Actuator Performance Prediction Method for Hidden Degradation of Nonlinear Closed-Loop Control Systems

Abstract

The present invention provides a method and device for predicting the performance of an actuator with implicit degradation in a nonlinear closed-loop control system, relating to the technical field of remaining useful life. The method comprises: acquiring nonlinear system control data, constructing implicit performance indicators for the implicit degradation characteristics of the actuator structure, and establishing a degradation model using a Wiener random process; designing a prediction algorithm architecture that integrates a FNN algorithm and an adaptive Kalman filter, analyzing the estimation error of the degradation state using unmodeled dynamics to obtain an error result; improving the filter structure of the adaptive Kalman filter, designing an online weight update mechanism for the FNN algorithm, and using the updated FNN algorithm weights to estimate the unmodeled dynamic terms in real time to obtain updated unmodeled dynamic terms; updating the covariance distribution of the implicit performance indicators and degradation model parameters, and calculating using a full probability formula to obtain a remaining useful life prediction result. The present invention can improve the accuracy of the remaining useful life prediction results for the system.

Description

Technical Field

[0001] The present invention relates to the technical field of remaining service life prediction, and in particular to a method and device for predicting the performance of an actuator of a nonlinear closed-loop control system with latent degradation. Background Art

[0002] The development of industrial modernization is a dominant force affecting the lifeblood of the national economy. In actual industrial production, industrial control systems issue control commands to controlled objects based on the requirements of industrial products. Actuators, represented by various motors, play a particularly important role as the crucial link between industrial controllers and controlled objects for signal transmission and conversion. Therefore, with increasing demands for safety and reliability, the need for necessary performance monitoring and prediction of degraded actuators, timely and accurate prediction of their remaining useful life, and the implementation of effective maintenance strategies can prevent significant financial losses and significantly enhance the reliability and safety of control systems. Current research on actuator performance degradation and remaining useful life prediction is typically based on measurable performance indicators, focusing on remaining useful life prediction for components in open-loop control systems. Currently, algorithms based on stochastic processes are widely used for remaining useful life prediction because they can explicitly describe degradation models and prediction uncertainty. However, due to sensor bias, external environmental variations, and internal uncertainty, discrepancies between established degradation models and actual production processes are inevitable. To achieve the remaining useful life for measurable performance indicators of open-loop systems, a degradation model incorporating the triple uncertainties of measurement-related, individual characteristic-related, and time-related variations can be established using a Wiener stochastic process fused with a Kalman filter. However, for predicting actuator performance degradation in closed-loop control systems, a generalized degradation model capable of characterizing uncertainty must be based on implicit degradation performance indicators that cannot be measured in real time. General nonlinear filters struggle to accurately describe degradation models from highly nonlinear data, and instead must account for unmodeled dynamics introduced during the linearization process. To predict actuator degradation in more complex and common nonlinear control systems, compensating for unmodeled dynamic terms in the evaluation of implicit performance indicators and the distribution of remaining useful life plays a crucial role in improving prediction results.

[0003] To improve filter accuracy without changing the system structure, artificial neural networks (ANNs) offer an effective solution for system modeling and nonlinear fitting within the filter framework. Extensive research on the fusion of filters and neural networks has focused on external and internal collaboration between the filter and the neural network. In the widely studied external collaboration model, independently operating filter and neural network modules are typically connected in parallel to obtain an accurate system model. This external collaboration model typically involves parameter identification and model fitting before and after the filter, while the filter design is unaffected by the neural network output errors. In contrast, in the internal collaboration model, which has seen relatively little research, the neural network is applied to each filter iteration, and the filter's gain matrix and auxiliary matrix vary with the neural network output. Currently, most such algorithms, based on the fusion of neural network algorithms and classical Kalman filters, only evaluate unmodeled dynamic terms. Further research is needed to design adaptive Kalman filters that account for neural network output errors and simultaneously estimate unmodeled dynamic terms and implicit performance indicators in the form of augmentation. Summary of the Invention

[0004] To address the existing technical problem of difficulty in modeling and predicting the implicit degradation performance of actuators in nonlinear closed-loop control systems under the influence of unmodeled dynamic terms, the present invention provides a method and apparatus for predicting the performance of actuators in nonlinear closed-loop control systems with implicit degradation. The technical solution is as follows:

[0005] In one aspect, a method for predicting the performance of an actuator of a nonlinear closed-loop control system with implicit degradation is provided. The method is implemented by a device for predicting the performance of an actuator of a nonlinear closed-loop control system with implicit degradation. The method comprises:

[0006] S1. Acquire nonlinear system control data, construct an implicit performance indicator capable of characterizing implicit degradation characteristics of an execution structure based on the system control data and a system state model, and establish a degradation model using a Wiener random process for the implicit performance indicator;

[0007] S2. Based on the degradation model, design a prediction algorithm architecture that integrates the FNN algorithm and the adaptive Kalman filter, and obtain an error result by analyzing the estimation error of the degradation state of the unmodeled dynamics; based on the error result, improve the filter structure of the adaptive Kalman filter to obtain an improved gain matrix of the adaptive Kalman filter and an updated adaptive Kalman filter structure;

[0008] S3. Designing an online weight update mechanism for the FNN algorithm based on the gain matrix of the improved adaptive Kalman filter, and updating the estimate of the unmodeled dynamic term in real time by using the updated weights of the FNN algorithm to obtain an updated unmodeled dynamic term;

[0009] S4. Update the covariance distribution of the implicit performance indicator and the degradation model parameters according to the updated adaptive Kalman filter structure, the improved gain matrix, and the updated unmodeled dynamic term to obtain independent distribution updates and conditional distribution updates of the implicit performance indicator and the degradation model parameters;

[0010] S5. Based on the independent distribution update and conditional distribution update of the implicit performance indicators and degradation model parameters, the full probability formula is used for calculation to obtain the remaining service life prediction result represented by the probability density function.

