Flexible part time-varying signal deconstruction method based on self-coordination local mode translation

Abstract

The invention discloses a flexible part time-varying signal deconstruction method based on self-coordination local mode translation. The method comprises the following steps of performing Hanning mapping on a flexible part time-varying signal; iteration based on a local modal translation model; performing self-coordination adaptive optimization on the local modal translation model; establishing a self-coordination local mode comprehensive evaluation index; and carrying out self-coordination deconstruction and analysis on the time-varying signal of the flexible component. According to the method, the complex time-varying signal of the flexible part can be effectively captured by constructing and iterating the local modal translation model; through a self-coordination adaptive optimization mechanism, the model can realize dynamic adjustment and accurately deconstruct a time-varying signal; dynamic information which is difficult to reveal in the time-varying signal is effectively revealed, a more accurate and reliable analysis tool is provided for deconstructing the time-varying signal, and the method has remarkable practical research significance.

Description

Technical Field

[0001] The present invention relates to a method for deconstructing time-varying signals of a high-sensitivity flexible component, and in particular to a method for deconstructing time-varying signals of a flexible component based on self-coordinated local modal translation. Background Art

[0002] With the rapid development of industrial equipment and complex engineering systems, traditional signal deconstruction methods face many challenges in dealing with time-varying signals of flexible components. Most existing methods rely on linear analysis methods, which makes it difficult to fully reveal the hidden problems in the operation of flexible components, especially the key information contained in nonlinear dynamic behavior. This limitation not only hinders the accurate assessment of the state of flexible components, but also may cause potential safety threats. In this context, it is necessary to propose a new method to accurately deconstruct the time-varying signals of flexible components. Summary of the invention

[0003] Purpose of the invention: The purpose of the present invention is to provide a method for deconstructing time-varying signals of flexible components based on self-coordinated local modal translation.

[0004] Technical solution: The present invention comprises the following steps:

[0005] (1) Hanning mapping of time-varying signals of flexible components;

[0006] (2) Iteration based on local modal translation model;

[0007] (3) Self-coordinated adaptive optimization of the local modal translation model;

[0008] (4) Establishment of comprehensive evaluation index of self-coordinating local modes;

[0009] (5) Self-coordinated deconstruction and analysis of time-varying signals of flexible components.

[0010] Furthermore, the step (1) comprises:

[0011] According to the running state of the flexible component, the time-varying signal is firstly extracted by a finite impulse response filter, and then the bandwidth of the signal is uniformly mapped using a Hanning window.

[0012] Further, extracting the time-varying signal by using a finite impulse response filter includes:

[0013] For a vibration signal x(t) of length N, a constrained optimization model is constructed to limit the physical and operational conditions in the Hanning mapping process. The constrained optimization problem is defined as:

[0014]

[0015]

[0016] Where M represents the shift order, u k (n) is the kth mode, f k (l) is the kth order filter, T s For the defined cycle period, CK represents signal mapping.

[0017] Furthermore, the uniform mapping process of the bandwidth of the signal using the Hanning window includes:

[0018] The Hanning window is used to evenly divide the signal bandwidth into multiple frequency bands:

[0019]

[0020] Among them, f l is the lower cutoff frequency, f u is the upper cutoff frequency,

[0021] The constrained optimization problem is solved by an iterative feature matrix decomposition algorithm, which uses matrix decomposition to gradually update the system features:

[0022]

[0023] By derivation of the characteristic matrix, the signal mapping CK is defined as:

[0024]

[0025] Among them, Q is the conjugate transpose operation, WM is used to control the weighted correlation matrix,

[0026] By iterating the characteristic matrix, the signal mapping is gradually updated to the optimal state, and the signal mapping result is updated as follows:

[0027]

[0028] Among them, R XWX and R XX denote the weighted correlation matrix and the correlation matrix, respectively.

