A Method for Assessing Uncertainty of Performance Degradation in Electromechanical Systems Based on Dynamic Comprehensive Indicators

Abstract

This invention relates to the field of intelligent operation and maintenance technology for mechanical equipment, specifically to a method for assessing the uncertainty of performance degradation in electromechanical systems based on dynamic comprehensive indicators. The method includes: acquiring real-time vibration signals of the electromechanical system and extracting features to construct real-time similarity feature indicators; constructing a model selection library containing multiple performance assessment fitting models; fitting each model based on the real-time similarity feature indicators; scoring the fitting effect of each model based on the PE criterion and selecting the optimal model; predicting the characteristic state of the electromechanical system based on the optimal model selected at the current moment, introducing process noise and performance monitoring noise, and calculating the uncertainty assessment results of the electromechanical system at different times; and predicting the performance of the electromechanical system based on the uncertainty assessment results. This invention can dynamically iterate models based on actual operating conditions and perform uncertainty assessment of the performance of electromechanical systems, improving the accuracy and reliability of performance assessment.

Description

Technical Field

[0001] This invention relates to the field of intelligent operation and maintenance technology for mechanical equipment, and more specifically to a method for assessing the uncertainty of performance degradation of electromechanical systems based on dynamic comprehensive indicators. Background Technology

[0002] In recent years, electromechanical systems have been widely used in modern mechanical equipment due to their advantages of high load-bearing capacity and efficiency. The performance of an electromechanical system directly affects the reliability and efficiency of the entire equipment. However, electromechanical systems often operate in harsh environments such as high temperatures and high loads. These harsh environments can easily lead to component failures, resulting in system failure. System failure can cause delays or even serious injuries or fatalities. Repairing an electromechanical system after failure significantly increases maintenance costs. Driven by this, performance evaluation has received increasing attention. Performance evaluation of electromechanical systems not only reduces the accident rate but also significantly lowers maintenance costs. Therefore, performance evaluation of electromechanical systems is of great importance.

[0003] In existing electromechanical system performance evaluation methods, model-based methods have received increasing attention. However, due to the complex and variable operating environment of electromechanical systems, it is difficult to accurately evaluate the performance of electromechanical systems using only a single model. To address this issue, more and more research focuses on adaptive evaluation methods, which are mainly divided into multi-stage evaluation methods and dynamic evaluation methods. Multi-stage evaluation methods divide the performance degradation process of electromechanical systems into three stages: operation, slow degradation, and accelerated degradation. However, in actual operation, not all electromechanical system degradation processes can be divided into three stages, such as abrupt and self-healing degradation processes. When using multi-stage methods to evaluate these two types of degradation processes, the evaluation effect will be greatly reduced. At the same time, the evaluation models used by existing dynamic evaluation methods are relatively simple, only selecting the evaluation model at the initial stage of operation. The selection criteria only consider the influence of fitting error, which is still difficult to accurately fit the degradation process of electromechanical systems. Furthermore, it is difficult to assess the uncertainty of the degradation process during the evaluation process. These shortcomings lead to poor performance evaluation results, which in turn affect subsequent maintenance decisions. Therefore, how to improve the fitting effect of the degradation process and assess the uncertainty of the degradation process has become a technical problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0004] In view of this, the present invention provides a method for evaluating the uncertainty of performance degradation of electromechanical systems based on dynamic comprehensive indicators, which solves the problems of poor fitting effect and low evaluation accuracy of existing methods in the degradation process.

[0005] To achieve the above objectives, the present invention adopts the following technical solution:

[0006] A method for assessing the uncertainty of performance degradation in electromechanical systems based on dynamic comprehensive indices includes the following steps:

[0007] Acquire real-time vibration signals of the electromechanical system and extract features;

[0008] Based on the feature distribution extracted in real time and the feature distribution extracted under normal conditions from the electromechanical system, a real-time similarity feature index is constructed.

[0009] Construct a model selection library containing multiple performance evaluation fitting models;

[0010] Based on real-time similarity feature metrics, a known dataset is constructed, and each model in the model selection library is fitted based on the known dataset;

[0011] The fitting effect of each model is scored based on the PE criterion, and the model with the lowest score is selected as the optimal model at the current time.

[0012] Based on the optimal model selected at the current moment, the instantaneous characteristic states of the electromechanical system at different future moments are predicted;

[0013] Process noise and performance monitoring noise are introduced into the prediction results, along with the uncertain evaluation results of the computer electrical system at different times.