[0011] On the other hand, a device for predicting the performance of an actuator of a nonlinear closed-loop control system with implicit degradation is provided. The device is applied to a method for predicting the performance of an actuator of a nonlinear closed-loop control system with implicit degradation. The device comprises:

[0012] an acquisition and construction unit for acquiring nonlinear system control data, constructing an implicit performance indicator capable of characterizing implicit degradation characteristics of the execution structure based on the system control data and a system state model, and establishing a degradation model for the implicit performance indicator using a Wiener random process;

[0013] A design unit is used to design a prediction algorithm architecture that integrates the FNN algorithm and the adaptive Kalman filter according to the degradation model, obtain an error result by analyzing the estimation error of the degradation state of the unmodeled dynamics; and improve the filter structure of the adaptive Kalman filter according to the error result to obtain an improved gain matrix of the adaptive Kalman filter and an updated adaptive Kalman filter structure;

[0014] The first updating unit is used to design an online weight updating mechanism of the FNN algorithm according to the gain matrix of the improved adaptive Kalman filter, and to update the estimation of the unmodeled dynamic term in real time by using the updated weight of the FNN algorithm to obtain an updated unmodeled dynamic term.

[0015] a second updating unit, which updates the covariance distribution of the implicit performance indicator and the degradation model parameters according to the updated adaptive Kalman filter structure, the improved gain matrix, and the updated unmodeled dynamic term, to obtain independent distribution updates and conditional distribution updates of the implicit performance indicator and the degradation model parameters;

[0016] The calculation unit uses the full probability formula to perform calculations based on the independent distribution update and conditional distribution update of the implicit performance indicators and degradation model parameters to obtain the remaining service life prediction result expressed as a probability density function.

[0017] On the other hand, a device for predicting the performance of an actuator of a nonlinear closed-loop control system with implicit degradation is provided. The device for predicting the performance of an actuator of a nonlinear closed-loop control system with implicit degradation comprises: a processor; and a memory, wherein the memory stores computer-readable instructions. When the computer-readable instructions are executed by the processor, any one of the above-mentioned methods for predicting the performance of an actuator of a nonlinear closed-loop control system with implicit degradation is implemented.

[0018] On the other hand, a computer-readable storage medium is provided, wherein the storage medium stores at least one instruction, and the at least one instruction is loaded and executed by a processor to implement any one of the above-mentioned actuator performance prediction methods for implicit degradation of a nonlinear closed-loop control system.

[0019] The beneficial effects brought about by the technical solution provided by the embodiment of the present invention include at least:

[0020] The embodiment of the present invention first obtains nonlinear system control data, and constructs an implicit performance indicator that can characterize the implicit degradation characteristics of the execution structure based on the system control data and the system state model. For the implicit performance indicator, a degradation model is established using the Wiener random process; secondly, based on the degradation model, a prediction algorithm architecture that integrates the FNN algorithm and the adaptive Kalman filter is designed, and an error result is obtained by analyzing the estimation error of the unmodeled dynamic to the degradation state; based on the error result, the filter structure of the adaptive Kalman filter is improved to obtain the gain matrix of the improved adaptive Kalman filter and the updated adaptive Kalman filter structure; based on the improved adaptive Kalman filter The gain matrix of the FNN algorithm is used to design an online weight update mechanism for the FNN algorithm. The estimation of the unmodeled dynamic term is updated in real time by the updated weights of the FNN algorithm to obtain the updated unmodeled dynamic term. According to the updated adaptive Kalman filter structure, the improved gain matrix and the updated unmodeled dynamic term, the covariance distribution of the implicit performance indicator and the degradation model parameters is updated to obtain the independent distribution update and conditional distribution update of the implicit performance indicator and the degradation model parameters. Finally, based on the independent distribution update and conditional distribution update of the implicit performance indicator and the degradation model parameters, the full probability formula is used for calculation to obtain the remaining useful life prediction result expressed as a probability density function.

[0021] An embodiment of the present invention introduces a feedforward neural network into a prediction framework based on an adaptive Kalman filter, establishes an online update mechanism for the neural network weights, and proves the convergence of the prediction error by establishing a Lyapunov function, thereby realizing compensation for the unmodeled dynamic term. Due to the introduction of the unmodeled dynamic compensation term, the gain term of the adaptive Kalman filter is improved, so that the degradation index and degradation model parameters are more accurately estimated. The distribution of the adaptive Kalman filter using the improved gain matrix is ​​re-derived, and the probability density function of the remaining useful life is also synchronously updated.

[0022] By integrating a neural network with nonlinear fitting and a Kalman filter robust to external uncertainties, the present invention accurately describes the degradation path of actuators in closed-loop control systems while simultaneously preventing the influence of unmodeled dynamic terms in the degradation model on the remaining useful life prediction results. This invention improves the accuracy of actuator performance prediction for latent degradation in nonlinear closed-loop control systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0024] Figure 1 1 is a schematic structural diagram of an actuator performance prediction for a nonlinear closed-loop control system with implicit degradation provided by an embodiment of the present invention;

[0025] Figure 2 This is a flow chart of a method for predicting the performance of an actuator of a nonlinear closed-loop control system with implicit degradation provided by an embodiment of the present invention;

[0026] Figure 3 This is a block diagram of a device for predicting the performance of an actuator of a nonlinear closed-loop control system with implicit degradation provided by an embodiment of the present invention;

[0027] Figure 4 It is a structural schematic diagram of an actuator performance prediction device for implicit degradation of a nonlinear closed-loop control system provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0028] The technical solution of the present invention is described below in conjunction with the accompanying drawings.

[0029] In the embodiments of the present invention, words such as "exemplarily" and "for example" are used to indicate examples, illustrations, or explanations. Any embodiment or design described as an "exemplary" in the present invention should not be interpreted as being preferred or advantageous over other embodiments or designs. Rather, the use of the word "exemplary" is intended to present concepts in a concrete manner. Furthermore, in the embodiments of the present invention, "and / or" can mean both or either of the two.

[0030] In the embodiments of the present invention, the terms "image" and "picture" may be used interchangeably. It should be noted that, when the distinction between them is not emphasized, their intended meanings are the same. The terms "of," "corresponding," and "corresponding" may be used interchangeably. It should be noted that, when the distinction between them is not emphasized, their intended meanings are the same.

[0031] In the embodiments of the present invention, sometimes a subscript such as W1 may be written as a non-subscript such as W1. When the difference is not emphasized, the meanings to be expressed are the same.

[0032] In order to make the technical problems, technical solutions and advantages to be solved by the present invention clearer, a detailed description will be given below with reference to the accompanying drawings and specific embodiments.

