Industrial equipment fault diagnosis method and system based on multi-view integration characteristics

Abstract

The invention discloses an industrial equipment fault diagnosis method and system based on multi-view integration characteristics, and relates to the technical field of industrial equipment fault diagnosis. The invention aims to solve the problems that the performance is obviously reduced when the existing method is used for processing high-noise, small-sample or non-stable working condition data, and the precision and reliability of fault diagnosis of industrial equipment are seriously restricted. According to the technical key points, the method comprises the following steps: obtaining a time sequence data set of an operation fault state of industrial equipment, carrying out feature extraction and enhancement on the time sequence data, carrying out statistical feature extraction on an original time sequence, a first-order difference time sequence and a time sequence sample set after expansion mapping to obtain feature vectors, and splicing the feature vectors into a high-dimensional initial feature vector set; performing multi-view integrated feature selection on the initial feature vector set, and performing multi-metric scoring and initial view generation on the high-dimensional feature set; and carrying out heterogeneous integrated classification and diagnosis, and outputting a result. Through three core technologies of multi-view feature enhancement, dynamic selection and heterogeneous integration, the key bottleneck of a traditional fault diagnosis method in industrial fault diagnosis is solved, and an innovative solution is provided for high-precision and low-cost equipment health management. According to the method, feature diversity, expression ability and model robustness can be considered, and the method adapts to harsh constraints of industrial scenes.

Description

Technical Field

[0001] The present invention belongs to the technical field of industrial equipment fault diagnosis, and in particular relates to an industrial equipment fault diagnosis method and system based on multi-view integrated features. Background Art

[0002] With the increasing intelligence and complexity of industrial equipment, fault diagnosis technology based on condition monitoring signals has become a key means to ensure the reliable operation of industrial systems. However, in actual industrial scenarios, only single-channel monitoring signals (such as current, vibration, etc.) can usually be obtained, and it is impossible to rely on multi-source data or domain expert knowledge for auxiliary analysis. In this context, data-driven univariate time series classification methods have become the core approach to solving fault diagnosis problems. Existing feature-based fault diagnosis methods extract descriptive statistical features (such as mean, variance, entropy, etc.) from time series and combine them with classifiers to achieve fault type discrimination. Its process is transparent and highly interpretable, and it has been widely used in industrial equipment health management.

[0003] In recent years, research on industrial equipment fault diagnosis has been increasing. For example, patent document CN117726320A provides a method and apparatus for industrial equipment fault diagnosis. This method first constructs a knowledge graph for industrial equipment fault diagnosis; then constructs a user-item interaction matrix and updates the values ​​of each element in the user-item interaction matrix based on the fault triples; then constructs a knowledge recommendation model based on graph neural networks and multi-task learning; and finally trains and validates the knowledge recommendation model based on graph neural networks and multi-task learning. This prior art introduces knowledge graphs and recommendation systems into the field of fault diagnosis. A knowledge recommendation model based on graph neural networks and multi-task learning extracts feature information from the industrial equipment fault diagnosis knowledge graph and embeds it into learning and knowledge reasoning. By training the user-item interaction matrix and predicting the user's goals, this method provides industrial equipment maintenance personnel with a fast, accurate, and personalized repair method when handling equipment faults. Patent document CN118296452A discloses an industrial equipment fault diagnosis method based on Transformer model optimization, which addresses the shortcomings of fault diagnosis models such as low accuracy, poor generalization, and weak anti-interference ability. The method includes the following steps: acquiring industrial equipment operating data; preprocessing the industrial equipment operating data; constructing an industrial equipment fault diagnosis model; training the industrial equipment fault diagnosis model; acquiring the industrial equipment operating data to be diagnosed; and diagnosing the industrial equipment fault. The method uses an eagle swarm optimization algorithm to optimize the Transformer model structure, ensuring the optimal number of encoder layers, number of attention heads, and number of algorithm model iterations during model operation. This increases the model's robustness and generalization, improving the accuracy of industrial equipment fault diagnosis. The prior art with document number CN118034261A discloses an industrial equipment fault diagnosis system and method based on artificial intelligence, which includes a fault detection unit: setting basic target operating data, and establishing a standard range of operating data, and then analyzing whether the basic target has a fault based on the range of change of the operating data. This document analyzes whether there is a fault in the industrial equipment line through the range of change of the operating data, and matches the corresponding fault solution, which can reduce the time for matching industrial equipment fault diagnosis solutions, realize real-time monitoring of fault diagnosis, and quickly locate the fault; the binary positioning of line nodes for maintenance can save the time of fault diagnosis and positioning to the maximum extent, and perform priority diagnosis in conjunction with the marked fault interval, and can make priority judgments based on previous fault data. Based on artificial intelligence analysis, the efficiency of industrial equipment fault diagnosis is effectively improved.

[0004] However, the existing technology still has the following significant defects:

[0005] Insufficient feature redundancy and scalability: Mainstream feature extraction tools (such as HCTSA, Catch22, and TSFresh) generate a large number of features using preset statistical formulas. However, these feature sets often contain highly redundant statistical patterns (such as repeated time-domain statistics). For example, of the more than 7,700 features extracted by HCTSA, approximately 60% are significantly correlated, making subsequent classifiers susceptible to interference from redundant features. Furthermore, existing frameworks lack flexible feature expansion mechanisms, making it difficult to introduce new low-redundancy features.

[0006] Limited feature expression capabilities: Traditional methods typically directly perform global statistics on the original time series or extract local features through sliding windows, but fail to effectively combine the complementary information of global and local patterns. For example, while TSFresh supports multi-scale feature extraction, its window division method is fixed and cannot adaptively enhance time series patterns of different granularities. Furthermore, existing methods ignore the structural relationships within the sequence (such as the spatial distribution of time series points and phase changes), making it difficult for features to capture the deep discriminative information of fault signals.

[0007] Simplified feature selection strategies: Existing methods often use embedded feature selection (such as LASSO regression and random forest importance scoring) or single filtering metrics (such as mutual information and variance thresholds), failing to consider the interaction effects and stability between features. For example, TSFresh uses hypothesis testing to select features, but its independence assumption often fails in real-world industrial data, leading to the inadvertent deletion of important features. Furthermore, existing methods lack dynamic adjustment mechanisms, making it impossible to generate robust feature subsets tailored to the characteristics of different datasets.

[0008] Insufficient classifier diversity: Current feature-based fault diagnosis algorithms often use a single classification model (such as SVM and random forest) or simply integrate homogeneous classifiers, which limits the model's generalization ability. For example, although FreshP improves classification performance by rotating forests, its base classifiers are all decision trees, which make it difficult to cover the discrimination boundaries under different data distributions.

[0009] These shortcomings severely limit the accuracy and reliability of industrial equipment fault diagnosis. Existing methods significantly degrade in performance, especially when dealing with high-noise, small sample sizes, or non-stationary operating conditions. Therefore, a new fault diagnosis framework is urgently needed that balances feature diversity, expressiveness, and model robustness while adapting to the stringent constraints of industrial scenarios. Summary of the Invention

[0010] Technical Problems to be Solved by the Invention

[0011] In view of the defects of the existing methods described in the background technology, the present invention needs to solve the following technical problems:

[0012] 1. Build a scalable feature enhancement framework to break through the redundancy bottleneck of traditional feature sets, support the flexible introduction of new low-redundancy features, and dynamically optimize feature combinations to adapt to the needs of different industrial scenarios;

[0013] 2. Achieve multi-view collaborative enhancement of global and local patterns. Through an innovative sequence mapping mechanism, it simultaneously captures the macro trends and micro fluctuations of time series signals, improving the discriminative ability of features.

