Equipment residual life prediction method and device considering coupling effect of degradation indexes

Abstract

The application discloses an equipment residual life prediction method and device considering coupling effects of degradation indexes, and relates to the technical field of equipment state monitoring and prediction. Monitoring signals are acquired, high-dimensional features are extracted from the monitoring signals to construct a high-dimensional feature set, and two high-dimensional features with the largest correlation are screened out as degradation indexes representing the degradation state of equipment. An adaptive signal conversion algorithm is used to convert the degradation indexes to a reference working condition to obtain working condition adaptive degradation indexes, thereby overcoming the problem of residual life prediction difficulty caused by working condition changes. A nonlinear binary Wiener process is constructed to describe the mutual dependence between the two working condition adaptive degradation indexes, and Monte Carlo simulation technology is used to simulate the coupling degradation indexes to obtain equipment residual life prediction. The influence of different degradation index coupling effects on the residual life is considered while the influence of working condition changes on the degradation indexes is adaptively eliminated, thereby realizing accurate equipment residual life.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of equipment state monitoring and prediction, and in particular to an equipment residual life prediction method and device considering coupling effects of degradation indicators. BACKGROUND

[0002] In the modern industrial field, the efficient and safe operation of various types of equipment is crucial to production activities. During the life cycle of the equipment, it is inevitable to be affected by various factors, among which the change of working conditions is a key factor. Frequent changes in working conditions can have a significant impact on the degradation data of the equipment, and thus affect the prediction of the residual life of the equipment. Traditional residual life prediction methods are often based on a single degradation indicator for modeling. However, this method is difficult to fully reflect the real degradation process of the equipment under complex working conditions. In addition, under the condition of time-varying working conditions, the influence of working condition changes on degradation indicators and the coupling effects between different degradation indicators all bring great challenges to the accurate prediction of residual life. Therefore, how to adaptively eliminate the influence of working condition changes on degradation indicators under the condition of time-varying working conditions, and fully consider the coupling effects between different degradation indicators, so as to realize the accurate prediction of the residual life of the equipment, is a technical problem to be solved at present.

[0003] In summary, the following problems still exist in the prior art:

[0004] 1. How to adaptively eliminate the influence of working condition changes on the amplitude of degradation data under the condition of frequent changes in working conditions during the life cycle of the equipment, in order to improve the accuracy of residual life prediction;

[0005] 2. How to fully consider the coupling effects between different degradation indicators in the modeling process, so as to more comprehensively reflect the degradation process of the equipment and realize accurate prediction of the residual life of the equipment. SUMMARY

[0006] Based on the problems raised in the above background technology, the purpose of the present application is to provide an equipment residual life prediction method and device considering coupling effects of degradation indicators, which solves the problem of how to adaptively eliminate the influence of working condition changes on degradation indicators while considering the influence of coupling effects of different degradation indicators on residual life under the condition of time-varying working conditions.

[0007] The present application is realized by the following technical solutions:

[0008] The present application provides an equipment residual life prediction method considering coupling effects of degradation indicators in the first aspect, comprising the following steps:

[0009] Step S1, obtaining a monitoring signal, extracting high-dimensional features from the monitoring signal and constructing a high-dimensional feature set;

[0010] Step S2, performing correlation analysis on the high-dimensional feature set to screen out two high-dimensional features with the largest correlation as degradation indicators representing the degradation state of the equipment;

[0011] Step S3, converting the degradation indicators to the reference working condition by using an adaptive signal conversion algorithm to obtain working condition adaptive degradation indicators;

[0012] Step S4, inputting the working condition adaptive degradation indicators into a pre-constructed nonlinear bivariate Wiener process degradation model to perform bivariate Wiener degradation analysis and obtain coupled degradation indicators;

[0013] Step S5, simulating the coupled degradation indicators by using a Monte Carlo simulation technique to obtain equipment residual life prediction results.

[0014] In the above technical solution, first, monitoring signals are obtained from multiple source sensors, and high-dimensional features are extracted from the monitoring signals to construct a high-dimensional feature set. Then, two high-correlation features are screened out from the high-dimensional feature set as degradation indicators representing the degradation state based on a correlation analysis method, wherein the two degradation indicators are high-representative features of the equipment degradation state.

[0015] Since the working condition frequently changes during the life cycle of the equipment, and the change of the working condition often affects the amplitude of the degradation data, making it more difficult to predict the residual life of the equipment under the time-varying working condition. To this end, the method divides the working condition into a reference working condition and a non-reference working condition. The reference working condition usually adopts the initial working condition, and the non-reference working condition includes all changed working conditions except the reference working condition. The adaptive signal conversion algorithm is used to convert the degradation indicators from other non-reference working conditions to the reference working condition to obtain working condition adaptive degradation indicators, thereby overcoming the problem of difficulty in predicting the residual life caused by the change of the working condition.

