral radius of s-index weakly positive tensors. AIMS Mathematics, 2024, 9(1): 205-217. doi: 10.3934/math.2024012
    [3] Yajun Xie, Changfeng Ma . The hybird methods of projection-splitting for solving tensor split feasibility problem. AIMS Mathematics, 2023, 8(9): 20597-20611. doi: 10.3934/math.20231050
    [4] Haroun Doud Soliman Adam, Khalid Ibrahim Adam, Sayed Saber, Ghulam Farid . Existence theorems for the dbar equation and Sobolev estimates on q-convex domains. AIMS Mathematics, 2023, 8(12): 31141-31157. doi: 10.3934/math.20231594
    [5] Tinglan Yao . An optimal Z-eigenvalue inclusion interval for a sixth-order tensor and its an application. AIMS Mathematics, 2022, 7(1): 967-985. doi: 10.3934/math.2022058
    [6] Rashad Ismail, Saira Hameed, Uzma Ahmad, Khadija Majeed, Muhammad Javaid . Unbalanced signed graphs with eigenvalue properties. AIMS Mathematics, 2023, 8(10): 24751-24763. doi: 10.3934/math.20231262
    [7] Qin Zhong . Some new inequalities for nonnegative matrices involving Schur product. AIMS Mathematics, 2023, 8(12): 29667-29680. doi: 10.3934/math.20231518
    [8] Yuanqiang Chen, Jihui Zheng, Jing An . A Legendre spectral method based on a hybrid format and its error estimation for fourth-order eigenvalue problems. AIMS Mathematics, 2024, 9(3): 7570-7588. doi: 10.3934/math.2024367
    [9] Ridho Alfarisi, Liliek Susilowati, Dafik . Local multiset dimension of comb product of tree graphs. AIMS Mathematics, 2023, 8(4): 8349-8364. doi: 10.3934/math.2023421
    [10] Paul Bracken . Applications of the lichnerowicz Laplacian to stress energy tensors. AIMS Mathematics, 2017, 2(3): 545-556. doi: 10.3934/Math.2017.2.545
  • Nonconforming events are rare in high-quality processes, and the time between events (TBE) may follow a skewed distribution, such as the gamma distribution. This study proposes one- and two-sided triple homogeneously weighted moving average charts for monitoring TBE data modeled by the gamma distribution. These charts are labeled as the THWMA TBE charts. Monte Carlo simulations are performed to approximate the run length distribution of the one- and two-sided THWMA TBE charts. The THWMA TBE charts are compared to competing charts like the DHWMA TBE, HWMA TBE, THWMA TBE, DEWMA TBE, and EWMA TBE charts at a single shift and over a range of shifts. For the single shift comparison, the average run length (ARL) and standard deviation run length (SDRL) measures are used, whereas the extra quadratic loss (EQL), relative average run length (RARL) and performance comparison index (PCI) measures are employed for a range of shifts comparison. The comparison reveals that the THWMA TBE charts outperform the competing charts at a single shift as well as at a certain range of shifts. Finally, two real-life data applications are presented to evaluate the applicability of the THWMA TBE charts in practical situations, one with boring machine failure data and the other with hospital stay time for traumatic brain injury patients.



    Let m and n be two positive integers with m2 and n2, [n]={1,2,,n}, R be the set of all real numbers, Rn be the set of all n-dimensional real vectors. Let x=(x1,x2,,xm)Rm and y=(y1,y2,,yn)Rn. If a fourth-order tensor A=(aijkl)R[m]×[n]×[m]×[n] satisfies the properties

    aijkl=akjil=ailkj=aklij,i,k[m],j,l[n],

    then we call A a partially symmetric tensor.

    It is well know that the tensor of the elastic modulus of elastic materials is just partially symmetrical [11]. And the components of a fourth-order partially symmetric tensor A can be regarded as the coefficients of the following biquadratic homogeneous polynomial optimization problem [6,19]:

    maxf(x,y)Axyxyi,k[m]j,l[n]aijklxiyjxkyl,s.t.xx=1,yy=1. (1.1)

    The optimization problem plays a great role in the analysis of nonlinear elastic materials and the entanglement problem in quantum physics [5,6,8,9,26]. To solve the problem, we would establish a new version based on the following definition:

    Definition 1.1. [11,20,21] Let A=(aijkl)R[m]×[n]×[m]×[n] be a partially symmetric tensor. If there are λR, xRm{0} and yRn{0} such that

    Ayxy=λx,Axyx=λy,xx=1,yy=1, (1.2)

    where

    (Ayxy)i=k[m]j,l[n]aijklyjxkyl,(Axyx)l=i,k[m]j[n]aijklxiyjxk,

    then we call λ an M-eigenvalue of A, x and y the left and right M-eigenvectors associated with λ, respectively. Let σ(A) be the set of all M-eigenvalues of A and λmax(A) be the largest M-eigenvalue of A, i.e.,

    λmax(A)=max{|λ|:λσ(A)}.