[0014] 3. Design an adaptive feature selection strategy based on the dual indicators of feature stability and diversity to dynamically screen high-value feature subsets to avoid overfitting or information loss caused by traditional single metrics;

[0015] 4. Construct a heterogeneous ensemble classifier, integrate classification models based on different principles (such as ridge regression and extreme random trees), and improve the model's generalization ability for complex fault modes through diversity enhancement mechanisms (such as hard voting and weighted fusion).

[0016] The technical solution adopted by the present invention to solve the above technical problems is:

[0017] The present invention provides an industrial equipment fault diagnosis method based on multi-view integrated features, which is characterized by comprising the following steps:

[0018] Step S1: Obtain a time series data set of the fault status of industrial equipment, wherein the time series data set includes a sample set S of a unary time series. L =(s1,s2,…,s n ) and the label set L=(l1,l2,…,l n ) consists of a training set Train=(S L ,L) and the sample set S of the univariate time series T =(s n+1 ,s n+2 ,…,s n+m ) composed of a test set Test=(S T ), n and m are the number of time series samples included in the training set and the test set respectively, and the time series sample s i =(s1,s2,…,s σ ) are all univariate time series with no missing values, equal intervals, and equal lengths, 1≤i≤n+m, σ is the length of the time series sample;

[0019] Step S2: extracting and enhancing features of the time series dataset, including:

[0020] Step S2.1: Extract statistical features from the original time series sample set S to obtain the feature vector FV RThe original time series sample set S includes the training set unary time series sample set S L and the test set univariate time series sample set S T ;

[0021] Perform c groups of expansion mappings of different scales on the original time series sample set S to obtain the sample set S after expansion mapping s1 ,…,S sc , for the sample set S s1 ,…,S sc Perform statistical feature extraction to obtain the feature vector FV Rs1 ,…,FV Rsc ;

[0022] Step S2.2: Perform first-order difference transformation on the original time series sample set S to obtain the first-order difference time series sample set S D ;

[0023] For the first-order difference time series sample set S D Perform statistical feature extraction to obtain the feature vector FV D ;

[0024] For the first-order difference time series sample set S D Perform c groups of expansion mappings of different scales to obtain the sample set S after expansion mapping Ds1 ,…,S Dsc , for the sample set S Ds1 ,…,S Dsc Perform statistical feature extraction to obtain the feature vector FV Ds1 ,…,FV Dsc ;

[0025] Step S2.3: The feature vector FV R , FV Rs1 ,…,FV Rsc , FV D , FV Ds1 ,…,FV Dsc Splicing into a high-dimensional initial feature vector set FV;

[0026] Step S3, performing multi-view integrated feature selection on the initial feature vector set FV, including:

[0027] Step S3.1: Multi-metric scoring and initial view generation for high-dimensional feature sets:

[0028] Use M independent feature evaluation indicators to score each feature in the initial feature vector set FV. Each feature evaluation indicator generates a corresponding scoring matrix, and M scoring matrices are obtained.

[0029] According to the scoring matrix corresponding to the i-th feature evaluation index, for each feature, sort in descending order and select the top K1, K2, ..., K T high-scoring features, and obtain the initial view feature set FV corresponding to the i-th feature evaluation index 1i ,FV 2i ,…,FV Ti , i is a positive integer less than M;

[0030] The i increases from 1 to M in sequence, i = 1, 2, ..., M, and the initial view feature set corresponding to the first feature evaluation index is obtained in sequence The initial view feature set FV corresponding to the second feature evaluation index 12 ,FV 22 ,…,FV T2 , ..., the initial view feature set corresponding to the Mth feature evaluation index

[0031] The number of features K1, K2, ..., K T According to the preset convergence threshold list thresholds=[threshold1,threshold2,…,threshold T ] Determine and traverse all candidate feature quantities K (K = 1, 2, ..., N feature ), calculated using the formula Select the values ​​that satisfy rate(K)≤threshold i The minimum number of features is K i , 1≤i≤T, where rate(K) is the relative change rate of the stability-diversity score (SD-score), N feature is the number of features of the initial feature vector set FV;

[0032] For each candidate feature number K, according to the scoring matrix corresponding to each feature evaluation index j, the top K high-scoring features are selected from the initial feature vector set FV to obtain the feature subset The j increases from 1 to M in sequence, j = 1, 2, ..., M, to obtain the feature subsets corresponding to all feature evaluation indicators. Calculate the feature subset corresponding to the number of candidate features K The absolute value matrix of the Pearson correlation coefficient Each sub-block of the Pearson correlation coefficient absolute value matrix A(K) are all K×K dimensional square matrices, Representing feature subsets With feature subset The absolute value matrix of the Pearson correlation coefficient between 1≤p≤M, 1≤q≤M;

[0033] The stability-diversity score (SD-score) corresponding to the number of candidate features K is calculated by the formula Calculate, where Represents a pair of blocks Sum all elements of ;

[0034] Step S3.2: Generate an intersection-optimized extended view feature set:

[0035] For the initial view feature set FV corresponding to the first feature evaluation index 11 ,FV 21 ,…,FV T1 , the initial view feature set FV corresponding to the second feature evaluation index 12 ,FV 22 ,…,FV T2 , ..., the initial view feature set FV corresponding to the Mth feature evaluation index 1M ,FV 2M ,…,FV TM , respectively take the feature index intersection of the initial view feature set corresponding to different feature evaluation indicators, and generate T common feature subsets in turn in

[0036] The common feature subset Spliced ​​into a common feature vector FV interse , and the common eigenvector FV intersection The initial view feature set FV corresponding to all feature evaluation indicators 11 ,FV 21 ,…,FV T1 , FV 12 ,FV 22 ,…,FV T2 ,……,FV 1M ,FV 2M ,…,FV TM Splicing is performed to generate N = T·M extended view feature sets FV' V1 ,FV' V2 ,…,FV' VN ;

[0037] Step S3.3: De-redundancy processing, outputting the final multi-view feature set:

[0038] For each extended view feature set FV' V1 ,FV' V2 ,…,FV' VNThe Pearson correlation coefficient analysis is performed on the features, and redundant features with correlation coefficients greater than the preset threshold are removed to obtain the final multi-view feature set FV'1, FV'2, ..., FV' N ;

[0039] Step S4: Heterogeneous integration classification and diagnosis:

[0040] Classifier training: using the training set Train = (S L ,L) Obtain the training set multi-view feature set through steps S2 and S3 Feature set in each view Train the classifier CL on i , 1≤i≤N;

[0041] Hard voting integration: using the test set Test = (S T ) After steps S2 and S3, the test set multi-view feature set is obtained The feature set of each view Input classifier CL i , get the multi-view prediction result L i , 1≤i≤N;

[0042] The classification results of all views L1, L2, L, L are integrated using the majority voting rule N , output the final fault label L T ;

[0043] Step S5, result output: the diagnostic label L T The information is transmitted to the industrial equipment control terminal through the communication module, triggering fault warnings or maintenance instructions.