[0016] A nonlinear bivariate Wiener process with random effects is constructed to describe the mutual dependence between the two working condition adaptive degradation indicators to find the coupling relationship between the degradation indicators. Finally, the Monte Carlo simulation technique is used to simulate the coupled degradation indicators to obtain the equipment residual life prediction, which considers the influence of the coupling effect of different degradation indicators on the residual life while adaptively eliminating the influence of the working condition change on the degradation indicators, thereby realizing accurate equipment residual life.

[0017] In an optional embodiment, converting the degradation indicators to the reference working condition by using an adaptive signal conversion algorithm includes the following steps:

[0018] Step S31, obtaining a sampling time point under a changed working condition, and performing linear interpolation on the degradation indicators under the reference working condition based on the sampling time point to obtain degradation indicator estimation values;

[0019] Step S32, calculating the error sum of squares between the degradation index and the degradation index estimate value, minimizing the error sum of squares to obtain a minimized error sum of squares function;

[0020] Step S33, optimizing the minimized error sum of squares function by a two-dimensional search optimization algorithm to obtain a relationship estimate value;

[0021] Step S34, substituting the relationship estimate value into the degradation index under the changed working condition to obtain a working condition adaptive degradation index.

[0022] In an alternative embodiment, the working condition adaptive degradation index is input into a pre-constructed nonlinear bivariate Wiener process degradation model to perform bivariate Wiener degradation analysis to obtain a coupling degradation index, including:

[0023] Step S41, logarithmically transforming the working condition adaptive degradation index to obtain a transformed working condition adaptive degradation index;

[0024] Step S42, using a bivariate Wiener process in the nonlinear bivariate Wiener process degradation model to describe the correlation of the transformed working condition adaptive degradation index to obtain a coupling degradation index; wherein the increment of the transformed working condition adaptive degradation index within any time interval in the bivariate Wiener process obeys a bivariate normal distribution, and the bivariate normal distribution is the coupling degradation index of the transformed working condition adaptive degradation index.

[0025] In an alternative embodiment, the step S4 further includes the following steps:

[0026] Obtaining historical data, and using the historical data to perform parameter estimation on the nonlinear bivariate Wiener process degradation model to obtain a model parameter estimate value;

[0027] Based on the model parameter estimate value, using Bayesian theorem to update a random parameter in the nonlinear bivariate Wiener process degradation model.

[0028] In an alternative embodiment, using the historical data to perform parameter estimation on the nonlinear bivariate Wiener process degradation model includes the following steps:

[0029] Performing availability screening on the historical data to obtain availability historical data; wherein the availability screening includes: screening N available historical data from the historical data, and the nth equipment degradation index in the N available historical data is available at time , N represents the number of available degradation indices;

[0030] obtain unbiased estimation parameters by using the historical binary index increment data based on multivariate normal theory to make unbiased estimation on the mean and covariance matrix in the nonlinear binary Wiener process degradation model;

[0031] generate bootstrap samples from the historical data, and make bootstrap estimation on the hyperparameters in the nonlinear binary Wiener process degradation model by combining the bootstrap samples to obtain bootstrap hyperparameter estimation values.

[0032] In an optional embodiment, the time intervals between adjacent time points are the same, and the interval time is set to .

[0033] It should be noted that in the present embodiment, the nonlinear binary Wiener process degradation model is parameter-estimated and updated, and the binary normal distribution after parameter estimation and the binary normal distribution after updating are respectively used in Monte Carlo simulation.

[0034] In an optional embodiment, the coupled degradation index is simulated by using the Monte Carlo simulation technology, including the following steps:

[0035] Step S51, obtaining the binary normal distribution after parameter estimation in the nonlinear binary Wiener process degradation model, and sampling a future drift term parameter from the binary normal distribution after parameter estimation;

[0036] Step S52, obtaining the binary normal distribution after updating in the nonlinear binary Wiener process degradation model, and sampling a future degradation increment from the binary normal distribution after updating based on the future drift term parameter;

[0037] Step S53, calculating a future degradation trajectory by using the future degradation increment;

[0038] Step S54, constructing a failure threshold, comparing the future degradation trajectory with the failure threshold, and repeating steps S51 to S54 if the future degradation trajectory is less than the failure threshold.