[0033] The embodiment of the present invention provides a method for predicting the performance of an actuator of a nonlinear closed-loop control system with implicit degradation. The method can be implemented by a device for predicting the performance of an actuator of a nonlinear closed-loop control system with implicit degradation. The device can be a terminal or a server. Figure 1The figure shows a structural schematic diagram of the performance prediction of an actuator of a nonlinear closed-loop control system with implicit degradation provided by an embodiment of the present invention. In a feasible implementation method, for the remaining useful life prediction process of the control system actuator under the action of a feedback controller, due to the existence of estimation errors of implicit performance degradation indicators, the unmodeled dynamic terms in the degradation model established based on the traditional nonlinear filter have a significant impact on the prediction results. Therefore, a performance prediction framework based on the unmodeled dynamic and implicit performance indicators of the degradation model with synchronous updates is designed by combining the FNN neural network with an adaptive Kalman filter. The gain matrix of the adaptive Kalman filter based on the least squares filtering and classical Kalman filtering principles is recalculated, and the convergence of the degradation information estimation value containing the implicit degradation indicator is also proved. In order to compensate for the missing unmodeled dynamic terms in the linearized model, the degradation state estimation residual is used to guide the weight update of the hidden layer of the neural network. Based on the real-time updated degradation state estimation value, degradation state distribution value and degradation model parameters, the probability density function of the remaining useful life of the actuator evolving in the form of implicit degradation is updated in real time.

[0034] like Figure 2 The flowchart of the method for predicting the performance of an actuator of a nonlinear closed-loop control system with implicit degradation is shown. The processing flow of the method may include the following steps:

[0035] S1. Obtain nonlinear system control data. Based on the system control data and the system state model, construct implicit performance indicators that can characterize the implicit degradation characteristics of the execution structure. For the implicit performance indicators, use the Wiener random process to establish a degradation model.

[0036] The nonlinear system control data includes: state data of the nonlinear system, control output data and system output data.

[0037] Optionally, the specific implementation process of S1 may include S11-S13:

[0038] S11, obtaining state data, control output data, and system output data of the nonlinear system;

[0039] Among them, the state model of the nonlinear control system can be expressed by formula (1):

[0040]

[0041] Among them, x k represents the state of the system at time k; u k represents the control output of the system at time k; y k represents the output of the system at time k; f(·) represents the nonlinear mapping function; C represents the output equation coefficient matrix; ω k~N(0,V1) represents the first vector of uncorrelated white noise; ω k ~N(0,V2) represents the second vector of white noise that is uncorrelated with each other; V1 represents the variance term of the first vector of white noise; V2 represents the variance term of the second vector of white noise; N represents the normal distribution function; k represents the sampling time.

[0042] S12. Using a classical nonlinear Kalman filter, the state model of the nonlinear control system is linearized to obtain a priori system model with implicit degradation performance indicators of the execution structure;

[0043] Among them, the equation of the implicit performance index with the implicit degradation characteristics of the execution structure can be expressed by formula (2):

[0044]

[0045] in, Indicates that the parameter matrix of the first state equation is the Jacobian matrix; Indicates that the parameter matrix of the second state equation is the Jacobian matrix; λ k Indicates the implicit degradation performance indicator with a value between 0 and 1.

[0046] Among them, subject to the unmeasurable λ k Due to the influence of , the state equation parameter matrix of the linearized process deviates from the actual model. The deviation is regarded as the unmodeled dynamics. According to formula (2), an accurate model that characterizes the system state including the unmodeled dynamics can be obtained, where the model can be expressed by the following formula (3):

[0047] x k =A k x k-1 +B k u k +E k λ k +Φ k (3)

[0048] Among them, Φ k represents unmodeled dynamics; A k Represents the state parameter matrix of the system state equation; B k Represents the input parameter matrix of the system state equation; E k Represents the implicit performance index parameter of the system state equation; x k-1 Represents the system state vector at the last sampling moment.

[0049] S13. Based on the equation of the implicit performance index with the implicit degradation characteristics of the execution structure, a degradation model of the implicitly degraded actuator is established using the Wiener random process.

[0050] When the performance index is not implicit, the degradation model of the implicitly degraded actuator is established using the Wiener random process. The process of using the Wiener process to establish the degradation model for the measurable performance index can be expressed by the following formula (4):

[0051]

[0052] Among them, λ k-1 Represents the implicit performance index value at the previous moment; θ k represents the drift parameter of the degradation model; θ k-1 represents the drift parameter of the degradation model at the previous moment; r k represents the first vector of white noise in the degradation model; κ k represents the second vector of white noise in the degradation model; T Time,k represents the running time of sampling k points; T Time,k-1 represents the running time of sampling k-1 points.

[0053] In a feasible implementation, due to the implicit performance indicator λ k It is unmeasurable. According to formula (3) and formula (4), the degradation model of the implicitly degraded actuator in the system can be established. The degradation model of the implicitly degraded actuator in the system can be expressed by the following formula (5):

[0054]

[0055] S2. Based on the degradation model, a prediction algorithm architecture that integrates the FNN algorithm and the adaptive Kalman filter is designed. The error result is obtained by analyzing the estimation error of the unmodeled dynamics on the degradation state. Based on the error result, the filter structure of the adaptive Kalman filter is improved to obtain the improved adaptive Kalman filter gain matrix and the updated adaptive Kalman filter structure.

[0056] Optionally, the specific implementation process of S2 may include S21-S23:

[0057] S21, using adaptive Kalman filter to evaluate degradation state;

[0058] In a feasible implementation, formula (4) is rearranged to obtain a generalized model of the state, wherein the generalized model of the state can be expressed by the following formula (6):

[0059] η k =Dη k-1 +ζ k (6)

[0060] in, represents the parameter matrix of the generalized model; ΔT Time =T Time,k -TTime,k-1 Represents the difference between two sampling moments; represents the white noise vector of the generalized model; Represents x k and η k Augmentation vector; η k Represents λ k and θ k The augmentation vector.

[0061] Among them, by sorting out formula (5), we can get x k and η k Augmentation vector The generalized model can be expressed by the following formula (7):

[0062]

[0063] in, F k =[E k 0], in, represents the state parameter matrix of the generalized model; represents the input parameter matrix of the generalized model; Φ k represents the unmodeled dynamics of the generalized model; F k Representation matrix The internal matrix of represents the white noise vector of the generalized model;

[0064] Among them, if there is no unmodeled dynamics in formula (7), the degradation model is only established for the time-varying system with implicit performance indicators.