[0029] Furthermore, the step (2) comprises:

[0030] According to the signal mapping CK determined in step (1), the fitness value of CK obeys the multivariate joint normal distribution characteristics, and the objective function of CK is defined as:

[0031] CK(x)~GP(m(x), k(x, x T ))

[0032] Among them, m(x) is the mean function, k(x, x T) is the covariance function, which is used to characterize the local modal translation process. The exponential square kernel function is selected as the kernel function of the local modal translation model:

[0033]

[0034] Where PP represents the input position x and x T The Euclidean distance between them, l is the characteristic length scale,

[0035] Based on the historical iteration output of the local modal translation model (x 1：t ,y 1：t ), define the latest iterative output and its joint distribution. Specifically, the joint distribution of the iterative output of the local modality translation model and the latest output (x t+1 ,y t+1 ) is defined as:

[0036]

[0037] Among them, m(x 1:t ) represents the mean vector of the previous output, m(x 1：t )=(m(x1), m(x2),..., m(x t )) T , k represents the covariance vector between the iterative output results, defined as:

[0038] k=(k(x t+1 , x1), k(x t+1 , x2),...,k(x t+1 , x t ))

[0039] The covariance matrix of the iterative output is defined as:

[0040]

[0041] According to the above iterative process, the conditional distribution y of the iterative output result of the local modal translation model is t+1 It is expressed as:

[0042]

[0043] Among them, the conditional mean u of the iterative output result of the local modal translation model is t (x t+1 ) and conditional variance Defined as:

[0044]

[0045] Furthermore, the step (3) comprises:

[0046] According to the conditional mean u of the iterative output of the local modal translation model determined in step (2) t (x t+1 ) and conditional variance will u t (x t+1 )and As a posterior measure of the self-coordinated fitness function,

[0047] The self-adaptive optimization process is achieved by optimizing the approximate value of the acquisition function, where the acquisition function is responsible for guiding the search path for the optimal solution. The self-adaptive optimization process is defined as follows:

[0048]

[0049] where χ is the bounded domain of x, To obtain the function, is the posterior distribution of the self-coordinated adaptation process,

[0050] In view of the advantages of the simplicity of the expected improvement (EI) strategy parameters, the expected improvement (EI) is selected as the acquisition function in the self-adaptive optimization process, and the conditional distribution y t+1 The improvement is defined as:

[0051] I(y t+1 ,λ)=max(0,λ-y t+1 )

[0052] in, represents the minimum CK(x) value of the current step, and the expected self-coordinated adaptive optimization can be derived as:

[0053]

[0054] Among them, γ is the standardized variable, φ(γ) is the standard normal cumulative distribution function, is the standard normal probability density function,

[0055] The optimized evaluation point x of the final model t+1 The optimization is determined by minimizing the acquisition function, and the entire optimization iteration process continues until a predetermined number of iterations are completed.

[0056] Furthermore, the step (4) comprises:

[0057] According to the self-adaptive optimization capability of the local modal translation model in step (3), a comprehensive evaluation index is established to effectively guide the self-adaptive optimization process, and a systematic evaluation is performed on the final optimization effect.

[0058] Furthermore, the step (4) comprises:

[0059] For the self-coordinated local mode a(t), the autocorrelation function of the envelope square mode is defined as:

[0060]

[0061] Where SE(a) is the envelope square mode, Hilbert(·) represents the Hilbert transform, and R SE(a) (τ) describes the autocorrelation of SE(x) at time delay i,

[0062] For each cycle iT, the ratio of the autocorrelation amplitude of the cycle to the total autocorrelation amplitude is defined as:

[0063]

[0064] Where Ω represents the number of sampling points in one cycle, and i = 1, 2, ..., Ω,

[0065] For each cycle iT, the cycle square evaluation metric is defined as:

[0066]

[0067] in,

[0068] The functionality of the cycle squared evaluation metric has been enhanced with cycle analysis.

[0069] Furthermore, the step (4) comprises:

[0070] The envelope characteristic frequency ratio is introduced as the core indicator for evaluating the richness of relevant dynamic information.