[0039] The second aspect of the present application provides an equipment residual life prediction system considering the coupling effect of degradation indexes, comprising:

[0040] a feature construction module configured to obtain a monitoring signal, extract high-dimensional features from the monitoring signal, and construct a high-dimensional feature set;

[0041] an index screening module configured to perform correlation analysis on the high-dimensional feature set, and screen out two high-dimensional features with the largest correlation as degradation indexes representing the degradation state of the equipment;

[0042] an adaptive conversion module, configured to convert the degradation indicators to a reference working condition by using an adaptive signal conversion algorithm to obtain working condition adaptive degradation indicators;

[0043] a coupling module, configured to input the working condition adaptive degradation indicators into a pre-constructed nonlinear bivariate Wiener process degradation model to perform bivariate Wiener degradation analysis and obtain coupled degradation indicators;

[0044] a simulation module, configured to simulate the coupled degradation indicators by using a Monte Carlo simulation technique to obtain equipment residual life prediction results.

[0045] The third aspect of the present application provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the equipment residual life prediction method considering the coupling effect of degradation indicators when executing the computer program.

[0046] The fourth aspect of the present application provides a computer readable storage medium, which stores a computer program, and the computer program implements the equipment residual life prediction method considering the coupling effect of degradation indicators when executed by a processor.

[0047] Compared with the prior art, the present application has the following advantages and beneficial effects:

[0048] 1. The present application converts the degradation indicators from other non-reference working conditions to the reference working condition by using an adaptive signal conversion algorithm to obtain working condition adaptive degradation indicators, thereby overcoming the problem of residual life prediction difficulty caused by working condition changes;

[0049] 2. A nonlinear bivariate Wiener process with random effects is constructed to describe the mutual dependence between two working condition adaptive degradation indicators, to find the coupling relationship between the degradation indicators, and finally to simulate the coupled degradation indicators by using a Monte Carlo simulation technique to obtain equipment residual life prediction, which considers the influence of different degradation indicators coupling effect on residual life while adaptively eliminating the influence of working condition changes on degradation indicators, thereby realizing accurate equipment residual life. BRIEF DESCRIPTION OF DRAWINGS

[0050] In order to more clearly illustrate the technical solutions of the exemplary embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiments. It should be understood that the following drawings only show some embodiments of the present application, and therefore should not be considered as a limitation on the scope, and for those skilled in the art, other related drawings can also be obtained without creative labor. In the drawings:

[0051] Figure 1A flowchart of an equipment residual life prediction method considering coupling effects of degradation indicators is provided for the embodiments of the present application.

[0052] Figure 2 A result distribution diagram of equipment residual life prediction is provided for the embodiments of the present application.

[0053] Figure 3 A structural schematic diagram of an electronic device is provided for the embodiments of the present application. DETAILED DESCRIPTION

[0054] In order to make the objectives, technical solutions and advantages of the present application clearer, further detailed description will be made to the present application in combination with embodiments and drawings, and the illustrative embodiments of the present application and the description thereof are only used to explain the present application and do not limit the present application.

[0055] The embodiments of the present application provide an equipment residual life prediction method considering coupling effects of degradation indicators, as shown in the method comprises the following steps: Figure 1

[0056] Step S1, a monitoring signal is acquired, high-dimensional features are extracted from the monitoring signal and a high-dimensional feature set is constructed;

[0057] Step S2, correlation analysis is performed on the high-dimensional feature set, and two high-dimensional features with the largest correlation are screened out as degradation indicators representing the degradation state of the equipment;

[0058] Step S3, an adaptive signal conversion algorithm is used to convert the degradation indicators to a reference working condition, and a working condition adaptive degradation indicator is obtained;

[0059] Step S4, the working condition adaptive degradation indicator is input into a pre-constructed nonlinear binary Wiener process degradation model for binary Wiener degradation analysis, and a coupling degradation indicator is obtained;

[0060] Step S5, the coupling degradation indicator is simulated by using Monte Carlo simulation technology, and an equipment residual life prediction result is obtained.

[0061] It should be noted that at present, there are two problems for equipment residual life prediction under variable working conditions: on the one hand, the working condition frequently changes in the life cycle of the equipment, and the change of the working condition often has an impact on the amplitude of the degradation data, making it more difficult to predict the residual life of the equipment under time-varying working conditions; on the other hand, modeling based on a single degradation indicator is difficult to comprehensively reflect the degradation process of the equipment. Therefore, under the condition of time-varying working conditions, how to adaptively eliminate the influence of working condition change on the degradation indicator, while considering the influence of coupling effects of different degradation indicators on the residual life, so as to realize accurate equipment residual life prediction is particularly key.