    In 2009, Wang, Qi and Zhang [24] pointed out that Problem (1.1) is equivalently transformed into calculating the largest M-eigenvalue of a fourth-order partially symmetric tensor. Based on this, Wang et al. [24] presented an algorithm (WQZ-algorithm) to find the largest M-eigenvalue of a fourth-order partially symmetric tensor.

    WQZ-algorithm [24,Algorithm 4.1]:

    Initial step: Input A=(aijkl)R[m]×[n]×[m]×[n] and unfold it into a matrix A=(Ast)R[mn]×[mn] by mapping Ast=aijkl with s=n(i1)+j,t=n(k1)+l.

    Substep 1: Take

    τ=1stmn|Ast|, (1.3)

    and set

    ¯A=τI+A, (1.4)

    where I=(δijkl)R[m]×[n]×[m]×[n] with δijkl=1 if i=k and j=l, otherwise, δijkl=0. Then unfold ¯A=(¯aijkl)R[m]×[n]×[m]×[n] into a matrix ¯A=(¯Ast)R[mn]×[mn].

    Substep 2: Compute the unit eigenvector w=(wi)mni=1Rmn of matrix ¯A associated with its largest eigenvalue, and fold vector w into the matrix W=(Wij)R[m]×[n] in the following way:

    Wij=wk,

    set i=k/n,j=(k1)modn+1,k=1,2,,mn.

    Substep 3: Compute the singular vectors u1 and v1 corresponding to the largest singular value σ1 of the matrix W. Specifically, the singular value decomposition of W is

    W=UTΣV=ri=1σiuivTi,

    where σ1σ2σr and r is the rank of W.

    Substep 4: Take x0=u1,y0=v1, and let k=0.

    Iterative step: Execute the following procedures alternatively until certain convergence criterion is satisfied and output x,y:

    ¯xk+1=¯Aykxkyk,xk+1=¯xk+1||¯xk+1||,¯yk+1=¯Axk+1ykxk+1,yk+1=¯yk+1||¯yk+1||,k=k+1.

    Final step: Output the largest M-eigenvalue of the tensor A:

    λmax(A)=f(x,y)τ,

    where

    f(x,y)=i,k[m]j,l[n]¯aijklxiyjxkyl,

    and the associated M-eigenvectors: x,y.

    The M-eigenvalues of tensors have a close relationship with the strong ellipticity condition in elasticity theory, which guarantees the existence of the solution to the fundamental boundary value problems of elastostatics [3,5,16]. However, when the dimensions m and n of tensors are large, it is not easy to calculate all M-eigenvalues. Thus, the problem of M-eigenvalue localization have attracted the attention of many researchers and many M-eigenvalue localization sets are given; see [2,4,13,14,15,17,18,23,27].

    For this, Wang, Li and Che [23] presented the following M-eigenvalue localization set for a partially symmetric tensor:

    Theorem 1.1. [23,Theorem 2.2] Let A=(aijkl)R[m]×[n]×[m]×[n] be a partially symmetric tensor. Then

    σ(A)H(A)=i[m]k[m],kiHi,k(A),

    where

    Hi,k(A)=[ˆHi,k(A)(¯Hi,k(A)Γi(A))],ˆHi,k(A)={zC:|z|Ri(A)Rki(A),|z|Rkk(A)},¯Hi,k(A)={zC:(|z|(Ri(A)Rki(A)))(|z|Rkk(A))Rki(A)(Rk(A)Rkk(A))},Ri(A)=k[m]j,l[n]|aijkl|,Rki(A)=j,l[n]|aijkl|.

    From the set H(A) in Theorem 1.1, we can obtain an upper bound of the largest M-eigenvalue λmax(A), which can be taken as an parameter τ in WQZ-algorithm. From Example 2 in [15], it can be seen that the smaller the upper bound of λmax(A), the faster WQZ-algorithm converges. In view of this, this paper intends to provide a smaller upper bound based on a new inclusion set and take this new upper bound as a parameter τ to make WQZ-algorithm converges to λmax(A) faster.

    The remainder of this paper is organized as follows. In Section 2, we provide an M-eigenvalue localization set for a partially symmetric tensor A and prove that the new set is tighter than some existing M-eigenvalue localization sets. In Section 3, based on the new set, we provide an upper bound for the largest M-eigenvalue of A. As an application, in order to make the sequence generated by WQZ-algorithm converge to the largest M-eigenvalue of A faster, we replace the parameter τ in WQZ-algorithm with the upper bound. In Section 4, we conclude this article.

    In this section, we provide a new M-eigenvalue localization set of a fourth-order partially symmetric tensor and prove that the new M-eigenvalue localization set is tighter than that in Theorem 1.1, i.e., Theorem 2.2 in [23]. Before that, the following conclusion in [1,25] is needed.

    Lemma 2.1. Let x=(x1,x2,,xn)Rn and y=(y1,y2,,yn)Rn. Then

    a) If x2=1, then |xi||xj|12 for i,j[n], ij;

    b) (i[n]xiyi)2i[n]x2ii[n]y2i.