[0071] For the self-coordinated local mode a(t), the Hilbert transform can be expressed as:

[0072]

[0073] where ι is the integration variable,

[0074] The analytical evaluation of the self-coordinated local mode a(t) can be expressed as:

[0075]

[0076] Where H(t) and ω(t) are the amplitude and phase, respectively, and are calculated as:

[0077]

[0078]

[0079] Through the above calculation, ECFR is defined as the ratio of the relevant dynamic information to the entire envelope mode, expressed as:

[0080]

[0081] Among them, E n (f ij ) represents the frequency f ij The energy at n (A) represents the total energy of the entire envelope mode, Represents the energy of the relevant dynamic information, N h and N c Represents the fundamental frequency number and fault category number respectively. ECFR provides a quantitative perspective to evaluate the total energy E of the envelope mode. n (A)

[0082] In view of the fluctuations caused by small changes in ECFR, an exponential function adjustment mechanism is adopted to keep the processed mode stable when evaluating relevant dynamic information. The comprehensive evaluation index is defined as:

[0083]

[0084] When the value of the self-coordinated local modal comprehensive evaluation index reaches the minimum value, it indicates that the local modal translation model has achieved the best self-coordinated adaptive optimization effect.

[0085] Furthermore, the step (5) comprises:

[0086] According to the self-coordinating local modal comprehensive evaluation index established in step (4), the self-coordinating deconstruction and analysis of the time-varying signal of the flexible component is completed.

[0087] The signal with time-varying characteristics is defined as:

[0088]

[0089] Among them, W n (t) represents amplitude modulation, φ n (t) represents the phase modulation and ε(n) is the noise term.

[0090] The expression for instantaneous frequency is as follows:

[0091]

[0092] The deconstruction coefficient is calculated by W n (p, q), where p represents the scale parameter and q represents the time offset. The deconstruction process is redistributed along the frequency axis.

[0093] The instantaneous frequency of the deconstruction point (p, q) is calculated by the following formula:

[0094]

[0095] The specific manifestation of the deconstruction process is that the time scale plane is compressed to [(q, p) → (q, W n (p, q))],

[0096] Self-coordinated deconstruction and analysis results of time-varying signals of flexible components x (W n , q) is defined as follows:

[0097]

[0098] Among them, a n represents discrete values, (Δa) n =a n -a n-1 , frequency ω l It is the deconstruction center.

[0099] Beneficial effects: Compared with the prior art, the present invention has the following significant advantages: by constructing and iterating a local modal translation model, it can effectively capture the complex time-varying signals of flexible components; through a self-coordinated adaptive optimization mechanism, the model can achieve dynamic adjustment and accurately deconstruct time-varying signals; it can effectively reveal dynamic information that is difficult to reveal in time-varying signals, and provide a more accurate and reliable analysis tool for deconstructing time-varying signals, which has significant practical research significance. BRIEF DESCRIPTION OF THE DRAWINGS

[0100] Figure 1 It is a flow chart of the present invention. DETAILED DESCRIPTION

[0101] The technical solution of the present invention is further described below in conjunction with the accompanying drawings.

[0102] like Figure 1 As shown, the present invention comprises the following steps:

[0103] (1) Hanning mapping of time-varying signals of flexible components,

[0104] According to the running state of the flexible component, the time-varying signal is firstly extracted by a finite impulse response filter, and then the bandwidth of the signal is uniformly mapped using a Hanning window.

[0105] For a vibration signal x(t) of length N, a constrained optimization model is constructed to limit the physical and operational conditions in the Hanning mapping process. The constrained optimization problem is defined as:

[0106]

[0107]

[0108] Where M represents the shift order, u k (n) is the kth mode, fk (l) is the kth order filter, T s For the defined cycle period, CK represents signal mapping.

[0109] The Hanning window is used to evenly divide the signal bandwidth into multiple frequency bands to improve the mapping effect of time-varying signals.

[0110]

[0111] Among them, f l is the lower cutoff frequency, f u is the upper cutoff frequency.