[0062] ​To solve the problems in the prior art, the embodiment provides a residual life prediction method of equipment considering coupling effects of degradation indexes. In the method, first, monitoring signals are acquired from multiple source sensors, and high-dimensional features are extracted from the monitoring signals to construct a high-dimensional feature set. Then, two high-correlation features are selected from the high-dimensional feature set as degradation indexes representing the degradation state of the equipment based on a correlation analysis method, wherein the two degradation indexes are high-representativeness features of the degradation state of the equipment.

[0063] Since the working conditions of the equipment frequently change during the life cycle of the equipment, and the change of the working conditions often affects the amplitude of the degradation data, it is more difficult to predict the residual life of the equipment under time-varying working conditions. To this end, the method divides the working conditions into a reference working condition and non-reference working conditions. The reference working condition usually adopts an initial working condition, and the non-reference working conditions include all changed working conditions except the reference working condition. The adaptive signal conversion algorithm is used to convert the degradation indexes from other non-reference working conditions to the reference working condition, so as to overcome the problem of difficulty in predicting the residual life caused by the change of the working conditions.

[0064] A nonlinear binary Wiener process with random effects is constructed to describe the mutual dependence between the two working condition adaptive degradation indexes, so as to find the coupling relationship between the degradation indexes. Finally, the Monte Carlo simulation technology is used to simulate the coupled degradation indexes, so as to obtain the residual life prediction of the equipment. The influence of the coupling effects of different degradation indexes on the residual life is considered while the influence of the change of the working conditions on the degradation indexes is adaptively eliminated, so that the residual life of the equipment is accurately realized.

[0065] In the embodiment, for the correlation analysis method, the Pearson correlation analysis method can be used for screening. Specifically, the screening process is as follows:

[0066] The features under the reference working condition are subjected to correlation analysis with the equipment degradation state or time sequence, the Pearson correlation coefficient between the features and the equipment degradation state is calculated, and a correlation matrix is obtained. The calculation formula of the correlation analysis is as follows:

[0067] , wherein, is the equipment degradation state, is the correlation matrix, i=1, 2,..., h; j=1, 2,..., D, h is the number of sensor signal channels, and D is the number of features of each channel. represents a covariance function, is a standard deviation, is a feature.

[0068] The correlation matrices of N historical equipments are integrated, two features with high correlation are selected, and the two features are used as degradation indexes representing the degradation state of the equipment, and the correlation matrix of the two features is represented as and is represented.

[0069] In an alternative embodiment, the degradation index is converted to the reference working condition by using an adaptive signal conversion algorithm, comprising the following steps:

[0070] In step S31, the sampling time points in the changed working condition are obtained, and the degradation index in the reference working condition is linearly interpolated based on the sampling time points to obtain a degradation index estimation value.

[0071] In step S32, the error sum of squares between the degradation index and the degradation index estimation value is calculated, and the error sum of squares is minimized to obtain a minimized error sum of squares function.

[0072] In step S33, the minimized error sum of squares function is optimized by a two-dimensional search optimization algorithm to obtain a relationship estimation value.

[0073] In step S34, the relationship estimation value is substituted into the degradation index in the changed working condition to obtain a working condition adaptive degradation index.

[0074] It should be noted that the purpose of this step is to design an adaptive signal conversion algorithm to eliminate the influence of working condition changes on the degradation index. Among them, considering the influence of working condition changes on the slope and amplitude of the degradation index, the degradation index in different working conditions is converted to the reference working condition by using the error sum of squares minimization algorithm.

[0075] In this embodiment, the first degradation index is taken as an example, and the relationship between the degradation index in the reference working condition and the index in other working conditions can be expressed as:

[0076] , wherein, is the degradation index of the tool in the reference working condition at time t, K is the current time, is the degradation index in the working condition , and is the degradation index in the working condition , and are adjustment parameters for adjusting the relationship between the degradation index in the reference working condition and the index in other working conditions.

[0077] Specifically, the sampling time points in the changed working condition p are obtained, and linear interpolation is performed within the amplitude range of the degradation index in the reference working condition to obtain a degradation index estimation value of the degradation index in the reference working condition .

[0078] The error sum of squares between the degradation index and the degradation index estimation value is calculated, and then the adjustment parameters are estimated by minimizing the error sum of squares function, and the minimized error sum of squares function is calculated as follows:

[0079] wherein, is an error sum of squares function, represents a set of sampling times at which the nth equipment operates under condition p. is a degradation index of the nth equipment under condition p at time K, is a degradation index of the nth equipment under condition p at time K, is a degradation index of the nth equipment under condition p at time K, is an estimated value of the degradation index of the nth equipment under condition p at time K, is a degradation index of the nth equipment under condition p at time K,

[0080] The estimated value of the degradation index under condition p is obtained by minimizing the error sum of squares function through a two-dimensional search optimization algorithm , and and the estimated values of the degradation index under the reference condition and the relationship between the indexes under other conditions are substituted into the relationship formula, to obtain the value of the degradation index under condition p, i.e. the transformed degradation index y 1( t ). Similarly, the above transformation operation is performed on the second degradation index to obtain the transformed degradation index y 2( t ). The two degradation indexes are used for subsequent residual life prediction.