[0014] Performance prediction of electromechanical systems is based on uncertainty assessment results.

[0015] Furthermore, the real-time vibration signal of the electromechanical system is represented as: ,in, For runtime, for Vibration signal values ​​monitored at all times For signal sampling points, ;

[0016] The formula for feature extraction from real-time vibration signals of electromechanical systems is as follows:

[0017]

[0018] in, Features extracted.

[0019] Furthermore, the formula for calculating the feature similarity index is as follows:

[0020]

[0021] in, Features extracted under normal conditions The distribution, Features extracted from vibration signal data at several consecutive time points updated in real time Distribution, through this calculation formula, features Normalization reflects the similarity between real-time updated features and features under normal operating conditions. The larger the value, the lower the similarity between the two, and vice versa.

[0022] Furthermore, the model selection library includes basic models and complex models. The basic model is represented as follows:

[0023]

[0024] Complex models are represented as:

[0025]

[0026] in, The feature index values ​​that need to be obtained by model fitting are... The characteristic index is calculated at the initial time. The parameters to be fitted, For runtime, This represents the standard Brownian motion process.

[0027] Furthermore, the process of fitting each model in the model selection library includes:

[0028] Based on real-time monitoring and calculation of similarity feature indicators Construct a known dataset ,in For runtime, for Feature values ​​extracted at runtime For signal sampling points, ;

[0029] Based on known datasets The simple and complex models were fitted respectively.

[0030] Furthermore, the formula for scoring the fitting effect of each model based on the PE criterion is as follows:

[0031]

[0032] in, The number of unknown parameters in the performance evaluation fitting model; The number of known data points used in the model fitting; In time The actual value obtained from it, To use the model in time The fitted value obtained at that point, This represents the average trend value of the characteristic indicator. Weights to account for the model degradation trend during dynamic iteration; Used to balance situations where different models contain different numbers of unknown parameters;

[0033] Whenever the vibration signal data of the electromechanical system is updated, the fitting models for different performance evaluations are calculated. Value, selection The model with the lowest score is used to evaluate the performance of the electromechanical system at that moment.

[0034] Furthermore, process noise and performance monitoring noise are introduced into the prediction results, along with uncertain evaluation results of the computer electrical system at different times, including:

[0035] Introduce process noise into the prediction results at a certain moment to construct a characteristic change equation;

[0036] Introduce performance monitoring noise and construct a performance monitoring function;

[0037] By introducing the covariance matrix, the optimal characteristic approximation value of the electromechanical system at the next time moment is estimated using the known optimal characteristic approximation value of the electromechanical system at the previous time moment, thus obtaining the uncertainty assessment results of the electromechanical system at different future times.

[0038] Furthermore, the expression for the characteristic change equation is:

[0039]

[0040] in, Indicates in The instantaneous characteristic state of the electromechanical system at any given moment. Is The instantaneous characteristic state of the electromechanical system at any given moment. Indicates from state to state The transfer matrix, For process noise, For the control matrix, For electromechanical systems Gain variable at time step;

[0041] The expression for the performance monitoring function is:

[0042]

[0043] in, In order to be in The characteristic values ​​of the time-sensitive electromechanical system In order to be in Time-time feature change matrix To monitor noise for its characteristics.

[0044] Furthermore, the formula for calculating the uncertainty assessment results of the electromechanical system at different future times is as follows:

[0045]

[0046] in, In order to be in The optimal characteristic approximation of the time-sensitive electromechanical system. In order to be in The prior error covariance matrix at time step, for The optimal characteristic approximation of the time-sensitive electromechanical system. Indicates from state to state The transfer matrix; In order to be in The feature change matrix at time step, In order to be in The feature change matrix at time step, In order to be in Transpose of the feature change matrix at time step 1 To monitor noise for characteristics The covariance matrix, In order to be in Characteristic values ​​of the electromechanical system at any given time;

[0047] In order to be in The covariance matrix of the prior error of the electromechanical system at any given time is calculated using the following formula:

[0048]

[0049] in, For the transfer matrix The transpose of the matrix, for The covariance matrix of the electromechanical system at any given time. Indicates in The variables influenced by external factors during the operation of the electromechanical system.