[0044] Furthermore, the first-order difference transformation of the original time series sample set S in step S2.2 can be implemented by a tracking differentiator with an adjustable filter factor k. The functional implementation process of the tracking differentiator with an adjustable filter factor is:

[0045] The calculation formula for the first column x1[1] of the matrix x1 and the first column x2[1] of the matrix x2 is:

[0046]

[0047] Wherein, time series sample s is the time series sample of the input tracking differentiator in the original time series sample set S, s[i] is the i-th value in the time series sample s, 1≤i≤σ, where σ is the length of the time series sample s;

[0048] For i=1,2,…,σ increases successively:

[0049] y i=x1[i]-s[i]+khx2[i];

[0050] If |y i |>k 2 h 2 r, then otherwise,

[0051] If |a|>rhk, then fh i = -r·sign(a), otherwise,

[0052] The formula for calculating the i+1th column x1[i+1] of the matrix x1 and the i+1th column x2[i+1] of the matrix x2 is:

[0053]

[0054] The differential time series is transformed by a tracking differentiator with an adjustable filter factor to obtain the following differential time series: Wherein, r is a preset speed factor parameter, h is a preset step size parameter, and k is an adjustable filter factor.

[0055] Furthermore, the iterative optimization process of the adjustable filter factor k of the tracking differentiator is:

[0056] (a) Set the optimization parameters: number of candidate solutions nc = 10, maximum number of iterations inter max =10, filter factor search range [b L ,b R ], where the search range [b L ,b R ]The default value is [1,3];

[0057] (b) Define the objective function f(k):

[0058] i. For the training set time series S L Apply filter factor k to generate differential sequence set

[0059] ii. Yes After sorting the rows in ascending order, the following sampling rate d rate =4 Extract quantile features

[0060] iii. Calculate feature set The silhouette score (S-score) of the label L is used as the objective function value f(k), and the filter factor k is iteratively optimized by maximizing the silhouette score (S-score);

[0061] The silhouette score (S-score) is defined as: Among them, n is the number of time series samples contained in the training set, SC(i) represents the feature set The silhouette coefficient of the eigenvector corresponding to the i-th time series sample in , Among them, η(i) represents the minimum distance between clusters, that is, the feature set The minimum value of the Euclidean distance between the feature vector corresponding to the i-th time series sample and the feature vectors corresponding to all time series samples with different labels, Represents the average distance within the cluster, that is, the feature set The average value of the Euclidean distance between the feature vector corresponding to the i-th time series sample and the feature vectors corresponding to all time series samples with the same label;

[0062] (c) Iterative optimization:

[0063] i. Initialize the candidate solution set {k1,k2,…,k nc}∈[b L ,b R ];

[0064] ii. According to k i When the objective function f(k) takes the value f(k i ) Sort candidate solutions from large to small, 1≤i≤nc, and select the k that makes the objective function f(k) take the largest, second largest, and third largest value in turn. i , as the optimal solution k α , suboptimal solution k β and the third optimal solution k δ ;

[0065] iii. Update the convergence coefficient and random vectors Where t is the current iteration number, is a vector whose modulus is randomly selected between [0,1], is a unit vector;

[0066] iv. Update candidate solution positions:

[0067]

[0068] in Representative vector The i-th item of k(t-1) is the value of the adjustable filter factor when the iteration number is t-1, k(t) is the value of the adjustable filter factor when the iteration number is t, k(0)=k α ;

[0069] v. Repeat the iteration until the maximum number of iterations is reached inter max , the optimal solution k for the adjustable filter factor of the output tracking differentiator adj =k(inter max ).

[0070] Furthermore, the feature evaluation indicators in step S3.1 include mutual information (MI) and analysis of variance (ANOVA).

[0071] Furthermore, the specific implementation steps of the expansion mapping in step S2.1 and step S2.2 are as follows:

[0072] (a) Bidirectional expansion mapping:

[0073] For the input time series s i =(s1,s2,…,s σ ) Perform global pattern enhancement according to the shuffle rate R, where R is an integer greater than 1. The specific operations include:

[0074] Forward downsampling: from s i =(s1,s2,…,s σ ) starts from the i-th data point, 1≤i≤R, and selects data points every R data points in turn to generate a subsequence s in the original order fdi ;

[0075] Reverse downsampling: from s i =(s1,s2,…,s σ ), starting from the i-th data point before the end of the sequence, 1≤i≤R, select data points in reverse order every R data points, and generate a subsequence s in reverse order. bdi ;

[0076] The forward downsampling and reverse downsampling operations are performed sequentially for i increasing from 1 to R, and s is obtained. fd1 ,s fd2 ,…,s fdR With s bd1 ,s bd2 ,…,s bdR ;

[0077] Will s fd1 ,s fd2 ,…,s fdR Splice in order to s fd , will s bdR ,…,s bd2 ,s bd1 Splice in order to s bd ;

[0078] (b) Bidirectional cross mapping:

[0079] i. Input time series s i =(s1,s2,…,s σ ) is evenly divided into R subsequences S1, S2, ..., S R , the length of each subsequence is or Indicates rounding down. Indicates rounding up to ensure that the length difference between subsequences does not exceed 1, and σ is the time series sample s i =(s1,s2,…,s σ ),

[0080] ii. Interleave the subsequences in forward order, and replace S1[j], S2[j], ..., S R [j] is spliced ​​into s in order fj ,S1[j],S2[j],…,S R [j] represent subsequences S1, S2, ..., S R The jth value of ;

[0081] Reverse cross-splicing: splice subsequences in reverse order, and R [j],…,S2[j],S1[j] are concatenated in order to form s bj

[0082] For j to increase from 1 to The forward cross-stitching and reverse cross-stitching operations are performed in sequence to obtain s f1 ,s f2 ,…,s fu With s b1 ,s b2 ,…,s bu , will s f1 ,s f2 ,…,s fu Splice in order to s f , will s bu ,…,s b2 ,s b1 Splice in order to s b ;

[0083] (c) Output a set of sequence sets {s fd ,s bd ,s f ,s b}, used for subsequent feature extraction.

[0084] Furthermore, the value range of the shuffling rate R is Where σ is the length of the time series sample, and the maximum value of the shuffle rate R is 16.

[0085] Furthermore, statistical features are extracted from the original sequence, the differential sequence, and the sequence after the original sequence and the differential sequence are expanded and mapped. The statistical features include mean, variance, kurtosis, skewness, and quantile features.

[0086] Furthermore, the classifier in step S4 includes a Ridge Regression classifier (RidgeCV i ) and the Extremely Randomized Trees Classifier (ExtraTrees i ).

[0087] Furthermore, the industrial equipment described in step S1 includes one or more general equipment, special equipment or special equipment in the fields of machinery, electrical electronics, computer networks, automobiles, chemicals or energy, and the equipment operating status information includes one or more of temperature, pressure, speed, torque, sound, vibration or electrical signals.

[0088] The present invention also provides an industrial equipment fault diagnosis system based on multi-view integrated features, characterized in that: the system has a program module corresponding to the above steps, and executes the steps in the above-mentioned industrial equipment fault diagnosis method based on multi-view integrated features during operation.

[0089] The present invention has the following beneficial technical effects:

[0090] Compared with the existing technology, the "industrial equipment fault diagnosis method based on multi-view integrated features" proposed in the present invention has the following significant advantages: 1. A breakthrough improvement in feature expression capabilities: Through expansion mapping, the signal is decomposed into multiple scales while retaining the original time series structure, and the global trend (such as the overall degradation process of the equipment) and local details (such as instantaneous fault pulses) are simultaneously enhanced, so that the feature set covers a more comprehensive fault mode; the present invention introduces a multi-view feature generation mechanism to generate a multi-view feature set with strong complementarity and low redundancy, which significantly improves the classification performance. 2. A dynamic adaptive feature optimization system: The present invention proposes a stability-diversity score-driven feature selector, which is based on the performance of features on different subsets (stability) and the mutual information (diversity) between features, and dynamically generates a robust multi-view feature subset through adaptive filters and set intersection operations; the present invention adopts feature interaction analysis based on the Pearson correlation coefficient to quantify the synergy and conflict relationship between features, avoiding the problem of mistaken deletion or redundant accumulation caused by ignoring feature correlation in traditional methods. 3. Performance advantages of heterogeneous ensemble classifiers: This invention integrates heterogeneous classifiers such as ridge regression (excels at linear patterns) and extreme random trees (capturing nonlinear relationships), outputs the final fault label through hard voting, and significantly improves the classification robustness under complex working conditions.