[0081] In an optional embodiment, the condition-adaptive degradation index is input into a pre-constructed nonlinear bivariate Wiener process degradation model for bivariate Wiener degradation analysis to obtain a coupled degradation index, including:

[0082] Step S41, logarithmically transforming the condition-adaptive degradation index to obtain a transformed condition-adaptive degradation index;

[0083] Step S42, using the bivariate Wiener process in the nonlinear bivariate Wiener process degradation model to describe the correlation of the transformed condition-adaptive degradation index to obtain a coupled degradation index; wherein the increment of the transformed condition-adaptive degradation index in any time interval in the bivariate Wiener process follows a bivariate normal distribution, and the bivariate normal distribution is the coupled degradation index of the transformed condition-adaptive degradation index.

[0084] It should be noted that it is difficult to comprehensively reflect the degradation process of the equipment based on a single degradation index, and therefore the purpose of this step is to describe the mutual dependence relationship of the degradation indexes through a nonlinear bivariate Wiener process with random effects, and to model the degradation process of the equipment using two degradation indexes, so as to comprehensively reflect the degradation process of the equipment.

[0085] ​Wherein, the nonlinear binary Wiener process random degradation model is as follows: Wherein, y(t) represents the value of the transformed degradation index at t time, is the degradation rate, is a diffusion term representing the random fluctuation of the degradation process, wherein is a standard Brownian motion, is a diffusion term parameter.

[0086] In order to facilitate calculation, the working condition adaptive degradation index is logarithmically transformed, and the working condition adaptive degradation index y 1( t ) and y 2( t ) are transformed into transformed working condition adaptive degradation indexes x1( t ) and x2( t ). The correlation between the two degradation indexes is described by a binary Wiener process, specifically, the increments of the logarithmically transformed values x1( t ) and x2( t ) of the two degradation indexes between any t and follow a binary normal distribution with mean and variance . .

[0087] That is , , wherein a, b, c, d and are all parameters of the nonlinear Wiener process random degradation model.

[0088] In an alternative embodiment, the step S4 further comprises the following steps:

[0089] Obtaining historical data, and using the historical data to estimate the parameters of the nonlinear binary Wiener process degradation model to obtain model parameter estimation values;

[0090] Based on the model parameter estimation values, using Bayesian theorem to update the random parameters in the nonlinear binary Wiener process degradation model.

[0091] In an alternative embodiment, using the historical data to estimate the parameters of the nonlinear binary Wiener process degradation model comprises the following steps:

[0092] Performing availability screening on the historical data to obtain available historical data; wherein the availability screening comprises: screening N available historical data from the historical data, and the nth equipment degradation index in the N available historical data is available at time . ​denotes the number of available degradation indicators;

[0093] calculating the historical binary indicator increment data of the nth equipment degradation indicator, performing unbiased estimation on the mean and covariance matrix in the nonlinear binary Wiener process degradation model based on the historical binary indicator increment data to obtain unbiased estimation parameters;

[0094] generating a bootstrap sample from the historical data, and performing bootstrap estimation on the hyperparameters in the nonlinear binary Wiener process degradation model to obtain a bootstrap hyperparameter estimation value.

[0095] In an optional embodiment, the time interval between adjacent time points is the same, and the interval time is set to .

[0096] It should be noted that in this embodiment, the nonlinear binary Wiener process degradation model is parameter estimated and updated, and the binary normal distribution after parameter estimation and the binary normal distribution after updating are respectively used in Monte Carlo simulation.

[0097] In this embodiment, the parameter estimation of the nonlinear binary Wiener process degradation model specifically includes:

[0098] First, the historical data is obtained, N available historical data is selected from the historical data, and the nth equipment degradation indicator is available at time , wherein denotes the number of available degradation indicators.

[0099] Then, the historical binary degradation indicator increment data of the nth equipment is denoted as , wherein the interval between two adjacent time points is set to .

[0100] Based on the historical binary indicator increment data, the mean and covariance matrix in the nonlinear binary Wiener process degradation model are unbiased estimated to obtain unbiased estimation parameters, and the calculation process is as follows:

[0101] ; ; wherein is the estimation result of the mean in the binary Wiener process degradation model, is the estimation result of the covariance in the binary Wiener process degradation model. Thus, we have:

[0102] .