[0050] Furthermore, the calculation formula for predicting the performance of the electromechanical system based on the uncertainty assessment results is as follows:

[0051]

[0052] in, In order to be in The estimated remaining runtime at each moment. Eigenvalues The lower limit, Is The eigenvalues ​​obtained after uncertainty assessment For the set failure characteristic values ​​of the electromechanical system, Indicated as from From the start of time until the evaluation of the eigenvalue Reaching the failure characteristic value The time interval.

[0053] As can be seen from the above technical solution, compared with the prior art, the present invention has the following beneficial effects:

[0054] 1. This invention can dynamically iterate the model as the electromechanical system continues to operate to adapt to different types of degradation processes. Furthermore, it combines the linear and nonlinear models commonly used in existing performance evaluation methods to fit the degradation process, thus adapting to various types of degradation processes.

[0055] 2. This invention uses the comprehensive index PE criterion to iteratively select models, while taking into account both fitting error and model degradation trend evaluation effects. This better captures the degradation process of electromechanical systems, reduces the impact of data mutations on performance evaluation, and improves the evaluation effect.

[0056] 3. This invention takes into account uncertainties such as noise during operation and performs uncertainty assessment on the degradation process, thereby improving the accuracy and reliability of performance assessment. Attached Figure Description

[0057] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0058] Figure 1 A flowchart of the method for assessing the uncertainty of performance degradation of electromechanical systems based on dynamic comprehensive indices provided by the present invention;

[0059] Figure 2 A framework diagram of the electromechanical system performance degradation uncertainty assessment method based on dynamic comprehensive index provided by the present invention;

[0060] Figure 3 This is a schematic diagram comparing the performance of the method of the present invention with existing prediction methods. Detailed Implementation

[0061] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0062] like Figures 1-2 As shown in the figure, this invention discloses a method for assessing the uncertainty of performance degradation of electromechanical systems based on dynamic comprehensive indices, including the following steps:

[0063] S1. Acquire the real-time vibration signal of the electromechanical system and extract its features;

[0064] S2. Based on the feature distribution extracted in real time from the electromechanical system and the feature distribution extracted under normal conditions, a real-time similarity feature index is constructed.

[0065] S3. Construct a model selection library containing multiple performance evaluation fitting models;

[0066] S4. Based on real-time similarity feature indicators, construct a known dataset and fit each model in the model selection library based on the known dataset;

[0067] S5. Based on the PE criterion, score the fitting effect of each model and select the model with the lowest score as the optimal model at the current moment.

[0068] S6. Based on the optimal model selected at the current moment, predict the instantaneous characteristic state of the electromechanical system at different future moments; introduce process noise and performance monitoring noise into the prediction results, calculate the uncertain evaluation results of the electromechanical system at different moments; and perform performance prediction of the electromechanical system based on the uncertain evaluation results.

[0069] The following provides further explanation of each of the above steps.

[0070] S1. Obtain the real-time vibration signal of the electromechanical system. The real-time vibration signal of the electromechanical system is represented as follows: ,in, For runtime, for Vibration signal values ​​monitored at all times For signal sampling points,

[0071] Feature extraction is performed on the real-time vibration signal of the electromechanical system. The extraction formula is as follows:

[0072]

[0073] in, Features extracted.

[0074] S2. Construction of similarity feature indicators.

[0075] After feature extraction from the vibration signal, a feature similarity index is further constructed, and the calculation formula is as follows:

[0076]

[0077] in, Features extracted under normal conditions Based on the monitored data, the electromechanical system was in normal operation during the initial operation phase, and faults gradually evolved thereafter. The "normal state" referred to here refers to the features extracted during the initial operation. The distribution, for example, taking the distribution of the first 10 data points as the distribution extracted under normal conditions is... . Features extracted from vibration signal data at several consecutive time points updated in real time The distribution can be the distribution of features extracted at the current time step and features extracted at several previous time steps. The specific number of features can be set by the user. This calculation formula is used to determine the feature distribution. Normalization reflects the similarity between real-time updated features and features under normal operating conditions. The distribution of the monitored data can be measured. Distribution of data in normal state Similarity, when hour, At this point, the two are completely similar. The larger the value, the lower the similarity between the two, and vice versa. By constructing similarity feature indicators, we can measure whether an electromechanical system is in an abnormal state, that is, whether it has begun to degrade.

[0078] S3. Construct a model selection library containing various performance evaluation fitting models. The model selection library contains two main categories: basic models and complex models. The basic models are represented as follows:

[0079]

[0080] Complex models are represented as:

[0081]

[0082] in, The feature index values ​​that need to be obtained by model fitting are... The characteristic index is calculated at the initial time. For the parameters that need to be fitted, For runtime, This represents the standard Brownian motion process.