[0091] In summary, this invention addresses the key bottlenecks of traditional fault diagnosis methods in industrial fault diagnosis through three core technologies: multi-view feature enhancement, adaptive feature selection, and classifier heterogeneous integration. This provides an innovative solution for high-precision, low-cost equipment health management. This invention balances feature diversity, expressiveness, and model robustness while adapting to the stringent constraints of industrial scenarios. Its application significantly improves the accuracy and reliability of industrial equipment fault diagnosis, particularly when processing high-noise, small sample, or non-stationary operating data. BRIEF DESCRIPTION OF THE DRAWINGS

[0092] Figure 1 This is a flow chart of an industrial equipment fault diagnosis method based on multi-view integrated features in an embodiment of the present invention;

[0093] Figure 2 This is a flow chart for implementing the tracking differentiator function of optimizing the filter factor based on the optimization algorithm in an embodiment of the present invention;

[0094] Figure 3 Schematic diagram of the expansion mapping (bidirectional expansion mapping, bidirectional cross mapping) process in an embodiment of the present invention;

[0095] Figure 4 The figures are scatter plots comparing the accuracy of the present invention and the existing methods in the embodiments of the present invention, wherein (a) is a scatter plot comparing the accuracy of the present invention and the Catch22 method, (b) is a scatter plot comparing the accuracy of the present invention and the Signatures method, (c) is a scatter plot comparing the accuracy of the present invention and the TSFresh method, (d) is a scatter plot comparing the accuracy of the present invention and the TSFel method, (e) is a scatter plot comparing the accuracy of the present invention and the FreshP method, and (e) is a scatter plot comparing the accuracy of the present invention and the TD-MVDC method;

[0096] Figure 5 This is a critical difference diagram of accuracy comparing the method of the present invention with other existing methods. DETAILED DESCRIPTION

[0097] Combined with attachment Figure 1 -Attached Figure 5 The implementation of the industrial equipment fault diagnosis method based on multi-view integrated features described in the present invention is described as follows:

[0098] Specific implementation method 1: The flowchart of the industrial equipment fault diagnosis method based on multi-view integrated features of the present invention is as shown in the attached figure. Figure 1 As shown, the specific steps include:

[0099] Step S1: Obtain a time series data set of the fault status of industrial equipment, wherein the time series data set includes a sample set S of a unary time series.L =(s1,s2,…,s n ) and the label set L=(l1,l2,…,l n ) consists of a training set Train=(S L ,L) and the sample set S of the univariate time series T =(s n+1 ,s n+2 ,…,s n+m ) composed of a test set Test=(S T ), n and m are the number of time series samples included in the training set and the test set respectively, and the time series sample s i =(s1,s2,…,s σ ) are all univariate time series with no missing values, equal intervals, and equal lengths, 1≤i≤n+m, σ is the length of the time series sample;

[0100] The industrial equipment includes one or more general-purpose equipment, special-purpose equipment, or specialized equipment in the fields of machinery, electrical electronics, computer networks, automobiles, chemicals, or energy. The equipment operating status information includes one or more of temperature, pressure, speed, torque, sound, vibration, or electrical signals.

[0101] Step S2: extracting and enhancing features of the time series dataset, including:

[0102] Step S2.1: Extract statistical features from the original time series sample set S to obtain the feature vector FV R The original time series sample set S includes the training set unary time series sample set S L and the test set univariate time series sample set S T ;

[0103] Perform c groups of expansion mappings of different scales on the original time series sample set S to obtain the sample set S after expansion mapping s1 ,…,S sc , for the sample set S s1 ,…,S sc Perform statistical feature extraction to obtain the feature vector FV Rs1 ,…,FV Rsc ;

[0104] Step S2.2: Perform first-order difference transformation on the original time series sample set S to obtain the first-order difference time series sample set S D ;

[0105] For the first-order difference time series sample set S D Perform statistical feature extraction to obtain the feature vector FV D ;

[0106] For the first-order difference time series sample set S D Perform c groups of expansion mappings of different scales to obtain the sample set S after expansion mapping Ds1 ,…,S Dsc , for the sample set S Ds1 ,…,S Dsc Perform statistical feature extraction to obtain the feature vector FV Ds1 ,…,FV Dsc ;

[0107] The statistical features of the original sequence, the differential sequence, and the sequence after the original sequence and the differential sequence are expanded and mapped are extracted, and the statistical features include mean, variance, kurtosis, skewness, and quantile features;

[0108] The first-order difference transformation of the original time series sample set S can be realized by a tracking differentiator with an adjustable filter factor k. The implementation flow chart of the tracking differentiator with an adjustable filter factor is shown in the attached figure. Figure 2 As shown, the specific function implementation process is as follows:

[0109] The calculation formula for the first column x1[1] of the matrix x1 and the first column x2[1] of the matrix x2 is:

[0110]

[0111] Wherein, time series sample s is the time series sample of the input tracking differentiator in the original time series sample set S, s[i] is the i-th value in the time series sample s, 1≤i≤σ, where σ is the length of the time series sample s;

[0112] For i=1,2,…,σ increases successively:

[0113] y i =x1[i]-s[i]+khx2[i];

[0114] If |y i |>k 2 h 2 r, then otherwise,

[0115] If |a|>rhk, then fh i = -r·sign(a), otherwise,

[0116] The formula for calculating the i+1th column x1[i+1] of the matrix x1 and the i+1th column x2[i+1] of the matrix x2 is:

[0117]

[0118] The differential time series is transformed by a tracking differentiator with an adjustable filter factor to obtain the following differential time series: Wherein, r is a preset speed factor parameter, h is a preset step size parameter, and k is an adjustable filter factor;

[0119] The adjustable filter factor k of the tracking differentiator is iteratively optimized by an iterative optimization algorithm, and the iterative optimization process is:

[0120] (a) Set the optimization parameters: number of candidate solutions nc = 10, maximum number of iterations inter max =10, filter factor search range [b L ,b R ], where the search range [b L ,b R ]The default value is [1,3];

[0121] (b) Define the objective function f(k):

[0122] i. For the training set time series S L Apply filter factor k to generate differential sequence set

[0123] ii. Yes After sorting the rows in ascending order, the following sampling rate d rate =4 Extract quantile features

[0124] iii. Calculate feature set The silhouette score (S-score) of the label L is used as the objective function value f(k), and the filter factor k is iteratively optimized by maximizing the silhouette score (S-score);

[0125] The silhouette score (S-score) is defined as: Among them, n is the number of time series samples contained in the training set, SC(i) represents the feature set The silhouette coefficient of the eigenvector corresponding to the i-th time series sample in , Among them, η(i) represents the minimum distance between clusters, that is, the feature set The minimum value of the Euclidean distance between the feature vector corresponding to the i-th time series sample and the feature vectors corresponding to all time series samples with different labels, Represents the average distance within the cluster, that is, the feature set The average value of the Euclidean distance between the feature vector corresponding to the i-th time series sample and the feature vectors corresponding to all time series samples with the same label;

[0126] (c) Iterative optimization:

[0127] i. Initialize the candidate solution set {k1,k2,…,k nc}∈[b L ,b R ];