[0065] Among them, the adaptive Kalman filter can be used to estimate the degradation state

[0066] S22. Based on the degradation assessment state and the degradation model, a fusion FNN adaptive Kalman filter algorithm is designed to obtain the error result by analyzing the estimation error of the degradation state of the unmodeled dynamics;

[0067] The process of designing the fusion FNN adaptive Kalman filter algorithm can be expressed by the following formula (8):

[0068]

[0069] in, represents the prior estimate of the system state at time k at time k-1; represents the posterior estimate of the system state at time k; represents the posterior estimate at time k-1; Denotes the k-1 time for η k-1 estimated value of; represents the effect of time k on η k-1 estimated value of; Denotes the k-time relationship with Φ k estimates; represents time k Φ k The estimation error of represents the system output error at time k; k K represents the first auxiliary matrix of the adaptive Kalman filter at time k; k represents the second auxiliary matrix of the adaptive Kalman filter at time k; S k Represents the third auxiliary matrix of the adaptive Kalman filter at time k.

[0070] The system output error at time k can be expressed by the following formula (9):

[0071]

[0072] In a feasible implementation, according to formula (5) and formula (8), the system state estimation error at time k can be obtained, which can be expressed by the following formula (10):

[0073]

[0074] S23. According to the error result, an auxiliary matrix is ​​established to improve the structure of the adaptive Kalman filter, a gain matrix of the improved adaptive Kalman filter is obtained, and the structure of the adaptive Kalman filter is updated.

[0075] In a feasible implementation, in order to clearly represent the relationship between the various states, an auxiliary matrix is ​​established, wherein the auxiliary matrix can be expressed by the following formula (11):

[0076]

[0077] Among them, ε k represents the first auxiliary vector of the adaptive Kalman filter at time k; S k-1 R represents the third auxiliary matrix of the adaptive Kalman filter at time k; k Represents the fourth auxiliary matrix of the adaptive Kalman filter at time k.

[0078] According to the auxiliary matrix, the state model of the nonlinear control system can be modified into the following formula (12):

[0079]

[0080] in, represents the system state estimation error at time k-1; ε k-1 represents the first auxiliary vector of the adaptive Kalman filter at time k-1; S k-1 represents the third auxiliary matrix of the adaptive Kalman filter at time k-1; R k-1 Represents the fourth auxiliary matrix of the adaptive Kalman filter at time k-1; Denotes the k-1 time for η k-1 estimated value of; represents k-1 time Φ k The estimation error.

[0081] Substituting formula (10) and formula (2) into formula (11), the expression of the first auxiliary vector of the adaptive Kalman filter at time k can be obtained, which is expressed by the following formula (13):

[0082]

[0083] Among them, according to the properties of the classic Kalman filter, we can get E(ω k )=0 and E(υ k )=0, wherein the second auxiliary matrix and the fourth auxiliary matrix of the adaptive Kalman filter can be expressed by the following formula (14):

[0084]

[0085] Among them, P k|k-1 represents the prior covariance of the system state at time k-1 to time k; P k-1|k-1 represents the posterior covariance of the system state at time k-1; P k|k represents the posterior covariance of the system state at time k; Σ k represents the fifth auxiliary matrix of the adaptive Kalman filter at time k; G k represents the sixth auxiliary matrix of the adaptive Kalman filter at time k; J k H represents the seventh auxiliary matrix of the adaptive Kalman filter at time k; k represents the eighth auxiliary matrix of the adaptive Kalman filter at time k; H k-1 represents the eighth auxiliary matrix of the adaptive Kalman filter at time k-1; when the representation of the third auxiliary matrix is ​​designed as the following formula (15), the error term can be proved to gradually tend to zero.

[0086] The third auxiliary matrix can be expressed by the following formula (15):

[0087]

[0088] S3. Based on the gain matrix of the improved adaptive Kalman filter, an online weight update mechanism of the FNN algorithm is designed. The estimation of the unmodeled dynamic term is updated in real time by the updated weight of the FNN algorithm to obtain an updated unmodeled dynamic term.

[0089] Optionally, after S3 updates the estimate of the unmodeled dynamic term in real time by using the updated weights of the FNN algorithm, it further includes:

[0090] The unmodeled dynamic terms are updated by updating the FNN weights, and the Lyapunov function is designed to verify the stability and convergence of the filter.

[0091] In a feasible implementation, based on the adaptive Kalman filter, the augmentation vector can be obtained: The estimated value at time k, where the estimated value of the augmented vector at time k can be expressed by the following formula (16):

[0092]

[0093] in, Indicates the augmented vector at time k estimated value of; Indicates the augmented vector at time k-1 Estimated value of Μ k Represents the parameter matrix of the output error at time k; Γ k represents the parameter matrix of the unmodeled dynamic estimation error at time k;

[0094] in,

[0095] Among them, e k-1 Indicates the augmented vector at time k-1 The estimation error of

[0096] in,

[0097] The state parameters can be expressed by the following formula (17):

[0098]

[0099] In one embodiment, when the weights are updated accurately, the actual unmodeled dynamic term can be expressed as follows using the FNN algorithm:

[0100]

[0101] Among them, Φ k represents the actual unmodeled dynamic term; W k Represents the actual weight from the hidden layer to the output layer of the FNN at time k; Indicates that at time k is a fixed Gaussian distribution function from the input to the FNN output layer and the hidden layer;

[0102] The estimation error can be expressed by the following formula (19):

[0103]

[0104] in, Indicates the response of W at time k k estimated value of; Indicates the response of W at time k k The estimation error of The first constant term has negligible influence;

[0105] The error of the weight can be expressed by the following formula (20):

[0106]

[0107] in,

[0108] in, Indicates that at time k is a fixed Gaussian distribution function from the input to the FNN output layer and the hidden layer; Indicates that at time k is a fixed Gaussian distribution function from the FNN output layer to the hidden layer as input; wherein, according to the above formula, the state parameter can be further rewritten and expressed by the following formula (21):

[0109]

[0110] Among them, e k Indicates the augmented vector at time k The estimation error of Ω k Represents the ninth auxiliary matrix of the adaptive Kalman filter at time k;

[0111] In a feasible implementation, the Lyapunov function can be expressed by the following formula (22):

[0112]

[0113] Where P represents the tuning constant matrix; V k represents the Lyapunov function at time k;

[0114] According to the Lyapunov function, the Lyapunov function at time k-1 can be obtained, which can be expressed by the following formula (23):

[0115]