[0083] S4. Fit each model in the model selection library. The specific fitting process includes:

[0084] Based on real-time monitoring and calculation of similarity feature indicators Construct a known dataset ,in For runtime, for Feature values ​​extracted at runtime For signal sampling points, ;

[0085] Based on known datasets The simple and complex models were fitted respectively.

[0086] Taking the double exponential model as an example, the formula for calculating the fitted curve is: .

[0087] in, Calculated for using the fitted model The characteristic index value at time t is used to establish an error function, and the calculation formula is as follows:

[0088]

[0089] in, The parameters obtained from the fitting, In search of A function with minimum parameter. This is achieved by utilizing a known dataset. The unknown parameters are estimated, and finally a well-fitted model is obtained.

[0090] S5. With real-time monitoring of electromechanical systems and dynamic model iteration using known datasets, and addressing the issue that existing methods typically only consider model fitting error, a comprehensive index, PE criterion, is constructed for scoring. The PE criterion considers both fitting error and model degradation trend effects. The PE criterion is used to score the model fitting effect, and the formula for calculating the PE value is as follows:

[0091]

[0092] in, The number of unknown parameters in the performance evaluation fitting model; The number of known data points used in the model fitting; In time The actual value obtained from it, To use the model in time The fitted value obtained at that point, This represents the average trend value of the characteristic indicator. To account for the model degradation trend effect during dynamic iteration, the weight is determined based on the degree to which the degradation trend effect needs to be considered in the actual situation. If the degradation trend effect is of greater concern, It can be set to greater than 0.5 if the fitting error and degradation trend are considered to be of equal importance. . Used to balance situations where different models contain different numbers of unknown parameters;

[0093] Whenever the vibration signal data of the electromechanical system is updated, the fitting models for different performance evaluations are calculated. Value, selection The model with the lowest score is taken as the optimal model at the current moment, and the performance of the electromechanical system at that moment is evaluated to achieve the effect of dynamic model iterative update.

[0094] S6. After determining the optimal model, the performance of the electromechanical system is evaluated based on the data acquired at this time. Since the electromechanical system operates under various external environmental influences, its performance needs to be evaluated for uncertainty degradation, specifically including:

[0095] S61. Due to the influence of process noise in the actual monitoring process, process noise is introduced, and a characteristic change equation is constructed to correct the characteristic values ​​predicted by the optimal model. The expression of the characteristic change equation is as follows:

[0096]

[0097] in, Indicates in The instantaneous characteristic state of the electromechanical system at any given moment, i.e., the optimal model at... The feature values ​​predicted at each time step, Is The instantaneous characteristic state of the electromechanical system at any given moment. Indicates from state to state The transfer matrix, For process noise, For the control matrix, For electromechanical systems Gain variable at time step. Control matrix. This indicates the degree of influence of external factors on state transmission, which is known. The control matrix can be obtained by fitting this function. The value, if there are no external influences. If the value is 0, the above function can be simplified to: .

[0098] Introducing performance monitoring noise, a performance monitoring function is constructed, the expression of which is:

[0099]

[0100] in, In order to be in The characteristic values ​​of the time-sensitive electromechanical system In order to be in Time-time feature change matrix The noise is characterized by being monitored. The noise is calculated based on the fluctuations of the forward known data. The noise is obtained by calculating the noise component of the measured data, such as the covariance of the difference between the measured value and the true value. This can be used to reflect the uncertainty of the actual operation of the system.

[0101] The uncertainty in electromechanical system performance evaluation is accommodated by introducing process noise and performance monitoring noise.

[0102] S62. To find the optimal characteristic approximation value of the electromechanical system, a covariance matrix is ​​introduced. Using the known optimal characteristic approximation value of the electromechanical system at the previous time step, the optimal characteristic approximation value of the electromechanical system at the next time step is estimated, thus obtaining the uncertainty assessment results of the electromechanical system at different future time steps. The calculation formula for the uncertainty assessment results of the electromechanical system at different future time steps is:

[0103]

[0104] in, In order to be in The optimal characteristic approximation of the time-sensor electromechanical system, i.e. Uncertainty assessment results of the time-sensitive electromechanical system In order to be in The prior error covariance matrix at time step, for The optimal characteristic approximation of the time-sensitive electromechanical system. Indicates from state to state The transfer matrix; In order to be in The feature change matrix at time step, In order to be in The feature change matrix at time step, In order to be in Transpose of the feature change matrix at time step 1 For characteristic monitoring of noise The covariance matrix, In order to be in The characteristic values ​​of the electromechanical system at any given time.