[0128] ii. According to k i When the objective function f(k) takes the value f(k i ) Sort candidate solutions from large to small, 1≤i≤nc, and select the k that makes the objective function f(k) take the largest, second largest, and third largest value in turn. i , as the optimal solution k α , suboptimal solution k β and the third optimal solution k δ ;

[0129] iii. Update the convergence coefficient and random vectors Where t is the current iteration number, is a vector whose modulus is randomly selected between [0,1], is a unit vector;

[0130] iv. Update candidate solution positions:

[0131]

[0132] in Representative vector The i-th item of k(t-1) is the value of the adjustable filter factor when the iteration number is t-1, k(t) is the value of the adjustable filter factor when the iteration number is t, k(0)=k α ;

[0133] v. Repeat the iteration until the maximum number of iterations is reached inter max , the optimal solution k for the adjustable filter factor of the output tracking differentiator adj =k(inter max );

[0134] The schematic diagram of the expansion mapping in step S2.1 and step S2.2 is shown in the attached figure. Figure 3 As shown, the specific implementation steps are:

[0135] (a) Bidirectional expansion mapping:

[0136] For the input time series s i =(s1,s2,…,s σ ) Perform global pattern enhancement according to the shuffle rate R, where R is an integer greater than 1. The specific operations include:

[0137] Forward downsampling: from s i =(s1,s2,…,s σ ) starts from the i-th data point, 1≤i≤R, and selects data points every R data points in turn to generate a subsequence s in the original order fdi ;

[0138] Reverse downsampling: from s i =(s1,s2,…,s σ ), starting from the i-th data point before the end of the sequence, 1≤i≤R, select data points in reverse order every R data points, and generate a subsequence s in reverse order. bdi ;

[0139] The forward downsampling and reverse downsampling operations are performed sequentially for i increasing from 1 to R, and s is obtained. fd1 ,s fd2 ,…,s fdR With s bd1 ,s bd2 ,…,s bdR ;

[0140] Will s fd1 ,s fd2 ,…,s fdR Splice in order to s fd , will s bdR ,…,s bd2 ,s bd1 Splice in order to s bd ;

[0141] (b) Bidirectional cross mapping:

[0142] i. Input time series s i =(s1,s2,…,s σ ) is evenly divided into R subsequences S1, S2, ..., S R , the length of each subsequence is or Indicates rounding down. Indicates rounding up to ensure that the length difference between subsequences does not exceed 1, and σ is the time series sample s i =(s1,s2,…,s σ ),

[0143] ii. Interleave the subsequences in forward order, and replace S1[j], S2[j], ..., S R [j] is spliced ​​into s in order fj ,S1[j],S2[j],…,S R [j] represent subsequences S1, S2, ..., S RThe jth value of ;

[0144] Reverse cross-splicing: splice subsequences in reverse order, and R [j],…,S2[j],S1[j] are concatenated in order to form s bj

[0145] For j to increase from 1 to The forward cross-stitching and reverse cross-stitching operations are performed in sequence to obtain s f1 ,s f2 ,…,s fu With s b1 ,s b2 ,…,s bu , will s f1 ,s f2 ,…,s fu Splice in order to s f , will s bu ,…,s b2 ,s b1 Splice in order to s b ;

[0146] (c) Output a set of sequence sets {s fd ,s bd ,s f ,s b}, used for subsequent feature extraction;

[0147] The value range of the shuffling rate R is Where σ is the length of the time series sample, and the maximum value of the shuffle rate R is 16;

[0148] Step S2.3: The feature vector FV R , FV Rs1 ,…,FV Rsc , FV D , FV Ds1 ,…,FV Dsc Splicing into a high-dimensional initial feature vector set FV;

[0149] Step S3, performing multi-view integrated feature selection on the initial feature vector set FV, including:

[0150] Step S3.1: Multi-metric scoring and initial view generation for high-dimensional feature sets:

[0151] Use M independent feature evaluation indicators (at least two independent feature evaluation indicators, such as mutual information MI and variance analysis ANOVA) to score each feature in the initial feature vector set FV. Each feature evaluation indicator generates a corresponding scoring matrix, and M scoring matrices are obtained.

[0152] According to the scoring matrix corresponding to the i-th feature evaluation index, for each feature, sort in descending order and select the top K1, K2, ..., K T high-scoring features, and obtain the initial view feature set FV corresponding to the i-th feature evaluation index 1i ,FV 2i ,…,FV Ti , i is a positive integer less than M;

[0153] The i increases from 1 to M in sequence, i = 1, 2, ..., M, and the initial view feature set corresponding to the first feature evaluation index is obtained in sequence The initial view feature set FV corresponding to the second feature evaluation index 12 ,FV 22 ,…,FV T2 , ..., the initial view feature set corresponding to the Mth feature evaluation index

[0154] The number of features K1, K2, ..., K T According to the preset convergence threshold list thresholds=[threshold1,threshold2,…,threshold T ] Determine and traverse all candidate feature quantities K (K = 1, 2, ..., N feature ), calculated using the formula Select the values ​​that satisfy rate(K)≤threshold i The minimum number of features is K i , 1≤i≤T, where rate(K) is the relative change rate of the stability-diversity score (SD-score), N feature is the number of features of the initial feature vector set FV;

[0155] For each candidate feature number K, according to the scoring matrix corresponding to each feature evaluation index j, the top K high-scoring features are selected from the initial feature vector set FV to obtain the feature subset The j increases from 1 to M in sequence, j = 1, 2, ..., M, to obtain the feature subsets corresponding to all feature evaluation indicators. Calculate the feature subset corresponding to the number of candidate features K The absolute value matrix of the Pearson correlation coefficient Each sub-block of the Pearson correlation coefficient absolute value matrix A(K) are all K×K dimensional square matrices, Representing feature subsets With feature subset The absolute value matrix of the Pearson correlation coefficient between 1≤p≤M, 1≤q≤M;

[0156] The stability-diversity score (SD-score) corresponding to the number of candidate features K is calculated by the formula Calculate, where Represents a pair of blocks Sum all elements of ;

[0157] The feature evaluation indicators include mutual information (MI) and analysis of variance (ANOVA);

[0158] Step S3.2: Generate an intersection-optimized extended view feature set:

[0159] For the initial view feature set FV corresponding to the first feature evaluation index 11 ,FV 21 ,…,FV T1 , the initial view feature set FV corresponding to the second feature evaluation index 12 ,FV 22 ,…,FV T2 , ..., the initial view feature set FV corresponding to the Mth feature evaluation index 1M ,FV 2M ,…,FV TM , respectively take the feature index intersection of the initial view feature set corresponding to different feature evaluation indicators, and generate T common feature subsets in turn in

[0160] The common feature subset Spliced ​​into a common feature vector FV interse , and the common eigenvector FV intersection The initial view feature set FV corresponding to all feature evaluation indicators 11 ,FV 21 ,…,FV T1 , FV 12 ,FV 22 ,…,FV T2 ,……,FV 1M ,FV 2M ,…,FV TM Splicing is performed to generate N = T·M extended view feature sets FV' V1 ,FV' V2 ,…,FV' VN ;

[0161] Step S3.3: De-redundancy processing, outputting the final multi-view feature set:

[0162] For each extended view feature set FV' V1,FV' V2 ,…,FV' VN The Pearson correlation coefficient analysis is performed on the features, and redundant features with correlation coefficients greater than the preset threshold are removed to obtain the final multi-view feature set FV'1, FV'2, ..., FV' N ;

[0163] Step S4: Heterogeneous integration classification and diagnosis:

[0164] Classifier training: using the training set Train = (S L ,L) Obtain the training set multi-view feature set through steps S2 and S3 Feature set in each view Train the classifier CL on i , 1≤i≤N;

[0165] The classifier includes a Ridge Regression classifier (RidgeCV i ) and the Extremely Randomized Trees Classifier (ExtraTrees i );

[0166] Hard voting integration: using the test set Test = (S T ) After steps S2 and S3, the test set multi-view feature set is obtained The feature set of each view Input classifier CL i , get the multi-view prediction result L i , 1≤i≤N;

[0167] The classification results of all views L1, L2, L, L are integrated using the majority voting rule N , output the final fault label L T ;

[0168] Step S5, result output: the diagnostic label L T The information is transmitted to the industrial equipment control terminal through the communication module, triggering fault warnings or maintenance instructions.