[0116] Among them, V k-1 represents the Lyapunov function at time k-1; tr represents the trace function; Indicates the K-1 time for W k-1 The estimation error of Γ k-1 represents the parameter matrix of the unmodeled dynamic estimation error at time k-1;

[0117] In one possible implementation, to prove stability, the discrete Lyapunov function needs to prove V k With V k-1 The difference ΔV k , meeting certain conditions, the specific implementation process may include:

[0118] make The following formula (24) can be obtained:

[0119]

[0120] Among them, ε′ is The second constant term has negligible influence; where ΔV k Indicates V k With V k-1 The difference,

[0121] Among them, if the vector is augmented at time k-1 The second norm of the estimation error ||e k-1 || and W at time k-1 k-1 The second norm of the estimation error Exceeding the maximum interval range, using the designed W for k+1 time k+1 The update rate of prediction weights The error of the unmodeled dynamic term can be limited. Among them, the improved filter gain and the online weight update mechanism of the FNN can ensure the accurate estimation of the expected and distributed degradation states.

[0122] S4. Update the covariance distribution of the implicit performance index and the degradation model parameters according to the updated adaptive Kalman filter structure, the improved gain matrix and the updated unmodeled dynamic terms to obtain independent distribution updates and conditional distribution updates of the implicit performance index and the degradation model parameters.

[0123] In a feasible implementation, according to the above formula (8), the k-time η k The predicted value can be expressed by formula (25):

[0124]

[0125] In one feasible implementation, Substituting into formula (25), we can obtain the following formula (26) for η at time k: k The prediction error

[0126]

[0127] in, Represents the k-1 time for η k-1 The prediction error of Represents a linear combination vector of white noise sequences containing independent distributions, at time k η k The estimated error covariance of The calculation process can be expressed by the following formula (27):

[0128]

[0129] in, Indicates that at time k-1 η k The estimated error covariance of ; Indicates that at time k-1 x k The estimated error covariance of It can be obtained through the covariance property of the classic Kalman filter;

[0130] make λ k Relative θ k The conditional distribution p(λ k |θ k ),λ k Relative θ k The distribution of p(θ k ) and λ k Relative λ k The distribution of p(λ k ), can be expressed by the following formula (28):

[0131]

[0132] S5. Based on the independent distribution update and conditional distribution update of the implicit performance indicators and degradation model parameters, the full probability formula is used for calculation to obtain the remaining service life prediction result represented by the probability density function.

[0133] Optionally, the specific implementation process of S5 includes:

[0134] According to the independent distribution update and conditional distribution update of implicit performance indicators and degradation model parameters, the inverse Gaussian distribution algorithm and the full probability formula are used for calculation to obtain the calculation result of the probability density function; among them, the calculation result of the probability density function is the remaining useful life prediction result.

[0135] In a feasible implementation, the process of calculating the probability density function of the remaining useful life according to the full probability formula can be expressed by the following formula (29):

[0136]

[0137] Among them, l k Indicates the remaining service life; k For λ k and θ k The conditional distribution of satisfies the inverse Gaussian distribution operation and can be expressed by the following formula (30):

[0138]

[0139] Among them, ω represents the threshold of RUL; σ 2 represents the variance of the diffusion parameter of the Wiener process; Formula (29) can be rewritten as the following formula (31):

[0140]

[0141] The auxiliary vector can be expressed by the following formula (32):

[0142]

[0143] Among them, Z1 is the first auxiliary matrix of PDF; Z2 is the second auxiliary matrix of PDF; Z3 is the third auxiliary matrix of PDF; Z4 is the fourth auxiliary matrix of PDF; Z5 is the fifth auxiliary matrix of PDF.

[0144] The embodiment of the present invention first obtains nonlinear system control data, and constructs an implicit performance indicator that can characterize the implicit degradation characteristics of the execution structure based on the system control data and the system state model. For the implicit performance indicator, a degradation model is established using the Wiener random process; secondly, based on the degradation model, a prediction algorithm architecture that integrates the FNN algorithm and the adaptive Kalman filter is designed, and an error result is obtained by analyzing the estimation error of the unmodeled dynamic to the degradation state; based on the error result, the filter structure of the adaptive Kalman filter is improved to obtain the gain matrix of the improved adaptive Kalman filter and the updated adaptive Kalman filter structure; based on the improved adaptive Kalman filter The gain matrix of the FNN algorithm is used to design an online weight update mechanism for the FNN algorithm. The estimation of the unmodeled dynamic term is updated in real time by the updated weights of the FNN algorithm to obtain the updated unmodeled dynamic term. According to the updated adaptive Kalman filter structure, the improved gain matrix and the updated unmodeled dynamic term, the covariance distribution of the implicit performance indicator and the degradation model parameters is updated to obtain the independent distribution update and conditional distribution update of the implicit performance indicator and the degradation model parameters. Finally, based on the independent distribution update and conditional distribution update of the implicit performance indicator and the degradation model parameters, the full probability formula is used for calculation to obtain the remaining useful life prediction result expressed as a probability density function.

[0145] An embodiment of the present invention introduces a feedforward neural network into a prediction framework based on an adaptive Kalman filter, establishes an online update mechanism for the neural network weights, and proves the convergence of the prediction error by establishing a Lyapunov function, thereby realizing compensation for the unmodeled dynamic term. Due to the introduction of the unmodeled dynamic compensation term, the gain term of the adaptive Kalman filter is improved, so that the degradation index and degradation model parameters are more accurately estimated. The distribution of the adaptive Kalman filter using the improved gain matrix is ​​re-derived, and the probability density function of the remaining useful life is also synchronously updated.

[0146] By integrating a neural network with nonlinear fitting and a Kalman filter robust to external uncertainties, the present invention accurately describes the degradation path of actuators in closed-loop control systems while simultaneously preventing the influence of unmodeled dynamic terms in the degradation model on the remaining useful life prediction results. This invention improves the accuracy of actuator performance prediction for latent degradation in nonlinear closed-loop control systems.

[0147] Figure 3 This is a block diagram of a device for predicting the performance of an actuator of a nonlinear closed-loop control system with implicit degradation according to an exemplary embodiment. The device is used in a method for predicting the performance of an actuator of a nonlinear closed-loop control system with implicit degradation. Figure 3The device includes an acquisition and construction unit 310, a design unit 320, a first updating unit 330, a second updating unit 340, and a calculation unit 350.