[0105] In order to be in The covariance matrix of the prior error of the electromechanical system at any given time is calculated using the following formula:

[0106]

[0107] in, For the transfer matrix The transpose of the matrix, for The covariance matrix of the electromechanical system at any given time. Indicates in The variables influenced by external factors during the operation of the electromechanical system.

[0108] S63. Based on the uncertainty assessment results, the performance of the electromechanical system is predicted. The specific calculation formula is as follows:

[0109]

[0110] in, In order to be in The estimated remaining runtime at each moment. Eigenvalues The lower limit, Is The eigenvalues ​​obtained after uncertainty assessment For the set failure characteristic values ​​of the electromechanical system, Indicated as from From the start of time until the evaluation of the eigenvalue Reaching the failure characteristic value The time interval.

[0111] Next, taking the performance evaluation of the gearbox in the electromechanical system of a rail train as an example, the effectiveness of the method of the present invention was verified by collecting the operating data of the gearbox in the electromechanical system through an experimental platform.

[0112] The experimental platform is powered by a drive motor and includes two gearboxes for transmission. One gearbox serves as a test unit for gearbox degradation, with a module of 1.5 and a width of 15 mm. The other gearbox provides auxiliary testing; its gears have modules of 2 and 3 and a width of 30 mm to improve reliability. Both gearboxes are made of S45C material with a tooth surface hardness below 194 HB.

[0113] This experiment collected data from 11 channels representing five different signal types: vibration, rotational speed, torque, sound, and current, from the point of operation until failure. The sampling frequency was set at 12.8 kHz, the sampling interval was 2 minutes, and each sampling duration was 2.56 seconds. Two types of operating conditions were set up: constant operating conditions and time-varying operating conditions, as shown in Table 1.

[0114] Table 1 Operating Condition Settings

[0115]

[0116] Table 2 provides detailed information for each test gearbox, including its operating conditions, lifespan, and failure type.

[0117] Table 2. Overview of Gear Lifetime Degradation Data

[0118]

[0119] The experiment employed the method proposed in this invention to dynamically iterate the gear running data and assess uncertainty degradation. The proposed method was compared with three other assessment methods. Method 1 used an adaptive two-stage degradation framework based on Gaussian regression for prediction; Method 2 established two models, a linear regression model and a power exponential model, to describe the slow degradation process and the accelerated degradation process, respectively; Method 3 used the BIC criterion for model selection but did not perform dynamic model updates. The four methods were compared using different gearboxes, as shown in Table 3. Figure 3 The comparison of the root mean square error and cumulative relative accuracy of three types of gearbox operating data is presented.

[0120] Table 3 Comparison of Root Mean Square Error of Four Methods

[0121]

[0122] The experimental results show that the proposed method for assessing the uncertainty of electromechanical system performance degradation based on dynamic comprehensive indexes achieves good assessment results, and the assessment effect is significantly better than the other three existing methods.

[0123] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to the method section.

[0124] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for assessing the uncertainty of performance degradation in electromechanical systems based on dynamic comprehensive indices, characterized in that, Includes the following steps: Acquire real-time vibration signals of the electromechanical system and extract features; Based on the feature distribution extracted in real time from the electromechanical system and the feature distribution extracted under normal conditions, a real-time feature similarity index is constructed. Construct a model selection library containing multiple performance evaluation fitting models; Based on real-time feature similarity metrics, a known dataset is constructed, and models in the model selection library are fitted based on the known dataset; The fit of each model is scored based on the PE criterion, and the model with the lowest score is selected as the best model at the current time step. The formula for scoring the fit of each model based on the PE criterion is as follows: ； in, The number of unknown parameters in the performance evaluation fitting model; The number of known data points used in the model fitting; In time The actual value obtained from it To use the model in time The fitted value obtained at that point, This represents the average trend value of the characteristic indicator. Weights to account for the model degradation trend during dynamic iteration; Used to balance situations where different models contain different numbers of unknown parameters; Whenever the vibration signal data of the electromechanical system is updated, the fitting models for different performance evaluations are calculated. Value, selection The model with the lowest score is worth evaluating the performance of the electromechanical system at this moment; Based on the optimal model selected at the current moment, the instantaneous characteristic states of the electromechanical system at different future moments are predicted; Process noise and performance monitoring noise are introduced into the prediction results. The uncertainty assessment results of the computer electrical system at different times are calculated as follows: Introduce process noise into the prediction results at a certain moment to construct a characteristic change equation; Introduce performance monitoring noise and construct a performance monitoring function; By introducing the covariance matrix, the optimal characteristic approximation value of the electromechanical system at the next time moment is estimated by knowing the optimal characteristic approximation value of the electromechanical system at the previous time moment, and the uncertainty assessment results of the electromechanical system at different future times are obtained. Performance prediction of electromechanical systems is based on uncertainty assessment results.