    Theorem 2.1. Let A=(aijkl)R[m]×[n]×[m]×[n] be a partially symmetric tensor. Then

    σ(A)Υ(A)=i[m]s[m],siΥi,s(A),

    where

    Υi,s(A)=[ˆΥi,s(A)(˜Υi,s(A)¯Υi,s(A))],ˆΥi,s(A)={zR:|z|<˜rsi(A),|z|<rss(A)},˜Υi,s(A)={zR:(|z|˜rsi(A))(|z|rss(A))rsi(A)˜rss(A)},¯Υi,s(A)={zR:|z|<˜rsi(A)+rsi(A)},

    and

    ˜rst(A)=12k[m],ksj,l[n],jl|atjkl|+k[m],ksl[n]a2tlkl,rst(A)=12j,l[n],jl|atjsl|+l[n]a2tlsl,t[m].

    Proof. Let λ be an M-eigenvalue of A, xRm{0} and yRn{0} be its left and right M-eigenvectors, respectively. Then xx=1. Let |xt|=maxi[m]|xi|. Then 0<|xt|1. For any given s[m] and st, by the t-th equation of (1.2), we have

    λxt=k[m]j,l[n]atjklyjxkyl=k[m],ksj,l[n],jlatjklyjxkyl+k[m],ksl[n]atlklylxkyl+j,l[n],jlatjslyjxsyl+l[n]atlslylxsyl.

    Taking the modulus of the above equation and using the triangle inequality and Lemma 2.1, one has

    |λ||xt|k[m],ksj,l[n],jl|atjkl||yj||xk||yl|+k[m],ksl[n]|atlkl||yl||xk||yl|+j,l[n],jl|atjsl||yj||xs||yl|+l[n]|atlsl||yl||xs||yl|12k[m],ksj,l[n],jl|atjkl||xt|+k[m],ksl[n]|atlkl||yl||xt|+12j,l[n],jl|atjsl||xs|+l[n]|atlsl||yl||xs|=12k[m],ksj,l[n],jl|atjkl||xt|+|xt|k[m],ks(l[n]|atlkl||yl|)+12j,l[n],jl|atjsl||xs|+|xs|l[n]|atlsl||yl|12k[m],ksj,l[n],jl|atjkl||xt|+|xt|k[m],ks(l[n]|atlkl|2l[n]|yl|2)+12j,l[n],jl|atjsl||xs|+|xs|l[n]|atlsl|2l[n]|yl|2=12k[m],ksj,l[n],jl|atjkl||xt|+|xt|k[m],ksl[n]a2tlkl+12j,l[n],jl|atjsl||xs|+|xs|l[n]a2tlsl=(12k[m],ksj,l[n],jl|atjkl|+k[m],ksl[n]a2tlkl)|xt|+(12j,l[n],jl|atjsl|+l[n]a2tlsl)|xs|=˜rst(A)|xt|+rst(A)|xs|,

    i.e.,

    (|λ|˜rst(A))|xt|rst(A)|xs|. (2.1)

    By (2.1), we have (|λ|˜rst(A))|xt|rst(A)|xt|, which leads to that |λ|˜rst(A)+rst(A), i.e., λ¯Υt,s(A).

    If |xs|>0, then by the s-th equation of (1.2), we have

    λxs=k[m]j,l[n]asjklyjxkyl=k[m],ksj,l[n],jlasjklyjxkyl+k[m],ksl[n]aslklylxkyl+j,l[n],jlasjslyjxsyl+l[n]aslslylxsyl.

    Taking the modulus of the above equation and using the triangle inequality and Lemma 2.1 yield

    |λ||xs|k[m],ksj,l[n],jl|asjkl||yj||xk||yl|+k[m],ksl[n]|aslkl||yl||xk||yl|+j,l[n],jl|asjsl||yj||xs||yl|+l[n]|aslsl||yl||xs||yl|12k[m],ksj,l[n],jl|asjkl||xt|+k[m],ksl[n]|aslkl||yl||xt|+12j,l[n],jl|asjsl||xs|+l[n]|aslsl||yl||xs|=12k[m],ksj,l[n],jl|asjkl||xt|+|xt|k[m],ks(l[n]|aslkl||yl|)+12j,l[n],jl|asjsl||xs|+|xs|l[n]|aslsl||yl|12k[m],ksj,l[n],jl|asjkl||xt|+|xt|k[m],ks(l[n]|aslkl|2l[n]|yl|2)+12j,l[n],jl|asjsl||xs|+|xs|l[n]|aslsl|2l[n]|yl|2=12k[m],ksj,l[n],jl|asjkl||xt|+|xt|k[m],ksl[n]a2slkl+12j,l[n],jl|asjsl||xs|+|xs|l[n]a2slsl=(12k[m],ksj,l[n],jl|asjkl|+k[m],ksl[n]a2slkl)|xt|+(12j,l[n],jl|asjsl|+l[n]a2slsl)|xs|=˜rss(A)|xt|+rss(A)|xs|,

    i.e.,

    (|λ|rss(A))|xs|˜rss(A)|xt|. (2.2)