[0112] The constrained optimization problem is solved by an iterative feature matrix decomposition algorithm, which uses matrix decomposition to gradually update the system features:

[0113]

[0114] By derivation of the characteristic matrix, the signal mapping CK can be defined as:

[0115]

[0116] Among them, Q is the conjugate transpose operation, and WM is used to control the weighted correlation matrix.

[0117] By iterating the characteristic matrix, the signal mapping is gradually updated to the optimal state, and the signal mapping result is updated as follows:

[0118]

[0119] Among them, R XWX and R XX denote the weighted correlation matrix and the correlation matrix, respectively.

[0120] (2) Iteration based on local modal translation model

[0121] According to the signal mapping CK determined in step S1 above, the fitness value of CK obeys the multivariate joint normal distribution characteristics. This statistical characteristic makes the result of signal deconstruction more consistent with the model expectations. The objective function of CK is defined as:

[0122] CK(x)~GP(m(x), k(x, x T ))

[0123] Among them, m(x) is the mean function, k(x, x T ) is the covariance function, which is used to characterize the local modal translation process.

[0124] In view of the superior performance of the exponential square kernel function in translation efficiency and accuracy, it is selected as the kernel function of the local modal translation model:

[0125]

[0126] Where PP represents the input position x and x T The Euclidean distance between them, I is the characteristic length scale.

[0127] Based on the historical iteration output of the local modal translation model (x 1：t ,y 1：t ), define the latest iterative output and its joint distribution. Specifically, the joint distribution of the iterative output of the local modality translation model and the latest output (x t+1 ,y t+1 ) is defined as:

[0128]

[0129] Among them, m(x 1：t ) represents the mean vector of the previous output, m(x 1：t )=(m(x1), m(x2),..., m(x t )) T k represents the covariance vector between the iterative output results, which is defined as:

[0130] k=(k(x t+1 , x1), k(x t+1 , x2),...,k(x t+1 , x t ))

[0131] The covariance matrix of the above iterative output can be defined as:

[0132]

[0133] According to the above iterative process, the conditional distribution y of the iterative output result of the local modal translation model is t+1 It can be expressed as:

[0134]

[0135] Among them, the conditional mean u of the iterative output result of the local modal translation model is t (x t+1 ) and conditional variance is defined as:

[0136]

[0137] (3) Self-coordinated adaptive optimization of local modal translation model

[0138] According to the conditional mean u of the iterative output of the local modal translation model determined in step (2) abovet (x t+1 ) and conditional variance will u t (x t+1 )and As a posterior measure of the self-coordinated fitness function, it is used to guide the dynamic adjustment of the self-coordinated adaptive optimization process.

[0139] The self-adaptive optimization process is implemented by optimizing the approximate value of the acquisition function, where the acquisition function is responsible for guiding the search path for the optimal solution. The self-adaptive optimization process is defined as follows:

[0140]

[0141] where χ is the bounded domain of x, To obtain the function. is the posterior distribution of the self-coordinated adaptation process.

[0142] In view of the advantages of the simple parameters of the expected improvement (EI) strategy, the expected improvement (EI) is selected as the acquisition function in the self-adaptive optimization process. t+1 The improvement is defined as:

[0143] I(y t+1 ,λ)=max(0,λ-y t+1 )

[0144] in, represents the minimum CK(x) value of the current step. The expected self-coordinated adaptive optimization can be derived as:

[0145]

[0146] Among them, γ is the standardized variable. φ(γ) is the standard normal cumulative distribution function, is the standard normal probability density function.

[0147] The optimized evaluation point x of the final model t+1 It is determined by minimizing the acquisition function. The entire optimization iteration process continues until the predetermined number of iterations is completed to ensure the accuracy of the optimization process and the final convergence of the model.

[0148] (4) Establishment of comprehensive evaluation index of self-coordinating local modal

[0149] According to the above step (3), the local modal translation model has the ability of self-coordination and adaptive optimization. By establishing a comprehensive evaluation index, the self-coordination and adaptive optimization process is effectively guided, and the final optimization effect is systematically evaluated.