[0148] An acquisition and construction unit 310 is configured to acquire nonlinear system control data, construct an implicit performance indicator capable of characterizing implicit degradation characteristics of the execution structure based on the system control data and a system state model, and establish a degradation model using a Wiener random process for the implicit performance indicator;

[0149] The design unit 320 is configured to design a prediction algorithm architecture that integrates the FNN algorithm and the adaptive Kalman filter based on the degradation model, obtain an error result by analyzing the estimation error of the degradation state of the unmodeled dynamics, and improve the filter structure of the adaptive Kalman filter based on the error result to obtain an improved gain matrix of the adaptive Kalman filter and an updated adaptive Kalman filter structure;

[0150] A first updating unit 330 is configured to design an online weight update mechanism for the FNN algorithm based on the gain matrix of the improved adaptive Kalman filter, and to update the estimate of the unmodeled dynamic term in real time using the updated weights of the FNN algorithm to obtain an updated unmodeled dynamic term;

[0151] A first updating unit 340 is configured to update the covariance distribution of the implicit performance indicator and the degradation model parameters according to the updated adaptive Kalman filter structure, the improved gain matrix, and the updated unmodeled dynamic term, to obtain independent distribution updates and conditional distribution updates of the implicit performance indicator and the degradation model parameters;

[0152] The calculation unit 350 is used to calculate based on the independent distribution update and conditional distribution update of the implicit performance index and the degradation model parameters using the full probability formula to obtain the remaining service life prediction result represented by the probability density function.

[0153] Optionally, the acquiring and constructing unit 310 is configured to:

[0154] Obtain state data, control output data, and system output data of nonlinear systems;

[0155] The classical nonlinear Kalman filter is used to linearize the state model of the nonlinear control system and obtain a priori system model with implicit degradation performance indicators of the execution structure.

[0156] According to the equation of implicit performance index with implicit degradation characteristics of actuator structure, a degradation model of implicit degradation actuator is established by using Wiener random process.

[0157] Optionally, the design unit 320 is configured to:

[0158] Adaptive Kalman filter is used to evaluate degradation status;

[0159] According to the degradation assessment state and the degradation model, a fusion FNN adaptive Kalman filter algorithm is designed to obtain the error result by analyzing the estimation error of the degradation state of the unmodeled dynamics;

[0160] According to the error results, an auxiliary matrix is ​​established to improve the structure of the adaptive Kalman filter, the gain matrix of the improved adaptive Kalman filter is obtained, and the structure of the adaptive Kalman filter is updated.

[0161] Optionally, after the estimation of the unmodeled dynamic term is updated in real time by using the updated weights of the FNN algorithm in S3, the method further includes:

[0162] The unmodeled dynamic terms are updated by updating the FNN weights, and the Lyapunov function is designed to verify the stability and convergence of the filter.

[0163] Optionally, the calculation unit 350 is configured to:

[0164] According to the independent distribution update and conditional distribution update of implicit performance indicators and degradation model parameters, the inverse Gaussian distribution algorithm and the full probability formula are used for calculation to obtain the calculation result of the probability density function; among them, the calculation result of the probability density function is the remaining useful life prediction result.

[0165] The embodiment of the present invention first obtains nonlinear system control data, and constructs an implicit performance indicator that can characterize the implicit degradation characteristics of the execution structure based on the system control data and the system state model. For the implicit performance indicator, a degradation model is established using the Wiener random process; secondly, based on the degradation model, a prediction algorithm architecture that integrates the FNN algorithm and the adaptive Kalman filter is designed, and an error result is obtained by analyzing the estimation error of the unmodeled dynamic to the degradation state; based on the error result, the filter structure of the adaptive Kalman filter is improved to obtain the gain matrix of the improved adaptive Kalman filter and the updated adaptive Kalman filter structure; based on the improved adaptive Kalman filter The gain matrix of the FNN algorithm is used to design an online weight update mechanism for the FNN algorithm. The estimation of the unmodeled dynamic term is updated in real time by the updated weights of the FNN algorithm to obtain the updated unmodeled dynamic term. According to the updated adaptive Kalman filter structure, the improved gain matrix and the updated unmodeled dynamic term, the covariance distribution of the implicit performance indicator and the degradation model parameters is updated to obtain the independent distribution update and conditional distribution update of the implicit performance indicator and the degradation model parameters. Finally, based on the independent distribution update and conditional distribution update of the implicit performance indicator and the degradation model parameters, the full probability formula is used for calculation to obtain the remaining useful life prediction result expressed as a probability density function.

[0166] An embodiment of the present invention introduces a feedforward neural network into a prediction framework based on an adaptive Kalman filter, establishes an online update mechanism for the neural network weights, and proves the convergence of the prediction error by establishing a Lyapunov function, thereby realizing compensation for the unmodeled dynamic term. Due to the introduction of the unmodeled dynamic compensation term, the gain term of the adaptive Kalman filter is improved, so that the degradation index and degradation model parameters are more accurately estimated. The distribution of the adaptive Kalman filter using the improved gain matrix is ​​re-derived, and the probability density function of the remaining useful life is also synchronously updated.

[0167] By integrating a neural network with nonlinear fitting and a Kalman filter robust to external uncertainties, the present invention accurately describes the degradation path of actuators in closed-loop control systems while simultaneously preventing the influence of unmodeled dynamic terms in the degradation model on the remaining useful life prediction results. This invention improves the accuracy of actuator performance prediction for latent degradation in nonlinear closed-loop control systems.

[0168] Figure 4 Schematic diagram of the structure of a device for predicting the performance of an actuator of a nonlinear closed-loop control system with implicit degradation provided by an embodiment of the present invention. Figure 4 As shown, the performance prediction device of the actuator of the nonlinear closed-loop control system with implicit degradation may include the above Figure 3The device for predicting the performance of an actuator of a nonlinear closed-loop control system with implicit degradation is shown. Optionally, the device for predicting the performance of an actuator of a nonlinear closed-loop control system with implicit degradation 410 may include a first processor 2001 .

[0169] Optionally, the device 410 for predicting performance of an actuator with implicit degradation in a nonlinear closed-loop control system may further include a memory 2002 and a transceiver 2003 .

[0170] The first processor 2001, the memory 2002 and the transceiver 2003 may be connected via a communication bus.