    When |λ|˜rst(A) or |λ|rss(A), multiplying (2.1) and (2.2) and eliminating |xt||xs|>0, we have

    (|λ|˜rst(A))(|λ|rss(A))rst(A)˜rss(A), (2.3)

    which implies that

    λ(˜Υt,s(A)¯Υt,s(A)). (2.4)

    When |λ|<˜rst(A) and |λ|<rss(A), it holds that

    λˆΥt,s(A). (2.5)

    It follows from (2.4) and (2.5) that

    λ[ˆΥt,s(A)(˜Υt,s(A)¯Υt,s(A))]=Υt,s(A). (2.6)

    If |xs|=0 in (2.1), then |λ|˜rst(A). When |λ|=˜rst(A), then (2.3) holds and consequently, (2.4) holds. When |λ|<˜rst(A), if |λ|rss(A), then (2.3) and (2.4) hold. If |λ|<rss(A), then (2.5) holds. Hence, (2.6) holds. By the arbitrariness of s[m], and st, we have

    λtsΥt,s(A)t[m]tsΥt,s(A),

    therefore, the assertion is proved.

    Next, we give the relationship between the localization set H(A) given in Theorem 1.1 and the set Υ(A) given in Theorem 2.1.

    Theorem 2.2. Let A=(aijkl)R[m]×[n]×[m]×[n] be a partially symmetric tensor. Then

    Υ(A)H(A).

    Proof. For any i,s[m] and is, it holds that

    ˜rsi(A)=12k[m],ksj,l[n],jl|aijkl|+k[m],ksl[n]a2ilklk[m],ksj,l[n]|aijkl|=Ri(A)Rsi(A); (2.7)

    and

    rsi(A)=12j,l[n],jl|aijsl|+l[n]a2ilslj,l[n]|aijsl|=Rsi(A). (2.8)

    Let zΥ(A). By Theorem 2.1, there is an index i[m] such that for any s[m], is, zΥi,s(A), which means that zˆΥi,s(A), or z˜Υi,s(A) and z¯Υi,s(A).

    Let zˆΥi,s(A), i.e., |z|<˜rsi(A) and |z|<rss(A). By (2.7) and (2.8), we have |z|Ri(A)Rsi(A) and |z|Rss(A), therefore, zˆHi,s(A).

    Let z˜Υi,s(A) and z¯Υi,s(A), i.e.,

    (|z|˜rsi(A))(|z|rss(A))rsi(A)˜rss(A), (2.9)

    and

    |z|<˜rsi(A)+rsi(A). (2.10)

    By (2.7), (2.8) and (2.10), one has |z|<˜rsi(A)+rsi(A)Ri(A), which means that zΓi(A). When |z|Ri(A)Rsi(A) and |z|Rss(A), by (2.7), (2.8) and (2.9), we have

    |z|˜rsi(A)|z|(Ri(A)Rsi(A))0,|z|rss(A)|z|Rss(A)0,

    then

    (|z|(Ri(A)Rsi(A)))(|z|Rss(A))(|z|˜rsi(A))(|z|rss(A))rsi(A)˜rss(A)Rsi(A)(Rs(A)Rss(A)),

    i.e.,

    (|z|(Ri(A)Rsi(A)))(|z|Rss(A))Rsi(A)(Rs(A)Rss(A)), (2.11)

    which means that z¯Hi,s(A). Thus, whether Ri(A)Rsi(A)|z|Rss(A) or Rss(A)|z|Ri(A)Rsi(A), (2.11) also holds. When |z|Ri(A)Rsi(A) and |z|Rss(A), it follows that zˆHi,s(A). i.e.,

    z[ˆHi,s(A)(¯Hi,s(A)Γi(A))]=Hi,s(A).

    From the arbitrariness of s[m], and si, we have

    zs[m],siHi,s(A)i[m]s[m],siHi,s(A),

    i.e., zH(A). Therefore, Υ(A)H(A).

    In order to show the validity of the set Υ(A) given in Theorem 2.1, we present a running example.

    Example 1. Let A=(aijkl)R[2]×[2]×[2]×[2] be a partially symmetric tensor with entries

    a1111=1,a1112=2,a1121=2,a1212=3,a1222=5,a1211=2,a1122=4,a1221=4,a2111=2,a2112=4,a2121=3,a2122=5,a2211=4,a2212=5,a2221=5,a2222=6.

    By Theorem 1.1, we have

    H(A)=i[m]k[m],kiHi,k(A)={zC:|z|29.4765}.

    By Theorem 2.1, we have

    Υ(A)=i[m]s[m],siΥi,s(A)={zC:|z|20.0035}.

    It is easy to see that Υ(A)H(A) and all M-eigenvalues are in [20.0035,20.0035]. In fact, all different M-eigenvalues of A are 1.2765, 0.0710, 0.1242, 0.2765, 0.3437 and 15.2091.