[0103] Random hyper-parameter estimation of the drift term. Due to the randomness of the individual service environment of the equipment, different individuals have variability in the degradation path. In order to describe this phenomenon, consider that a and c obey normal distribution with mean and variance , that is , .

[0104] Next, the hyper-parameters of a and c are estimated. First, H bootstrap samples are generated from the historical data . Combined with the bootstrap samples, the bootstrap estimates of a and c are obtained as and , where i = 1, 2,..., H. Since a and c obey normal distribution, the log-likelihood function of them can be expressed as: ; .

[0105] The hyper-parameter estimates of a and c can be obtained by maximum likelihood estimation .

[0106] Further, based on Bayes' theorem, the random parameters in the model are updated based on the online degradation indicators to update the two-dimensional Wiener process random degradation model;

[0107] Based on Bayes' theorem, the posterior distribution of can be expressed as: .

[0108] The posterior distribution further can be expressed as:

[0109] .

[0110] For convenience, let .

[0111] Based on this, further derivation is as follows: .

[0112] Finally, we get:

[0113] where A, B, D, E, F, G are intermediate parameters and have no practical significance.

[0114] In an alternative embodiment, the coupling degradation indicator is simulated using Monte Carlo simulation technology, including the following steps:

[0115] Step S51: Obtain the bivariate normal distribution after parameter estimation in the nonlinear bivariate Wiener process degradation model, and sample the future drift term parameters from the bivariate normal distribution after parameter estimation.

[0116] Step S52: Obtain the updated binary normal distribution in the nonlinear binary Wiener process degradation model, and sample the future degradation increment from the updated binary normal distribution based on the future drift term parameter;

[0117] Step S53: Calculate the future degradation trajectory using the future degradation increment;

[0118] Step S54: Construct a failure threshold, compare the future degradation trajectory with the failure threshold, and if the future degradation trajectory is less than the failure threshold, repeat steps S51 to S54.

[0119] It should be noted that obtaining Parameters of the drift term at time step: from the bivariate normal distribution Mid-sampling yields the drift term parameters for the j-th future step. .

[0120] Obtaining future degradation increments based on future drift term parameters: from a binary normal distribution The increment for the j-th step is obtained by sampling. .

[0121] Calculate the future degradation trajectory using future degradation increments: ,in, and For the future j-step degradation trajectory and For the future J-1 step degradation trajectory.

[0122] Build failure threshold and Failure determination is based on failure thresholds:

[0123] if or Then let ;

[0124] Otherwise, j = j + 1, and return S52.

[0125] The procedure is as follows:

[0126] if or ( and (Failure threshold);

[0127] ;

[0128] break;

[0129] else;

[0130] j = j + 1, return S52.

[0131] After repeating steps S51 to S54 L times to obtain L simulated remaining life prediction results, the distribution of the remaining life and important indicators such as the mean, median and confidence interval can be approximately estimated by the simulation results, wherein, the remaining life is represented by .

[0132] Further, the method proposed in the embodiment is verified by using the full-life tool milling data.

[0133] First, the features of each channel data are extracted from the time domain, frequency domain and time-frequency domain. The time domain includes 5 dimensioned features and 6 dimensionless features, including mean, variance, root mean square, average absolute value and peak value, waveform index, peak index, pulse index, margin, skewness and kurtosis. The frequency domain signal is composed of full-band signal and sub-band signal, and 11 features are extracted from the full-band and sub-band, including spectrum sum, spectrum mean, spectrum variance, spectrum entropy, spectrum root mean square, spectrum kurtosis, spectrum skewness, spectral shape mean, spectral shape variance, spectral shape third moment and spectral shape fourth moment. 8 time-frequency domain features are generated by Haar wavelet packet transform and three-layer decomposition.

[0134] By calculating the Pearson correlation coefficient of each feature and the tool wear value, high representation features are selected as degradation indicators. In working condition 1, the variance of the X-axis vibration signal 1000-2000Hz band (correlation coefficient 0.964) and the time domain variance of the Y-axis vibration signal (correlation coefficient 0.940) are selected as degradation indicators. In working condition 2, the variances of the X-axis cutting force signal and the X-axis vibration signal 1000-2000Hz band (correlation coefficients are 0.949 and 0.937, respectively) are selected as degradation indicators. The signal conversion method is used to convert the degradation indicators to the reference machining parameters. In addition, the moving average algorithm is used to further smooth the converted signal to reduce the influence of random noise, and the window size of the moving average algorithm is set to 3.