[0169] Specific embodiment 2: The present invention also provides an industrial equipment fault diagnosis system based on multi-view integrated features, characterized in that: the system has a program module corresponding to the above steps, and executes the steps in the above-mentioned industrial equipment fault diagnosis method based on multi-view integrated features during operation.

[0170] Specifically, it includes the following modules:

[0171] (a) Data acquisition and preprocessing module: used to obtain univariate time series data on the operating status of industrial equipment. The data includes a training set and a test set. The training set consists of standardized time series samples with equal intervals, equal lengths, and no missing values, and their corresponding discrete fault labels. The test set consists of unlabeled standardized time series samples.

[0172] The module is also used to standardize the time series data so that its mean is 0 and its variance is 1;

[0173] (b) Multi-view feature extraction module, including:

[0174] i. Adaptive parameter optimization unit: Based on the swarm intelligence optimization algorithm, with the silhouette score as the objective function, dynamically adjusts the filter factor k of the tracking differentiator adj , generate a set of differential sequences;

[0175] ii. Sequence Transformation Unit: Performs bidirectional expansion mapping and bidirectional cross mapping on the original sequence and the differential sequence. The bidirectional expansion mapping generates a global trend enhancement sequence through forward downsampling and backward downsampling, and the bidirectional cross mapping generates a local detail enhancement sequence by evenly splitting the sequence and interleaving it in forward and reverse order.

[0176] iii. Statistical feature extraction unit: uses a preset statistical feature library to extract mean, variance, and quantile features from the original sequence, differential sequence, and transformed sequence, and concatenates them into a high-dimensional initial feature vector set;

[0177] (c) Adaptive multi-view selection module, including:

[0178] i. Multi-metric scoring unit: uses feature evaluation indicators such as mutual information (MI) and analysis of variance (ANOVA) to independently score features and generate a scoring matrix;

[0179] ii. Intersection Optimization Unit: Adaptively determines the number of features based on the stability-diversity score (SD-score), generates a common feature subset by taking the index intersection of the features under each metric, and concatenates it with the initial view feature set to output a multi-view feature set.

[0180] (d) Heterogeneous integrated classification module, including:

[0181] i. Classifier training unit: trains a ridge regression classifier (RidgeCV) and an extreme random tree classifier (ExtraTrees) on each view feature set in the training set;

[0182] ii. Hard Voting Integration Unit: This unit inputs the multi-view features of the test set into the trained ridge regression classifier and extreme randomized tree classifier, obtains multi-view prediction results, performs majority voting (hard voting), and outputs the final fault diagnosis label.

[0183] (e) Result interaction module:

[0184] Used to transmit diagnostic tags to industrial equipment control terminals to trigger fault warnings or maintenance instructions.

[0185] It has been verified that the method proposed in the present invention solves the technical problem proposed in the present invention. The method of the present invention has been verified through practical application of the technical effect and practicality claimed by the present invention.

[0186] The algorithm (method) proposed in the present invention is the underlying technical core of the present invention, and various products can be derived based on the algorithm.

[0187] Based on the algorithm (method) proposed in the present invention, an industrial equipment fault diagnosis system based on multi-view integrated features is developed using a programming language. The system has program modules corresponding to the steps of the above-mentioned technical solution, and executes the steps of the above-mentioned industrial equipment fault diagnosis method based on multi-view integrated features during operation.

[0188] The developed system (software) computer program is stored on a computer-readable storage medium. The computer program is configured to implement the steps of the aforementioned method for diagnosing industrial equipment faults based on multi-view integration features when invoked by a processor. This materializes the present invention on a carrier, becoming a computer program product.

[0189] A device for diagnosing industrial equipment faults based on multi-view integrated features, the device comprising at least one processor and a memory communicatively connected to the at least one processor, wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the above-mentioned method for diagnosing industrial equipment faults based on multi-view integrated features, thereby realizing fault diagnosis of industrial equipment.

[0190] Various implementations of the systems and techniques described herein can be realized in digital electronic circuit systems, integrated circuit systems, dedicated ASICs (application specific integrated circuits), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system that includes at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0191] The computer programs (also referred to as programs, software, software applications, or code) of the present invention include machine instructions for a programmable processor and can be implemented using high-level procedural and / or object-oriented programming languages, and / or assembly / machine languages. As used herein, the terms "machine-readable medium" and "computer-readable medium" refer to any computer program product, device, and / or means (e.g., a magnetic disk, optical disk, memory, programmable logic device (PLD)) for providing machine instructions and / or data to a programmable processor, including machine-readable media that receive machine instructions as machine-readable signals. The term "machine-readable signal" refers to any signal for providing machine instructions and / or data to a programmable processor.

[0192] Example

[0193] In this embodiment, the existing methods compared with the method of the present invention include:

[0194] (1) TSFresh Pipeline with Rotation Forest Classifier (FreshP), which is derived from the reference M. Middlehurst, A. Bagnall, “The FreshPRINCE: A Simple Transformation Based Pipeline Time Series Classifier,” in Proceedings of International Conference on Pattern Recognition and Artificial Intelligence 2022 (ICPRAI 2022), pp. 150-161, May 2022;

[0195] (2) A feature extraction algorithm based on 22 canonical time-series characteristics (catch22), which is derived from C. H. Lubba, S. S. Sethi, P. Knaute, et al., “catch22: CAnonical Time-series C Haracteristics Selected through highly comparative time-series analysis,” Data Mining Knowledge Discovery, vol. 33, no. 6, pp. 1821–1852, Nov. 2019.

[0196] (3) Time series feature extraction on basis of scalable hypothesis tests method (TSFresh), which is derived from M. Christ, N. Braun, J. Neuffer, and AW Kempa-Liehr, “Time Series FeatuRe Extraction on basis of Scalable Hypothesis tests (tsfresh – A Python package),” Neurocomputing, vol. 307, pp. 72–77, Sep. 2018;

[0197] (4) A generalized signature method for multivariate time series feature extraction (abbreviated as: Signatures) The method is derived from the reference J.Morrill, A.Fermanian, P.Kidger, et al., “A generalized signature method for multivariate time series feature extraction,” arXiv, 2020, https: / / doi.org / 10.48550 / arXiv.2006.00873 ;

[0198] (5) A time series feature extraction library (TSFel), which is derived from the reference M.Barandas, D.Folgado, L.Fernandes, et al., “Tsfel: Time series feature extraction library,” SoftwareX, vol. 11, pp. 100456, Jan. 2020;

[0199] (6) Chinese invention patent CN118644702A discloses an industrial equipment fault diagnosis method based on multi-view expansion statistical features (abbreviated as: TD-MVDC).