    In this section, based on the set in Theorem 2.1, we provide an upper bound for the largest M-eigenvalue of a fourth-order partially symmetric tensor A. As an application, we apply the upper bound as a parameter τ to the WQZ-algorithm to make the sequence generated by the WQZ-algorithm converges to the largest M-eigenvalue of A faster.

    Theorem 3.1. Let A=(aijkl)R[m]×[n]×[m]×[n] be a partially symmetric tensor. Then

    ρ(A)Ω(A)=maxi[m]mins[m],isΩi,s(A),

    where

    Ωi,s(A)=max{min{˜rsi(A),rss(A)},min{˜rsi(A)+rsi(A),ˆΩi,s(A)}},

    and

    ˆΩi,s(A)=12{˜rsi(A)+rss(A)+(rss(A)˜rsi(A))2+4rsi(A)˜rss(A)}.

    Proof. By Theorem 2.1 and ρ(A)σ(A), it follows that there exists an index i[m] such that for any s[m] and si, ρ(A)ˆΥi,s(A), or ρ(A)(˜Υi,s(A)¯Υi,s(A)). If ρ(A)ˆΥi,s(A), that is, ρ(A)<˜rsi(A) and ρ(A)<rss(A), then

    ρ(A)<min{˜rsi(A),rss(A)}. (3.1)

    If ρ(A)(˜Υi,s(A)¯Υi,s(A)), that is,

    ρ(A)<˜rsi(A)+rsi(A)<min{˜rsi(A)+rsi(A)}, (3.2)

    and

    (ρ(A)˜rsi(A))(ρ(A)rss(A))rsi(A)˜rss(A). (3.3)

    Solving Inequality (3.3), we have

    ρ(A)ˆΩi,s(A)min{ˆΩi,s(A)}. (3.4)

    Combining (3.2) and (3.4), we have

    ρ(A)min{˜rsi(A)+rsi(A),ˆΩi,s(A)}. (3.5)

    Hence, by (3.1) and (3.5), we have

    ρ(A)max{min{˜rsi(A),rss(A)},min{˜rsi(A)+rsi(A),ˆΩi,s(A)}}=Ωi,s(A).

    Furthermore, by the arbitrariness of s, we have

    ρ(A)mins[m],isΩi,s(A).

    Since we do not know which i is appropriate to ρ(A), we can only conclude that

    ρ(A)maxi[m]mins[m],isΩi,s(A).

    This proof is complete.

    Remark 3.1. In Theorem 3.1, we obtain an upper bound Ω(A) for the largest M-eigenvalue of a fourth order partially symmetric tensor A. Now, we take Ω(A) as the parameter τ in WQZ-algorithm to obtain a modified WQZ-algorithm. That is, the only difference between WQZ-algorithm and the modified WQZ-algorithm is the selection of τ, in particular, τ=1stmn|Ast| in WQZ-algorithm and τ=Ω(A) in the modified WQZ-algorithm.

    Next, we take Ω(A) and some existing upper bounds of the largest M-eigenvalue as τ in WQZ-algorithm to calculate the largest M-eigenvalue of a fourth-order partially symmetric tensor A.

    Example 2. Consider the tensor A in Example 4.1 of [24], where

    A(:,:,1,1)=[0.97270.31690.34370.63320.78660.42570.33500.98960.4323],
    A(:,:,2,1)=[0.63320.78660.42570.73870.68730.32480.79860.59880.9485],
    A(:,:,3,1)=[0.33500.98960.43230.79860.59880.94850.58530.59210.6301],
    A(:,:,1,2)=[0.31690.61580.01840.78660.01600.00850.98960.66630.2559],
    A(:,:,2,2)=[0.78660.01600.00850.68730.51600.02160.59880.04110.9857],
    A(:,:,3,2)=[0.98960.66630.25590.59880.04110.98570.59210.29070.3881],
    A(:,:,1,3)=[0.34370.01840.56490.42570.00850.14390.43230.25590.6162],
    A(:,:,2,3)=[0.42570.00850.14390.32480.02160.00370.94850.98570.7734],
    A(:,:,3,3)=[0.43230.25590.61620.94850.98570.77340.63010.38810.8526].

    By (1.3), we have τ=1st9|Ast|=23.3503. By Corollary 1 of [17], we have

    ρ(A)16.6014.

    By Theorem 3.5 of [23], we have

    ρ(A)15.4102.

    By Corollary 2 of [17], we have

    ρ(A)14.5910.

    By Corollary 1 of [15], where Sm=Sn=1, we have

    ρ(A)13.8844.

    By Corollary 2 of [15], where Sm=Sn=1, we have

    ρ(A)11.7253.

    By Theorem 3.1, we have

    ρ(A)8.2342.

    From [24], it can be seen that λmax(A)=2.3227.

    Taking τ=23.3503, 16.6014, 15.4102, 14.5910, 13.8844, 11.7253 and 8.2342 respectively, numerical results obtained by the WQZ-algorithm are shown in Figure 1.