[0135] As shown in Figure 2 , (a) is the result distribution diagram of the tool at 0 sampling points, Figure 2 (b) is the result distribution diagram of the tool at 220 sampling points, Figure 2 (c) is the result distribution diagram of the tool at 110 sampling points, Figure 2 (c) is the result distribution diagram of the tool at 110 sampling points, Figure 2 ​(d) is a result distribution diagram of the tool at 330 sampling points, and the four parts described above draw the approximate remaining life distribution prediction results of C1-1 at four sampling time points (t k =0, 110, 220, 330) respectively. It can be seen that as the degradation data accumulates, the variance of the remaining life gradually decreases, and the point estimate result (mean) is closer and closer to the true value.

[0136] The embodiment of the present application provides an equipment remaining life prediction system considering the coupling effect of degradation indexes, which is used to realize the equipment remaining life prediction method considering the coupling effect of degradation indexes provided in the above embodiment.

[0137] Specifically, the equipment remaining life prediction system considering the coupling effect of degradation indexes comprises:

[0138] A feature construction module is configured to acquire a monitoring signal, extract high-dimensional features from the monitoring signal, and construct a high-dimensional feature set;

[0139] An index screening module is configured to perform correlation analysis on the high-dimensional feature set, and screen out two high-dimensional features with the largest correlation as degradation indexes representing the degradation state of the equipment;

[0140] An adaptive conversion module is configured to convert the degradation indexes to a reference working condition by using an adaptive signal conversion algorithm to obtain working condition adaptive degradation indexes;

[0141] A coupling module is configured to input the working condition adaptive degradation indexes into a pre-constructed nonlinear binary Wiener process degradation model to perform binary Wiener degradation analysis, and obtain coupled degradation indexes;

[0142] A simulation module is configured to simulate the coupled degradation indexes by using a Monte Carlo simulation technique to obtain equipment remaining life prediction results.

[0143] The embodiment of the present application also provides an electronic device, as shown in the figure, the electronic device comprises a processor 21, a memory 22, an input device 23 and an output device 24; the number of processors 21 in the computer device can be one or more, Figure 3 for example, one processor 21; the processor 21, the memory 22, the input device 23 and the output device 24 in the electronic device can be connected through a bus or other means, Figure 3 for example, connected through a bus. Figure 3

[0144] ​The memory 22 can be used to store software programs, computer executable programs and modules as a computer readable storage medium. The processor 21 executes various functions and data processing of the electronic device by running the software programs, instructions and modules stored in the memory 22, that is, implements the equipment residual life prediction method considering coupling effect of degradation indicators as described in the above embodiments.

[0145] The memory 22 can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system and at least one application required by a function; the data storage area can store data created according to the use of the terminal, etc. In addition, the memory 22 can include a high-speed random access memory, and can also include a non-volatile memory, such as at least one magnetic disk storage device, a flash memory device, or other non-volatile solid-state memory device. In some examples, the memory 22 can further include a memory remotely arranged with respect to the processor 21, which can be connected to the electronic device through a network. Examples of the above network include but are not limited to the Internet, an intranet, a local area network, a mobile communication network, and a combination thereof.

[0146] The input device 23 can be used to receive the id and password input by the user, etc. The output device 24 is used to output the network configuration page.

[0147] The embodiments of the present application also provide a computer readable storage medium, and the computer executable instructions are used to implement the equipment residual life prediction method considering coupling effect of degradation indicators as provided in the above embodiments when executed by a computer processor.

[0148] The storage medium provided by the embodiments of the present application contains computer executable instructions, which are not limited to the method operations provided in the above embodiments, and can also perform related operations in the equipment residual life prediction method considering coupling effect of degradation indicators provided by any embodiment of the present application.

[0149] The above specific embodiments further explain the purpose, technical solutions and beneficial effects of the present application. It should be understood that the above description is only a specific embodiment of the present application and is not used to limit the protection scope of the present application. Any modification, equivalent replacement, improvement, etc. within the spirit and principle of the present application should be included in the protection scope of the present application.