[0150] For the self-coordinated local mode a(t), the autocorrelation function of the envelope square mode is defined as:

[0151]

[0152] Where SE(a) is the envelope square mode, Hilbert(·) represents the Hilbert transform, and R SE(a) (τ) describes the autocorrelation of SE(x) at a delay of τ.

[0153] For each cycle iT, the ratio of the autocorrelation amplitude of the cycle to the total autocorrelation amplitude is defined as:

[0154]

[0155] Wherein, Ω represents the number of sampling points in one cycle, and i=1, 2, ..., Ω.

[0156] For each cycle iT, the cycle square evaluation metric (CSEE) is defined as:

[0157]

[0158] in,

[0159] The functionality of the cycle square evaluation index is enhanced through cycle analysis. In addition, the envelope characteristic frequency ratio (ECFR) is introduced as a core index to evaluate the richness of relevant dynamic information.

[0160] For the self-coordinated local mode a(t), the Hilbert transform can be expressed as:

[0161]

[0162] where ι is the integration variable,

[0163] The analytical evaluation of the self-coordinated local mode a(t) can be expressed as:

[0164]

[0165] Where H(t) and ω(t) are the amplitude and phase, respectively, and are calculated as:

[0166]

[0167]

[0168] Through the above calculation, ECFR is defined as the ratio of the relevant dynamic information to the entire envelope mode, expressed as:

[0169]

[0170] Among them, E n (f ij ) represents the frequency f ij The energy at n (A) represents the total energy of the entire envelope mode, Represents the energy of the relevant dynamic information. N h and N c Represents the fundamental frequency number and fault category number respectively. ECFR provides a quantitative perspective to evaluate the total energy E of the envelope mode. n (A).

[0171] In view of the fluctuations caused by small changes in ECFR, an exponential function adjustment mechanism is adopted to keep the processed mode stable when evaluating relevant dynamic information. The comprehensive evaluation index is defined as:

[0172]

[0173] When the value of the self-coordinated local modal comprehensive evaluation index reaches the minimum value, it indicates that the local modal translation model has achieved the best self-coordinated adaptive optimization effect.

[0174] (5) Self-coordinated deconstruction and analysis of time-varying signals of flexible components

[0175] According to the self-coordinated local modal comprehensive evaluation index established in the above step (4), the self-coordinated deconstruction and analysis of the time-varying signal of the flexible component is completed under the action of the local modal translation model with self-coordinated adaptive optimization capability.

[0176] The signal with time-varying characteristics is defined as:

[0177]

[0178] Among them, W n (t) represents amplitude modulation, φ n (t) represents the phase modulation and ε(n) is the noise term.

[0179] The expression for instantaneous frequency is as follows:

[0180]

[0181] The deconstruction coefficient is calculated by W n (p, q), where p represents the scale parameter and q represents the time offset. This deconstruction process redistributes along the frequency axis, effectively reducing the blurring effect.

[0182] The instantaneous frequency of the deconstruction point (p, q) is calculated by the following formula:

[0183]

[0184] The specific manifestation of the deconstruction process is that the time scale plane is compressed to [(q, p) → (q, W n (p, q))].

[0185] Self-coordinated deconstruction and analysis results of time-varying signals of flexible components x (W n , q) is defined as follows:

[0186]

[0187] Among them, a n represents discrete values, (Δa) n =a n -a n-1 , frequency ω l It is the deconstruction center.

Claims

1. A method for deconstructing time-varying signals of flexible components based on self-coordinated local modal translation, characterized in that: The method comprises the following steps: (1) Hanning mapping of time-varying signals of flexible components; (2) Iteration based on local modal translation model; (3) Self-coordinated adaptive optimization of the local modal translation model; (4) Establishment of comprehensive evaluation index of self-coordinating local modes; (5) Self-coordinated deconstruction and analysis of time-varying signals of flexible components.