    Figure 1.  Numerical results for the WQZ-algorithm with different τ.

    Numerical results in Figure 1 shows that :

    1) When we take τ=8.2342, the sequence more rapidly converges to the largest M-eigenvalue λmax(A) than taking τ=23.3503, τ=16.6014, τ=15.4102, τ=14.5910, τ=13.8844 and τ=11.7253, respectively.

    2) When we take τ=23.3503, 16.6014, 15.4102, 14.5910, 13.8844, 11.7253 and 8.2342, the WQZ-algorithm can get the largest M-eigenvalue λmax(A) after finite iterations. However, under the same stopping criterion, if we take τ=23.3503, 16.6014, 15.4102, 14.5910, 13.8844 and 11.7253, it can be seen that the WQZ-algorithm needs more iterations to obtain the largest M-eigenvalue, and when τ=8.2342, WQZ-algorithm can obtain the largest M-eigenvalue λmax(A) faster.

    3) The choice of the parameter τ in WQZ-algorithm has a significant impact on the convergence speed of the WQZ-algorithm. When τ is larger, the convergence speed of WQZ-algorithm is slower. When τ is smaller and τ is greater than the largest M-eigenvalue, the WQZ-algorithm converges faster. In other words, the faster the largest M-eigenvalue can be obtained.

    4) The numerical result of the upper bound of the M-spectral radius obtained by Theorem 3.1 is of great help to the WQZ-algorithm. Therefore, it shows that the results we get have a certain effect.

    Now, we consider a real elasticity tensor, which is derived from the study of self-anisotropic materials [10] for explanation.

    In anisotropy materials, the components of the tensor of elastic moduli C=(cijkl)R[3]×[3]×[3]×[3] satisfy the following symmetry:

    cijkl=cjikl=cijlk=cjilk,cijkl=cklij,1i,j,k,l3,

    which is also called an elasticity tensor. After a lot of research, we know that there are many anisotropic materials, of which crystal is one of its typical examples. We classify from the crystal homologues [22], the elasticity tensor C=(cijkl)R[3]×[3]×[3]×[3] of some crystals for trigonal system, such as CaCO3 and HgS also satisfy

    c1112=c2212=c3323=c3331=c3312=c2331=0,c2222=c1111,c3131=c2323,c2233=c1133,c2223=c1123,c2231=c1131,c3112=2c1123,c2312=2c1131,c1212=c1111c1122.

    This shows that the triangular system of anisotropic materials has only 7 elasticities. In fact, CaMg(CO3)2-dolomite and CaCO3-calcite have similar crystal structures, in which the atoms along any triplet are alternated with magnesium and calcium. In [22], we can know that the elasticity tensor of CaMg(CO3)2-dolomite is as follows.

    c2222=c1111=196.6,c3131=c2323=83.2,c2233=c1133=54.7,c2223=c1123=31.7,c2231=c1131=25.3,c3112=44.8,c2312=35.84,c1212=132.2,c3333=110,c1122=64.4.

    Next, we transform the elastic tensor C into a partially symmetric tensor A through the following double mapping, and the M-eigenvalue of A after transformation is the same as the M-eigenvalue of C [7,12]:

    aijkl=aikjl,1i,j,k,l3.

    In order to illustrate the validity of the results we obtained, we take the above-mentioned partial symmetry tensor of the CaMg(CO3)2-dolomite elasticity tensor transformation as an example.

    Example 3. Consider the tensor A2=(aijkl)R[3]×[3]×[3]×[3] in Example 3 of [17], where

    a2222=a1111=196.6,a3311=a2233=83.2,a2323=a3232=a1313=a3131=54.7,a2223=a2232=a1213=a2131=31.7,a3333=110,a1212=a2121=64.4,a1122=132.2,a2321=a1232=a1311=a1131=25.3,a3112=a1321=44.8,a2132=a1223=35.84,

    and other aijkl=0.

    The data results of Example 2 show that the upper bound of the largest M-eigenvalue in Theorem 3.1 is sharper than the existing results.Here, we only calculate the upper bound of the largest M-eigenvalue of A2 by Theorem 3.1, and use it as the parameter τ in the WQZ-algorithm to calculate the largest M-eigenvalue of A2. Here, in order to distinguish different values of τ, we calculate the result by Theorem 3.1 and record it as τ2, that is, WQZ-algorithm τ=τ2.

    By Theorem 3.1, we can get τ2=647.6100.

    By Eq (1.3), we can get

    τ=1st9|Ast|=1998.6000.

    In the WQZ-algorithm, when we take τ=1998.6000 and 647.6100 respectively, the numerical results we get are shown in Figure 2.

    Figure 2.  Numerical results for the WQZ-algorithm with different τ.

    As we can see in Figure 2, in the WQZ-algorithm, when we regard τ2 as τ, it makes the convergence sequence in the WQZ-algorithm converges faster than τ=1st9|Ast|, so that the largest M-eigenvalue can be calculated faster.That is to say, in this article, the result we provide as the parameter τ in the WQZ-algorithm can speed up the convergence speed, so that the largest M-eigenvalue can be calculated quickly.