2. The method for assessing the uncertainty of electromechanical system performance degradation based on dynamic comprehensive indicators according to claim 1, characterized in that, The real-time vibration signal of the electromechanical system is represented as ,in, For runtime, for Vibration signal values ​​monitored at all times For signal sampling points, ; The formula for feature extraction from real-time vibration signals of electromechanical systems is as follows: ； in, Features extracted.

3. The method for assessing the uncertainty of electromechanical system performance degradation based on dynamic comprehensive indicators according to claim 1, characterized in that, The formula for calculating the feature similarity index is: ； in, Features extracted under normal conditions The distribution, Features extracted from vibration signal data at several consecutive time points updated in real time Distribution, through this calculation formula, features Normalization reflects the similarity between real-time updated features and features under normal operating conditions. The larger the value, the lower the similarity between the two, and vice versa.

4. The method for assessing the uncertainty of electromechanical system performance degradation based on dynamic comprehensive indicators according to claim 1, characterized in that, The model selection library contains basic and complex models. The basic model is represented as follows: ； Complex models are represented as: ； in, The feature index values ​​that need to be obtained by model fitting are... The characteristic index is calculated at the initial time. The parameters to be fitted, For runtime, This represents the standard Brownian motion process.

5. The method for assessing the uncertainty of electromechanical system performance degradation based on dynamic comprehensive indicators according to claim 1, characterized in that, The process of fitting each model in the model selection library includes: Based on real-time monitoring and calculated feature similarity index Construct a known dataset ,in For runtime, for Feature values ​​extracted at runtime For signal sampling points, ; Based on known datasets The simple and complex models were fitted respectively.

6. The method for assessing the uncertainty of electromechanical system performance degradation based on dynamic comprehensive indicators according to claim 1, characterized in that, The expression for the characteristic change equation is: ； in, Indicates in The instantaneous characteristic state of the electromechanical system at any given moment. Is The instantaneous characteristic state of the electromechanical system at any given moment. Indicates from state to state The transfer matrix, For process noise, For the control matrix, For electromechanical systems Gain variable at time step; The expression for the performance monitoring function is: ； in, In order to be in The characteristic values ​​of the time-sensitive electromechanical system In order to be in Time-time feature change matrix Noise for performance monitoring.

7. The method for assessing the uncertainty of electromechanical system performance degradation based on dynamic comprehensive indicators according to claim 6, characterized in that, The formula for calculating the uncertainty assessment results of the electromechanical system at different future times is as follows: ； in, In order to be in The optimal characteristic approximation of the time-sensitive electromechanical system. for The optimal characteristic approximation of the time-sensitive electromechanical system. Indicates from state to state The transfer matrix; In order to be in The feature change matrix at time step, In order to be in The feature change matrix at time step, In order to be in Transpose of the feature change matrix at time step 1 For performance monitoring noise The covariance matrix, In order to be in Characteristic values ​​of the electromechanical system at any given time; In order to be in The covariance matrix of the prior error of the electromechanical system at any given time is calculated using the following formula: ； in, For the transfer matrix The transpose of the matrix, for The covariance matrix of the electromechanical system at any given time. Indicates in The variables influenced by external factors during the operation of the electromechanical system.

8. The method for assessing the uncertainty of electromechanical system performance degradation based on dynamic comprehensive indicators according to claim 7, characterized in that, The formula for predicting the performance of electromechanical systems based on uncertainty assessment results is as follows: ； in, In order to be in The estimated remaining runtime at each moment. for The lower limit, Is The eigenvalues ​​obtained after uncertainty assessment For the set failure characteristic values ​​of the electromechanical system, Indicated as from From the start of time until the evaluation of the eigenvalue Reaching the failure characteristic value The time interval.

Images