2. The method for deconstructing time-varying signals of flexible components based on self-coordinated local modal translation according to claim 1, characterized in that: The step (1) comprises: According to the running state of the flexible component, the time-varying signal is firstly extracted by a finite impulse response filter, and then the bandwidth of the signal is uniformly mapped using a Hanning window.

3. The method for deconstructing time-varying signals of flexible components based on self-coordinated local modal translation according to claim 2, characterized in that: The extracting of the time-varying signal by a finite impulse response filter comprises: For a vibration signal x(t) of length N, a constrained optimization model is constructed to limit the physical and operational conditions in the Hanning mapping process. The constrained optimization problem is defined as: Where M represents the shift order, u k (n) is the kth mode, f k (l) is the kth order filter, T s For the defined cycle period, CK represents signal mapping.

4. The method for deconstructing time-varying signals of flexible components based on self-coordinated local modal translation according to claim 2, characterized in that: The method of uniformly mapping the bandwidth of a signal using a Hanning window comprises: The Hanning window is used to evenly divide the signal bandwidth into multiple frequency bands: Among them, f l is the lower cutoff frequency, f u is the upper cutoff frequency, The constrained optimization problem is solved by an iterative feature matrix decomposition algorithm, which uses matrix decomposition to gradually update the system features: By derivation of the characteristic matrix, the signal mapping CK is defined as: Among them, Q is the conjugate transpose operation, WM is used to control the weighted correlation matrix, By iterating the characteristic matrix, the signal mapping is gradually updated to the optimal state, and the signal mapping result is updated as follows: Among them, R XWX and R XX denote the weighted correlation matrix and the correlation matrix, respectively.

5. The method for deconstructing time-varying signals of flexible components based on self-coordinated local modal translation according to claim 1, characterized in that: The step (2) comprises: According to the signal mapping CK determined in step (1), the fitness value of CK obeys the multivariate joint normal distribution characteristics, and the objective function of CK is defined as: CK(x)~GP(m(x),k(x,x T )) Among them, m(x) is the mean function, k(x, x T ) is the covariance function, which is used to characterize the local modal translation process. Select the exponential square kernel function as the kernel function of the local modal translation model: Where P·P represents the input position x and x T The Euclidean distance between them, I is the characteristic length scale, Based on the historical iteration output of the local modal translation model (x 1：t ,y 1：t ), define the latest iterative output and its joint distribution. Specifically, the joint distribution of the iterative output of the local modality translation model and the latest output (x t+1 ,y t+1 ) is defined as: Among them, m(x 1:t ) represents the mean vector of the previous output, m(x 1：t )=(m(x1), m(x2),..., m(x t )) T , k represents the covariance vector between the iterative output results, defined as: k=(k(x t+1 ,x1),k(x t+1 ,x2),...,k(x t+1 x t )) The covariance matrix of the iterative output is defined as: According to the above iterative process, the conditional distribution y of the iterative output result of the local modal translation model is t+1 It is expressed as: Among them, the conditional mean u of the iterative output result of the local modal translation model is t (x t+1 ) and conditional variance Defined as:

6. The method for deconstructing time-varying signals of flexible components based on self-coordinated local modal translation according to claim 1, characterized in that: The step (3) comprises: According to the conditional mean u of the iterative output of the local modal translation model determined in step (2) t (x t+1 ) and conditional variance will u t (x t+1 )and As a posterior measure of the self-coordinated fitness function, The self-adaptive optimization process is achieved by optimizing the approximate value of the acquisition function, where the acquisition function is responsible for guiding the search path for the optimal solution. The self-adaptive optimization process is defined as follows: where χ is the bounded domain of x, To obtain the function, is the posterior distribution of the self-coordinated adaptation process, In view of the advantages of the simplicity of expected improvement strategy parameters, the expected improvement is selected as the acquisition function in the self-adaptive optimization process, and the conditional distribution y t+1 The improvement is defined as: I(y t+1 ,λ)=max(0,λ-y t+1 ) in, represents the minimum CK(x) value of the current step, and the expected self-coordinated adaptive optimization can be derived as: Among them, γ is the standardized variable, Φ(γ) is the standard normal cumulative distribution function, is the standard normal probability density function, The optimized evaluation point x of the final model t+1 The optimization is determined by minimizing the acquisition function, and the entire optimization iteration process continues until a predetermined number of iterations are completed.