    In this paper, we first in Theorem 2.1 provided an M-eigenvalue localization set Υ(A) for a fourth-order partially symmetric tensor A, and then proven that the set Υ(A) is tighter than the set H(A) in Theorem 2.2 of [23]. Secondly, based on the set Υ(A), we derived an upper bound for the M-spectral radius of A. As an application, we took the upper bound of the M-spectral radius as a parameter τ in the WQZ-algorithm to make the sequence generated by this algorithm converge to the largest M-eigenvalue of A faster. Finally, two numerical examples are given to show the effectiveness of the set Υ(A) and the upper bound Ω(A).

    The author sincerely thanks the editors and anonymous reviewers for their insightful comments and constructive suggestions, which greatly improved the quality of the paper. The author also thanks Professor Jianxing Zhao (Guizhou Minzu University) for guidance. This work is supported by Science and Technology Plan Project of Guizhou Province (Grant No. QKHJC-ZK[2021]YB013).

    The author declares no conflict of interest.



    [1] W. A. Shewhart, Economic control of quality of manufactured product, New York: D. Van Nostrand Company, 1931.
    [2] V. Alevizakos, C. Koukouvinos, A. Lappa, Monitoring of time between events with a double generally weighted moving average control chart, Qual. Reliab. Eng. Int., 35 (2019), 685–710. https://doi.org/10.1002/qre.2430 doi: 10.1002/qre.2430
    [3] M. Aslam, M. Khan, S. M. Anwar, B. Zaman, A homogeneously weighted moving average control chart for monitoring time between events, Qual. Reliab. Eng. Int., 38 (2022), 1013–1044. https://doi.org/10.1002/qre.3032 doi: 10.1002/qre.3032
    [4] J. M. Lucas. Counted data CUSUM's, Technometrics, 27 (1985), 129–144.
    [5] S. Vardeman, D. Ray, Average run lengths for CUSUM schemes when observations are exponentially distributed, Technometrics, 27 (1985), 145–150.
    [6] F. F. Gan, Designs of one- and two-sided exponential EWMA charts, J. Qual. Technol., 30 (1998), 55–69. https://doi.org/10.1080/00224065.1998.11979819 doi: 10.1080/00224065.1998.11979819
    [7] L. Y. Chan, M. Xie, T. N. Goh, Cumulative quantity control charts for monitoring production processes, Int. J. Prod. Res., 38 (2000), 397–408. https://doi.org/10.1080/002075400189482 doi: 10.1080/002075400189482
    [8] M Xie, T. N. Goh, P. Ranjan, Some effective control chart procedures for reliability monitoring, Reliab. Eng. Syst. Safe., 77 (2002), 143–150. https://doi.org/10.1016/S0951-8320(02)00041-8 doi: 10.1016/S0951-8320(02)00041-8
    [9] C. Pehlivan, M. C. Testik, Impact of model misspecification on the exponential EWMA charts: a robustness study when the time-between-events are not exponential, Qual. Reliab. Eng. Int., 26 (2010), 177–190. https://doi.org/10.1002/qre.1033 doi: 10.1002/qre.1033
    [10] N. Chakraborty, S. W. Human, N. Balakrishnan, A generally weighted moving average chart for time between events, Commun. Stat.-Simul. Comput., 46 (2017), 7790–7817. https://doi.org/10.1080/03610918.2016.1252397 doi: 10.1080/03610918.2016.1252397
    [11] V. Alevizakos, C. Koukouvinos, A double exponentially weighted moving average chart for time between events., Commun. Stat.-Simul. Comput., 49 (2000), 2765–2784. https://doi.org/10.1080/03610918.2018.1532516 doi: 10.1080/03610918.2018.1532516
    [12] V. Alevizakos, K. Chatterjee, C. Koukouvinos, A triple exponentially weighted moving average control chart for monitoring time between events, Qual. Reliab. Eng. Int., 37 (2021), 1059–1079. https://doi.org/10.1002/qre.2781 doi: 10.1002/qre.2781
    [13] S. W. Roberts, Control chart tests based on geometric moving averages, Technometrics, 42 (2000), 97–101.
    [14] J. S. Hunter, The exponentially weighted moving average, J. Qual. Technol., 18 (1986), 203–210. https://doi.org/10.1080/00224065.1986.11979014 doi: 10.1080/00224065.1986.11979014
    [15] N. Abbas, Homogeneously weighted moving average control chart with an application in substrate manufacturing process, Comput. Ind. Eng., 120 (2018), 460–470. https://doi.org/10.1016/j.cie.2018.05.009 doi: 10.1016/j.cie.2018.05.009
    [16] M. Abid, A. Shabbir, H. Z. Nazir, R. A. K. Sherwani, M. Riaz, A double homogeneously weighted moving average control chart for monitoring of the process mean, Qual. Reliab. Eng. Int., 36 (2020), 1513–1527.