[0200] To validate the effectiveness of our method, we conducted a computer simulation classification comparison experiment on 112 datasets from the UCR open-source dataset. Each dataset used the default training / testing set partitioning, and each sample was normalized. Classification accuracy was used as the performance metric for evaluation, and the Wilcoxon signed-rank test with a Holm correction at a p-value of 0.02 was used to measure the significance of the differences between the algorithms. The classification accuracy results are shown in Table 1, where the data in Table 1 are expressed in %.

[0201] Table 1

[0202]

[0203]

[0204]

[0205]

[0206]

[0207] The accuracy scatter plot of the method of the present invention compared with the existing methods on 112 UCR data sets is as follows: Figure 4 As shown, the points falling above / above / below the diagonal line respectively represent that the accuracy of the method of the present invention on the data set is worse than / the same as / better than the accuracy of the prior art method, wherein W represents the number of data sets whose classification accuracy of the method is better than that of the comparison method, T represents the number of data sets whose classification accuracy of the method is equal to that of the comparison method, and L represents the number of data sets whose classification accuracy of the method is worse than that of the comparison method.

[0208] From Table 1 and Figure 4The results show that the method of the present invention is not only more accurate than the existing method on most data sets, but also significantly more accurate on some data sets.

[0209] Figure 5 A plot showing the critical difference in accuracy between our method and prior art methods is shown, with methods with no statistically significant difference connected by a thick black line. The numbers represent the average rankings of our method and prior art methods calculated on 112 UCR datasets. Our method achieved the highest average ranking in terms of accuracy, demonstrating its superiority.

[0210] It should be understood that the various forms of the above-described processes may be used, and steps may be reordered, added, or deleted. For example, the steps described in this application may be performed in parallel, sequentially, or in a different order. As long as the desired results of the technical solutions disclosed in this application can be achieved, they are all within the scope of protection of the present invention.

Claims

1. A method for industrial equipment fault diagnosis based on multi-view integrated features, characterized in that: The following steps are involved: Step S1: Obtain a time series data set of the fault status of industrial equipment, wherein the time series data set includes a sample set S of a unary time series. L =(s1,s2,…,s n ) and the label set L=(l1,l2,…,l n ) consists of a training set Train=(S L ,L) and the sample set S of the univariate time series T =(s n+1 ,s n+2 ,…,s n+m ) composed of a test set Test=(S T ), n and m are the number of time series samples included in the training set and the test set respectively, and the time series sample s i =(s1,s2,…,s σ ) are all univariate time series with no missing values, equal intervals, and equal lengths, 1≤i≤n+m, and σ is the length of the time series sample; Step S2: extracting and enhancing features of the time series dataset, including: Step S2.1: Extract statistical features from the original time series sample set S to obtain the feature vector FV R The original time series sample set S includes the training set unary time series sample set S L and the test set univariate time series sample set S T ; Perform c groups of expansion mappings of different scales on the original time series sample set S to obtain the sample set S after expansion mapping s1 ,…,S sc , for the sample set S s1 ,…,S sc Perform statistical feature extraction to obtain the feature vector FV Rs1 ,…,FV Rsc ; Step S2.2: Perform first-order difference transformation on the original time series sample set S to obtain the first-order difference time series sample set S D ; For the first-order difference time series sample set S D Perform statistical feature extraction to obtain the feature vector FV D ; For the first-order difference time series sample set S D Perform c groups of expansion mappings of different scales to obtain the sample set S after expansion mapping Ds1 ,…,S Dsc , for the sample set S Ds1 ,…,S Dsc Perform statistical feature extraction to obtain the feature vector FV Ds1 ,…,FV Dsc ; Step S2.3: The feature vector FV R , FV Rs1 ,…,FV Rsc , FV D , FV Ds1 ,…,FV Dsc Splicing into a high-dimensional initial feature vector set FV; Step S3, performing multi-view integrated feature selection on the initial feature vector set FV, including: Step S3.1: Multi-metric scoring and initial view generation for high-dimensional feature sets: Use M independent feature evaluation indicators to score each feature in the initial feature vector set FV. Each feature evaluation indicator generates a corresponding scoring matrix, and M scoring matrices are obtained. According to the scoring matrix corresponding to the i-th feature evaluation index, for each feature, sort in descending order and select the top K1, K2, ..., K T high-scoring features, and obtain the initial view feature set FV corresponding to the i-th feature evaluation index 1i ,FV 2i ,…,FV Ti , i is a positive integer less than M; The i increases from 1 to M in sequence, i = 1, 2, ..., M, and the initial view feature set FV corresponding to the first feature evaluation index is obtained in sequence 11 ,FV 21 ,…,FV T1 , the initial view feature set FV corresponding to the second feature evaluation index 12 ,FV 22 ,…,FV T2 , ..., the initial view feature set FV corresponding to the Mth feature evaluation index 1M ,FV 2M ,…,FV TM ; The number of features K1, K2, ..., K T According to the preset convergence threshold list thresholds=[threshold1,threshold2,…,threshold T ] Determine and traverse all candidate feature quantities K (K = 1, 2, ..., N feature ), calculated using the formula Select the values ​​that satisfy rate(K)≤threshold i The minimum number of features is K i , 1≤i≤T, where rate(K) is the relative change rate of the stability-diversity score (SD-score), N feature is the number of features of the initial feature vector set FV; For each candidate feature number K, according to the scoring matrix corresponding to each feature evaluation index j, the top K high-scoring features are selected from the initial feature vector set FV to obtain the feature subset The j increases from 1 to M in sequence, j = 1, 2, ..., M, to obtain the feature subsets corresponding to all feature evaluation indicators. Calculate the feature subset corresponding to the number of candidate features K The absolute value matrix of the Pearson correlation coefficient Each sub-block of the Pearson correlation coefficient absolute value matrix A(K) are all K×K dimensional square matrices, Representing feature subsets With feature subset The absolute value matrix of the Pearson correlation coefficient between The stability-diversity score (SD-score) corresponding to the number of candidate features K is obtained by the formula Calculate, where Represents a pair of blocks Sum all elements of ; Step S3.2: Generate an intersection-optimized extended view feature set: For the initial view feature set FV corresponding to the first feature evaluation index 11 ,FV 21 ,…,FV T1 , the initial view feature set FV corresponding to the second feature evaluation index 12 ,FV 22 ,…,FV T2 , ..., the initial view feature set FV corresponding to the Mth feature evaluation index 1M ,FV 2M ,…,FV TM , respectively take the feature index intersection of the initial view feature set corresponding to different feature evaluation indicators, and generate T common feature subsets in turn in The common feature subset Spliced ​​into a common feature vector FV intersection , and the common eigenvector FV intersection The initial view feature set FV corresponding to all feature evaluation indicators 11 ,FV 21 ,…,FV T1 , FV 12 ,FV 22 ,…,FV T2 ,……,FV 1M ,FV 2M ,…,FV TM Splice and generate N=T·M extended view feature sets FV' V1 ,FV' V2 ,…,FV' VN ; Step S3.3: De-redundancy processing, outputting the final multi-view feature set: For each extended view feature set FV' V1 ,FV' V2 ,…,FV' VN The Pearson correlation coefficient analysis is performed on the features, and redundant features with correlation coefficients greater than the preset threshold are removed to obtain the final multi-view feature set FV'1, FV'2, ..., FV' N ; Step S4: Heterogeneous integration classification and diagnosis: Classifier training: using the training set Train = (S L ,L) Obtain the training set multi-view feature set through steps S2 and S3 In each view feature set FV i L Train the classifier CL on i , 1≤i≤N; Hard voting integration: using the test set Test = (S T ) After steps S2 and S3, the test set multi-view feature set is obtained The feature set FV of each view i T Input classifier CL i , get the multi-view prediction result L i , 1≤i≤N; The classification results of all views L1, L2, ..., L are integrated using the majority voting rule. N , output the final fault label L T ; Step S5, result output: the diagnostic label L T The information is transmitted to the industrial equipment control terminal through the communication module, triggering fault warnings or maintenance instructions.