7. The method for deconstructing time-varying signals of flexible components based on self-coordinated local modal translation according to claim 1, characterized in that: The step (4) comprises: According to the self-adaptive optimization capability of the local modal translation model in step (3), a comprehensive evaluation index is established to effectively guide the self-adaptive optimization process, and a systematic evaluation is performed on the final optimization effect.

8. The method for deconstructing time-varying signals of flexible components based on self-coordinated local modal translation according to claim 1, characterized in that: The step (4) comprises: For the self-coordinated local mode a(t), the autocorrelation function of the envelope square mode is defined as: Where SE(a) is the envelope square mode, Hilbert(·) represents the Hilbert transform, and R SE(a) (v) describes the autocorrelation of SE(x) at a delay of τ, For each cycle iT, the ratio of the autocorrelation amplitude of the cycle to the total autocorrelation amplitude is defined as: Where Ω represents the number of sampling points in one cycle, and i = 1, 2, ..., Ω, For each cycle iT, the cycle square evaluation metric is defined as: in, The functionality of the cycle squared evaluation metric has been enhanced with cycle analysis.

9. The method for deconstructing time-varying signals of flexible components based on self-coordinated local modal translation according to claim 1, characterized in that: The step (4) comprises: The envelope characteristic frequency ratio is introduced as the core indicator for evaluating the richness of relevant dynamic information. For the self-coordinated local mode a(t), the Hilbert transform can be expressed as: where ι is the integration variable, The analytical evaluation of the self-coordinated local mode a(t) can be expressed as: Where H(t) and ω(t) are the amplitude and phase, respectively, and are calculated as: Through the above calculation, ECFR is defined as the ratio of the relevant dynamic information to the entire envelope mode, expressed as: Among them, E n (f ij ) represents the frequency f ij The energy at n (A) represents the total energy of the entire envelope mode, Represents the energy of the relevant dynamic information, N h and N c Represents the fundamental frequency number and fault category number respectively. ECFR provides a quantitative perspective to evaluate the total energy E of the envelope mode. n (A) In view of the fluctuations caused by small changes in ECFR, an exponential function adjustment mechanism is adopted to keep the processed mode stable when evaluating relevant dynamic information. The comprehensive evaluation index is defined as: When the value of the self-coordinated local modal comprehensive evaluation index reaches the minimum value, it indicates that the local modal translation model has achieved the best self-coordinated adaptive optimization effect.

10. The method for deconstructing time-varying signals of flexible components based on self-coordinated local modal translation according to claim 1, characterized in that: The step (5) comprises: According to the self-coordinating local modal comprehensive evaluation index established in step (4), the self-coordinating deconstruction and analysis of the time-varying signal of the flexible component is completed. The signal with time-varying characteristics is defined as: Among them, W n (t) represents amplitude modulation, φ n (t) represents the phase modulation and ε(n) is the noise term. The expression for instantaneous frequency is as follows: The deconstruction coefficient is calculated by W n (p, q), where p represents the scale parameter and q represents the time offset. The deconstruction process is redistributed along the frequency axis. The instantaneous frequency of the deconstruction point (p, q) is calculated by the following formula: The specific manifestation of the deconstruction process is that the time scale plane is compressed to [(q, p) → (q, W n (p, q))], Self-coordinated deconstruction and analysis results of time-varying signals of flexible components x (W n , q) is defined as follows: Among them, a n Represents discrete values, (Δa) n =a n -a n-1 , frequency ω l It is the deconstruction center.

Images