    [17] M. Riaz, Z. Abbas, H. Z. Nazir, M. Abid, On the development of triple homogeneously weighted moving average control chart, Symmetry, 13 (2021), 360. https://doi.org/10.3390/sym13020360 doi: 10.3390/sym13020360
    [18] Z. Rasheed, H. Y. Zhang, S. M. Anwar, B. Zaman, Homogeneously mixed memory charts with application in the substrate production process, Math. Prob. Eng., 2021 (2021), 2582210. https://doi.org/10.1155/2021/2582210 doi: 10.1155/2021/2582210
    [19] Z. Rasheed, H. Y. Zhang, M. Arslan, B. Zaman, S. M. Anwar, M. Abid, et al., An efficient robust nonparametric triple EWMA Wilcoxon signed-rank control chart for process location, Math. Probl. Eng., 2021 (2021), 2570198. https://doi.org/10.1155/2021/2570198 doi: 10.1155/2021/2570198
    [20] Z. Rasheed, M. Khan, N. L. Abiodun, S. M. Anwar, G. Khalaf, S. A. Abbasi, Improved nonparametric control chart based on ranked set sampling with application of chemical data modelling, Math. Probl. Eng., 2022 (2022), 7350204. https://doi.org/10.1155/2022/7350204 doi: 10.1155/2022/7350204
    [21] V. Alevizakos, C. Koukouvinos, A progressive mean control chart for monitoring time between events, Qual. Reliab. Eng. Int., 36 (2020), 161–186. https://doi.org/10.1002/qre.2565 doi: 10.1002/qre.2565
    [22] V. Alevizakos, C. Koukouvinos, Monitoring reliability for a gamma distribution with a double progressive mean control chart, Qual. Reliab. Eng. Int., 37 (2021), 199–218. https://doi.org/10.1002/qre.2730 doi: 10.1002/qre.2730
    [23] D. C. Montgomery, Introduction to Statistical Quality Control, Hoboken: John Wiley & Sons, 2013.
    [24] M. Awais, A. Haq, A new cumulative sum control chart for monitoring the process mean using varied L ranked set sampling, J. Ind. Prod. Eng., 35 (2018), 74–90. https://doi.org/10.1080/21681015.2017.1417787 doi: 10.1080/21681015.2017.1417787
    [25] Z. Wu, J. X. Jiao, M. Yang, Y. Liu, Z. J. Wang, An enhanced adaptive Cusum control chart, ⅡE Trans., 41 (2009), 642–653. https://doi.org/10.1080/07408170802712582 doi: 10.1080/07408170802712582
    [26] M. Khan, M. Aslam, S. M. Anwar, B. Zaman, A robust hybrid exponentially weighted moving average chart for monitoring time between events, Qual. Reliab. Eng. Int., 38 (2022), 895–923. https://doi.org/10.1002/qre.3021 doi: 10.1002/qre.3021
    [27] Y. J. Ou, Z. Wu, F. Tsung, A comparison study of effectiveness and robustness of control charts for monitoring process mean, Int. J. Prod. Econ., 135 (2012), 479–490.
    [28] P. N. Rao, Manufacturing Technology: Metal Cutting and Machine Tools, McGraw-Hill Education (India) Pvt Limited, 2000.
    [29] M. P. Hossain, M. H. Omar, M. Riaz, New V control chart for the Maxwell distribution, J. Stat. Comput. Sim., 87 (2016), 594–606. https://doi.org/10.1080/00949655.2016.1222391 doi: 10.1080/00949655.2016.1222391
    [30] M. P. Hossain, R. A. Sanusi, M. H. Omar, M. Riaz, On designing Maxwell CUSUM control chart: an efficient way to monitor failure rates in boring processes, Int. J. Adv. Manuf. Technol., 100 (2019), 1923–1930.
    [31] R. H. Bonow, J. Barber, N. R. Temkin, W. Videtta, C. Rondina, G. Petroni, et al., The outcome of severe traumatic brain injury in Latin America, World Neurosurg., 111 (2018), e82–e90. https://doi.org/10.1016/j.wneu.2017.11.171 doi: 10.1016/j.wneu.2017.11.171
    [32] P. W. Owens, N. P. Lynch, D. P. O'Leary, A. J. Lowery, M. J. Kerin, Six-year review of traumatic brain injury in a regional trauma unit: demographics, contributing factors and service provision in Ireland, Brain Injury, 32 (2018), 900–906. https://doi.org/10.1080/02699052.2018.1466366 doi: 10.1080/02699052.2018.1466366
    [33] P. L. Ramos, D. C. Nascimento, P. H. Ferreira, K. T. Weber, T. E. G. Santos, F. Louzada, Modeling traumatic brain injury lifetime data: Improved estimators for the generalized Gamma distribution under small samples, Plos One, 14 (2019), e0221332.
  • Reader Comments
  • © 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)
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(1698) PDF downloads(51) Cited by(5)

Figures and Tables

Figures(9)  /  Tables(8)

/

DownLoad:  Full-Size Img  PowerPoint
Return
Return

Catalog