Claims

1. A method for predicting the residual life of equipment considering the coupling effect of degradation indicators, characterized in that, The method comprises the following steps: Step S1, obtaining a monitoring signal, extracting high-dimensional features from the monitoring signal, and constructing a high-dimensional feature set; Step S2, performing correlation analysis on the high-dimensional feature set, and screening out two high-dimensional features with the largest correlation as degradation indexes representing the degradation state of the equipment; Step S3, converting the degradation indexes to a reference working condition by using an adaptive signal conversion algorithm to obtain working condition adaptive degradation indexes; Step S4, input the working condition adaptive degradation index into the pre-constructed nonlinear bivariate Wiener process degradation model for bivariate Wiener degradation analysis to obtain the coupling degradation index; wherein, inputting the working condition adaptive degradation index into the pre-constructed nonlinear bivariate Wiener process degradation model for bivariate Wiener degradation analysis to obtain the coupling degradation index comprises: step S41, logarithmically transforming the working condition adaptive degradation index to obtain a transformed working condition adaptive degradation index; step S42, using the bivariate Wiener process in the nonlinear bivariate Wiener process degradation model to describe the correlation of the transformed working condition adaptive degradation index to obtain the coupling degradation index; wherein, the increment of the transformed working condition adaptive degradation index in any time interval in the bivariate Wiener process obeys a bivariate normal distribution, and the bivariate normal distribution is the coupling degradation index of the transformed working condition adaptive degradation index; wherein, the step S4 further comprises the following steps: obtaining historical data, using the historical data to perform parameter estimation on the nonlinear bivariate Wiener process degradation model to obtain a model parameter estimation value; based on the model parameter estimation value, using Bayes theorem to update the random parameters in the nonlinear bivariate Wiener process degradation model; wherein, using the historical data to perform parameter estimation on the nonlinear bivariate Wiener process degradation model comprises the following steps: performing availability screening on the historical data to obtain availability historical data; wherein, the availability screening comprises: screening N available historical data from the historical data, and the nth equipment degradation index in the N available historical data is available at time t n, represents the number of available degradation indexes; calculating the historical bivariate index increment data of the nth equipment degradation index, using the historical bivariate index increment data to perform unbiased estimation on the mean and the covariance matrix in the nonlinear bivariate Wiener process degradation model based on multivariate normal theory to obtain an unbiased estimation parameter; generating a bootstrap sample from the historical data, combining the bootstrap sample to perform bootstrap estimation on the hyperparameters in the nonlinear bivariate Wiener process degradation model to obtain a bootstrap hyperparameter estimation value; The stochastic degradation model of the nonlinear binary Wiener process is as follows: Where y(t) represents the value of the degradation index after transformation at time t. For degradation rate, The diffusion term characterizes the random fluctuations of the degradation process, where For standard Brownian motion, The parameters are for the diffusion term. The adaptive degradation indices of the operating condition are logarithmically transformed, transforming the adaptive degradation indices y1(t) and y2(t) into transformed adaptive degradation indices x1(t) and x2(t). The logarithmically transformed values ​​x1(t) and x2(t) of the two degradation indices at any t and between The increment follows the mean. variance is bivariate normal distribution ;in, , Among them, a, b, c, d and These are all parameters of a nonlinear Wiener process stochastic degradation model; Step S5, simulating the coupling degradation indexes by using a Monte Carlo simulation technique to obtain an equipment residual life prediction result. 2.The equipment residual life prediction method considering coupling effects of degradation indicators according to claim 1, wherein, The method for converting the degradation indexes to the reference working condition by using the adaptive signal conversion algorithm comprises the following steps: Step S31, obtaining a sampling time point under a changed working condition, performing linear interpolation on the degradation indexes under the reference working condition based on the sampling time point to obtain degradation index estimation values; Step S32, calculating the error sum of squares between the degradation indexes and the degradation index estimation values, minimizing the error sum of squares to obtain a minimized error sum of squares function; Step S33, optimizing the minimized error sum of squares function by using a two-dimensional search optimization algorithm to obtain a relationship estimation value; Step S34, substituting the relationship estimation value into the degradation indexes under the changed working condition to obtain the working condition adaptive degradation indexes. 3.The equipment residual life prediction method considering coupling effects of degradation indicators according to claim 1, wherein, The time The interval time between adjacent time points is the same, and the interval time is set as . 4.The method of claim 1, wherein, The method for simulating the coupling degradation indexes by using the Monte Carlo simulation technique comprises the following steps: Step S51, obtaining a bivariate normal distribution after parameter estimation in the non-linear bivariate Wiener process degradation model, and sampling future drift term parameters from the bivariate normal distribution after parameter estimation; Step S52, obtaining a bivariate normal distribution after updating in the non-linear bivariate Wiener process degradation model, and sampling future degradation increments from the bivariate normal distribution after updating based on the future drift term parameters; Step S53, calculating a future degradation trajectory by using the future degradation increments; Step S54, constructing a failure threshold, comparing the future degradation trajectory with the failure threshold, and repeating steps S51 to S54 if the future degradation trajectory is smaller than the failure threshold.

5. An electronic device, comprising: The equipment residual life prediction method considering the coupling effect of degradation indexes comprises a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the equipment residual life prediction method considering the coupling effect of degradation indexes according to any one of claims 1 to 4 when executing the computer program.

Images