[0171] The following combination Figure 4 The components of the actuator performance prediction device 410 for latent degradation of a nonlinear closed-loop control system are described in detail:

[0172] The first processor 2001 is the control center of the actuator performance prediction device 410 for implicit degradation of a nonlinear closed-loop control system, and can be a single processor or a collective term for multiple processing elements. For example, the first processor 2001 can be one or more central processing units (CPUs), or an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement an embodiment of the present invention, such as one or more digital signal processors (DSPs) or one or more field programmable gate arrays (FPGAs).

[0173] Optionally, the first processor 2001 may execute various functions of the actuator performance prediction device 410 for implicit degradation of a nonlinear closed-loop control system by running or executing a software program stored in the memory 2002 and calling data stored in the memory 2002 .

[0174] In a specific implementation, as an embodiment, the first processor 2001 may include one or more CPUs, such as Figure 4 CPU0 and CPU1 are shown in FIG.

[0175] In a specific implementation, as an embodiment, the actuator performance prediction device 410 of the nonlinear closed-loop control system with implicit degradation may also include multiple processors, such as Figure 41 and 2. The first processor 2001 and the second processor 2004 are shown in FIG. Each of these processors can be a single-core processor (single-CPU) or a multi-core processor (multi-CPU). A processor herein can refer to one or more devices, circuits, and / or processing cores for processing data (e.g., computer program instructions).

[0176] The memory 2002 is used to store the software program for executing the solution of the present invention, and is controlled by the first processor 2001 for execution. The specific implementation method can refer to the above method embodiment and will not be repeated here.

[0177] Alternatively, the memory 2002 may be a read-only memory (ROM) or other type of static storage device capable of storing static information and instructions, a random access memory (RAM) or other type of dynamic storage device capable of storing information and instructions, or an electrically erasable programmable read-only memory (EEPROM), a compact disc read-only memory (CD-ROM) or other optical disc storage, an optical disc storage (including a compact disc, laser disc, optical disc, digital versatile disc, Blu-ray disc, etc.), a magnetic disk storage medium or other magnetic storage device, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and capable of being accessed by a computer, but not limited thereto. The memory 2002 may be integrated with the first processor 2001 or may exist independently and be accessed through the interface circuit ( Figure 4 (not shown) is coupled to the first processor 2001, which is not specifically limited in this embodiment of the present invention.

[0178] The transceiver 2003 is used to communicate with a network device or a terminal device.

[0179] Optionally, the transceiver 2003 may include a receiver and a transmitter ( Figure 4 (not shown separately in the figure). The receiver is used to implement a receiving function, and the transmitter is used to implement a sending function.

[0180] Optionally, the transceiver 2003 may be integrated with the first processor 2001 or may exist independently and be used to predict the performance of the actuator of the nonlinear closed-loop control system with implicit degradation through the interface circuit ( Figure 4(not shown) is coupled to the first processor 2001, which is not specifically limited in this embodiment of the present invention.

[0181] It should be noted that Figure 4 The structure of the actuator performance prediction device 410 for implicit degradation of the nonlinear closed-loop control system shown in the figure does not constitute a limitation on the router. The actual knowledge structure recognition device may include more or fewer components than shown in the figure, or combine certain components, or arrange the components differently.

[0182] In addition, the technical effects of the performance prediction device 410 for an actuator of a nonlinear closed-loop control system with implicit degradation can refer to the technical effects of the performance prediction method for an actuator of a nonlinear closed-loop control system with implicit degradation described in the above method embodiment, and will not be repeated here.

[0183] It should be understood that the first processor 2001 in the embodiment of the present invention may be a central processing unit (CPU), and the processor may also be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor, etc.

[0184] It should also be understood that the memory in the embodiments of the present invention may be a volatile memory or a non-volatile memory, or may include both volatile and non-volatile memories. Among them, the non-volatile memory may be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), or a flash memory. The volatile memory may be a random access memory (RAM), which is used as an external cache. By way of example and not limitation, many forms of random access memory (RAM) are available, such as static RAM (SRAM), dynamic random access memory (DRAM), synchronous DRAM (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link DRAM (SLDRAM), and direct rambus RAM (DR RAM).

[0185] The above embodiments can be implemented in whole or in part through software, hardware (such as circuits), firmware, or any other combination. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer program are loaded or executed on a computer, the processes or functions described in accordance with the embodiments of the present invention are generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via a wired method (such as infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that contains a collection of one or more available media. The available medium can be a magnetic medium (such as a floppy disk, hard disk, or magnetic tape), an optical medium (such as a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state drive.

[0186] It should be understood that the term "and / or" as used herein simply describes a relationship between associated objects, indicating that three possible relationships exist. For example, "A and / or B" can represent: A alone, A and B together, or B alone. A and B can be singular or plural. Furthermore, the character " / " as used herein generally indicates an "or" relationship between the associated objects, but it may also indicate an "and / or" relationship. For specific understanding, please refer to the context.

[0187] In this disclosure, "at least one" means one or more, and "plurality" means two or more. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single or plural items. For example, "at least one of a, b, or c" can mean: a, b, c, ab, ac, bc, or abc, where a, b, and c can be single or plural.

[0188] It should be understood that in various embodiments of the present invention, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0189] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present invention.

[0190] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described equipment, devices and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0191] In the several embodiments provided by the present invention, it should be understood that the disclosed devices, apparatuses and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another device, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interface, indirect coupling or communication connection of the device or unit, which can be electrical, mechanical or other forms.

[0192] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0193] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0194] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0195] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in the present invention should be included in the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be based on the scope of protection of the claims.

Claims

1. A method for predicting the performance of an actuator with implicit degradation in a nonlinear closed-loop control system, characterized in that: The method comprises: S1. Acquire nonlinear system control data, construct an implicit performance indicator capable of characterizing implicit degradation characteristics of an execution structure based on the system control data and a system state model, and establish a degradation model using a Wiener random process for the implicit performance indicator; S2. Based on the degradation model, design a prediction algorithm architecture that integrates the FNN algorithm and the adaptive Kalman filter, and obtain an error result by analyzing the estimation error of the degradation state of the unmodeled dynamics; based on the error result, improve the filter structure of the adaptive Kalman filter to obtain an improved gain matrix of the adaptive Kalman filter and an updated adaptive Kalman filter structure; S3. Designing an online weight update mechanism for the FNN algorithm based on the gain matrix of the improved adaptive Kalman filter, and updating the estimate of the unmodeled dynamic term in real time by using the updated weights of the FNN algorithm to obtain an updated unmodeled dynamic term; S4. Update the covariance distribution of the implicit performance indicator and the degradation model parameters according to the updated adaptive Kalman filter structure, the improved gain matrix, and the updated unmodeled dynamic term to obtain independent distribution updates and conditional distribution updates of the implicit performance indicator and the degradation model parameters; S5. Based on the independent distribution update and conditional distribution update of the implicit performance indicators and degradation model parameters, the full probability formula is used for calculation to obtain the remaining service life prediction result represented by the probability density function.