2. The method for industrial equipment fault diagnosis based on multi-view integrated features according to claim 1 is characterized in that: The first-order difference transformation of the original time series sample set S in step S2.2 can be implemented by a tracking differentiator with an adjustable filter factor k. The functional implementation process of the tracking differentiator with an adjustable filter factor is as follows: The calculation formula for the first column x1[1] of the matrix x1 and the first column x2[1] of the matrix x2 is: Wherein, time series sample s is the time series sample of the input tracking differentiator in the original time series sample set S, s[i] is the i-th value in the time series sample s, 1≤i≤σ, where σ is the length of the time series sample s; For i=1,2,…,σ increases successively: y i =x1[i]-s[i]+xx2[i]; If |y i |>k 2 h 2 r, then otherwise, If |a|>rhk, then fh i = -r·sign(a), otherwise, The formula for calculating the i+1th column x1[i+1] of the matrix x1 and the i+1th column x2[i+1] of the matrix x2 is: The differential time series is transformed by a tracking differentiator with an adjustable filter factor to obtain the following differential time series: Wherein, r is a preset speed factor parameter, h is a preset step size parameter, and k is an adjustable filter factor.

3. The method for industrial equipment fault diagnosis based on multi-view integrated features according to claim 2 is characterized in that: The iterative optimization process of the adjustable filter factor k of the tracking differentiator is: (a) Set the optimization parameters: number of candidate solutions nc = 10, maximum number of iterations inter max =10, filter factor search range [b L ,b R ], where the search range [b L ,b R ]The default value is [1,3]; (b) Define the objective function f(k): i. For the training set time series S L Apply filter factor k to generate differential sequence set ii. Yes After sorting the rows in ascending order, the following sampling rate d rate =4 Extract quantile features iii. Calculate feature set The silhouette score (S-score) of the label L is used as the objective function value f(k), and the filter factor k is iteratively optimized by maximizing the silhouette score (S-score); The silhouette score (S-score) is defined as: Among them, n is the number of time series samples contained in the training set, SC(i) represents the feature set The silhouette coefficient of the eigenvector corresponding to the i-th time series sample in , Among them, η(i) represents the minimum distance between clusters, that is, the feature set The minimum value of the Euclidean distance between the feature vector corresponding to the i-th time series sample and the feature vectors corresponding to all time series samples with different labels, θ(i) represents the average distance within the cluster, that is, the feature set The average value of the Euclidean distance between the feature vector corresponding to the i-th time series sample and the feature vectors corresponding to all time series samples with the same label; (c) Iterative optimization: i. Initialize the candidate solution set {k1,k2,…,k nc }∈[b L ,b R ]; ii. According to k i When the objective function f(k) takes the value f(k i ) Sort candidate solutions from large to small, 1≤i≤nc, and select the k that makes the objective function f(k) take the largest, second largest, and third largest value in turn. i , as the optimal solution k α , suboptimal solution k β and the third optimal solution k δ ; iii. Update the convergence coefficient and random vectors Where t is the current iteration number, is a vector whose modulus is randomly selected between [0,1], is a unit vector; iv. Update candidate solution positions: in Representative vector The i-th item of k(t-1) is the value of the adjustable filter factor when the iteration number is t-1, k(t) is the value of the adjustable filter factor when the iteration number is t, k(0)=k α ; v. Repeat the iteration until the maximum number of iterations is reached inter max , the optimal solution k for the adjustable filter factor of the output tracking differentiator adj =k(inter max ).

4. The method for industrial equipment fault diagnosis based on multi-view integrated features according to claim 3 is characterized in that: The feature evaluation indicators in step S3.1 include mutual information (MI) and analysis of variance (ANOVA).

5. The method for industrial equipment fault diagnosis based on multi-view integrated features according to claim 4 is characterized in that: The specific implementation steps of the expansion mapping in step S2.1 and step S2.2 are: (a) Bidirectional expansion mapping: For the input time series s i =(s1,s2,…,s σ ) Perform global pattern enhancement according to the shuffle rate R, where R is an integer greater than 1. The specific operations include: Forward downsampling: from s i =(s1,s2,…,s σ ) starts from the i-th data point, 1≤i≤R, and selects data points every R data points in turn to generate a subsequence s in the original order fdi ; Reverse downsampling: from s i =(s1,s2,…,s σ ), starting from the i-th data point before the end of the sequence, 1≤i≤R, select data points in reverse order every R data points, and generate a subsequence s in reverse order. bdi ; The forward downsampling and reverse downsampling operations are performed in sequence for i increasing from 1 to R, and s is obtained. fd1 ,s fd2 ,…,s fdR With s bd1 ,s bd2 ,…,s bdR ; Will s fd1 ,s fd2 ,…,s fdR Splice in order to s fd , will s bdR ,…,s bd2 ,s bd1 Splice in order to s bd ; (b) Bidirectional cross mapping: i. Input time series s i =(s1,s2,…,s σ ) is evenly divided into R subsequences S1, S2, ..., S R , the length of each subsequence is or Indicates rounding down. Indicates rounding up to ensure that the length difference between subsequences does not exceed 1, and σ is the time series sample s i =(s1,s2,…,s σ ), ii. Interleave the subsequences in forward order, and replace S1[j], S2[j], ..., S R [j] is spliced ​​into s in order fj ,S1[j],S2[j],…,S R [j] represent subsequences S1, S2, ..., S R The jth value of ; Reverse cross-splicing: splice subsequences in reverse order, and R [j],…,S2[j],S1[j] are concatenated in order to form s bj For j to increase from 1 to The forward cross-stitching and reverse cross-stitching operations are performed in sequence to obtain s f1 ,s f2 ,…,s fu With s b1 ,s b2 ,…,s bu , will s f1 ,s f2 ,…,s fu Splice in order to s f , will s bu ,…,s b2 ,s b1 Splice in order to s b ; (c) Output a set of sequence sets {s fd ,s bd ,s f ,s b }, used for subsequent feature extraction.

6. The method for industrial equipment fault diagnosis based on multi-view integrated features according to claim 5, characterized in that: The value range of the shuffling rate R is Where σ is the length of the time series sample, and the maximum value of the shuffle rate R is 16.

7. The method for industrial equipment fault diagnosis based on multi-view integrated features according to claim 6, characterized in that: Statistical features are extracted from the original sequence, the differential sequence, and the sequence after the original sequence and the differential sequence are expanded and mapped. The statistical features include mean, variance, kurtosis, skewness, and quantile features; 8. The method for industrial equipment fault diagnosis based on multi-view integrated features according to claim 7, characterized in that: The classifier in step S4 includes a Ridge Regression classifier (RidgeCV i ) and the Extremely Randomized Trees Classifier (ExtraTrees i ).

9. The method for industrial equipment fault diagnosis based on multi-view integrated features according to claim 8, characterized in that: The industrial equipment described in step S1 includes one or more general equipment, special equipment or special equipment in the fields of machinery, electrical electronics, computer networks, automobiles, chemicals or energy, and the equipment operating status information includes one or more of temperature, pressure, speed, torque, sound, vibration or electrical signals.

10. An industrial equipment fault diagnosis system based on multi-view integrated features, characterized by: The system has a program module corresponding to the steps of any one of claims 1 to 9, and executes the steps of the above-mentioned industrial equipment fault diagnosis method based on multi-view integrated features when running.

Images