2. The method for predicting the performance of an actuator with latent degradation in a nonlinear closed-loop control system according to claim 1, characterized in that: The step S1 acquires nonlinear system control data, constructs an implicit performance indicator capable of characterizing implicit degradation characteristics of the execution structure based on the system control data and the system state model, and establishes a degradation model using the Wiener random process for the implicit performance indicator, including: S11, obtaining state data, control output data, and system output data of the nonlinear system; S12. Using a classical nonlinear Kalman filter, linearize the state model of the nonlinear control system to obtain a linearized system state model with implicit degradation performance indicators of the actuator; S13. Based on the linearized system state model of the implicit performance index with the implicit degradation characteristics of the execution structure, a degradation model of the implicitly degraded actuator is established using the Wiener random process.

3. The method for predicting the performance of an actuator with latent degradation in a nonlinear closed-loop control system according to claim 1, characterized in that: According to the degradation model, the S2 designs a prediction algorithm architecture integrating the FNN algorithm and the adaptive Kalman filter, and obtains an error result by analyzing the estimation error of the degradation state by the unmodeled dynamics; According to the error results, the adaptive Kalman filter structure is improved to obtain an improved adaptive Kalman filter gain matrix and an updated adaptive Kalman filter structure, including: S21, using adaptive Kalman filter to evaluate degradation state; S22. Designing a fusion FNN adaptive Kalman filter algorithm based on the degradation assessment state and the degradation model, and obtaining an error result by analyzing the estimation error of the degradation state by the unmodeled dynamics; S23. According to the error result, an auxiliary matrix is ​​established to improve the structure of the adaptive Kalman filter, a gain matrix of the improved adaptive Kalman filter is obtained, and the structure of the adaptive Kalman filter is updated.

4. The method for predicting the performance of an actuator with latent degradation in a nonlinear closed-loop control system according to claim 1, characterized in that: After the estimation of the unmodeled dynamic term is updated in real time by using the updated weights of the FNN algorithm in S3, the following steps are further included: The unmodeled dynamic terms are updated by updating the FNN weights, and the Lyapunov function is designed to verify the stability and convergence of the filter.

5. The method for predicting the performance of an actuator with latent degradation in a nonlinear closed-loop control system according to claim 1, characterized in that: The S5 is based on the independent distribution update and conditional distribution update of the implicit performance index and the degradation model parameters, and uses the full probability formula to calculate and obtain the remaining service life prediction result represented by the probability density function, including: According to the independent distribution update and conditional distribution update of implicit performance indicators and degradation model parameters, the inverse Gaussian distribution algorithm and the full probability formula are used for calculation to obtain the calculation result of the probability density function; among them, the calculation result of the probability density function is the remaining useful life prediction result.

6. A device for predicting the performance of an actuator of a nonlinear closed-loop control system with latent degradation, wherein the device is used to implement the method for predicting the performance of an actuator of a nonlinear closed-loop control system with latent degradation according to any one of claims 1 to 5, characterized in that: The device comprises: an acquisition and construction unit for acquiring nonlinear system control data, constructing an implicit performance indicator capable of characterizing implicit degradation characteristics of the execution structure based on the system control data and a system state model, and establishing a degradation model for the implicit performance indicator using a Wiener random process; A design unit is used to design a prediction algorithm architecture that integrates the FNN algorithm and the adaptive Kalman filter according to the degradation model, obtain an error result by analyzing the estimation error of the degradation state of the unmodeled dynamics; and improve the filter structure of the adaptive Kalman filter according to the error result to obtain an improved gain matrix of the adaptive Kalman filter and an updated adaptive Kalman filter structure; A first updating unit is configured to design an online weight updating mechanism of the FNN algorithm based on the gain matrix of the improved adaptive Kalman filter, and to update the estimate of the unmodeled dynamic term in real time by using the updated weight of the FNN algorithm to obtain an updated unmodeled dynamic term; a second updating unit, which updates the covariance distribution of the implicit performance indicator and the degradation model parameters according to the updated adaptive Kalman filter structure, the improved gain matrix, and the updated unmodeled dynamic term, to obtain independent distribution updates and conditional distribution updates of the implicit performance indicator and the degradation model parameters; The calculation unit uses the full probability formula to perform calculations based on the independent distribution update and conditional distribution update of the implicit performance indicators and degradation model parameters to obtain the remaining service life prediction result expressed as a probability density function.

7. The performance prediction device for actuators with latent degradation in nonlinear closed-loop control systems according to claim 6, characterized in that: The acquisition and construction unit is used to: Obtain state data, control output data, and system output data of nonlinear systems; The classical nonlinear Kalman filter is used to linearize the state model of the nonlinear control system and obtain a priori system model with implicit degradation performance indicators of the execution structure. According to the equation of implicit performance index with implicit degradation characteristics of actuator structure, a degradation model of implicit degradation actuator is established by using Wiener random process.

8. The performance prediction device for actuators with latent degradation in nonlinear closed-loop control systems according to claim 6, characterized in that: The design unit is used to: Adaptive Kalman filter is used to evaluate degradation status; According to the degradation assessment state and the degradation model, a fusion FNN adaptive Kalman filter algorithm is designed to obtain the error result by analyzing the estimation error of the degradation state of the unmodeled dynamics; According to the error results, an auxiliary matrix is ​​established to improve the structure of the adaptive Kalman filter, the gain matrix of the improved adaptive Kalman filter is obtained, and the structure of the adaptive Kalman filter is updated.

9. A device for predicting the performance of an actuator with latent degradation in a nonlinear closed-loop control system, characterized in that: The performance prediction device for the actuator of the nonlinear closed-loop control system with implicit degradation includes: processor; A memory having computer-readable instructions stored thereon, wherein when the computer-readable instructions are executed by the processor, the method according to any one of claims 1 to 5 is implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores program code, which can be called by a processor to execute the method according to any one of claims 1 to 5.

Images