Failure prediction method and system for factory manufacturing equipment based on data driving

Abstract

The invention relates to the technical field of intelligent manufacturing, in particular to a fault prediction method and system for factory manufacturing equipment based on data driving, and the method comprises the steps: collecting vibration signal data and tool wear data, carrying out the preprocessing of the collected data, carrying out the time-frequency domain feature extraction of the collected vibration signal data, and extracting the wear features of a tool; fusing the extracted tool wear features, vibration signal data features and cutting parameter features to form a high-dimensional feature vector, selecting first k feature vectors through principal component analysis (PCA), and reducing the dimension of the high-dimensional feature vector; constructing and training a numerical control machine tool fault prediction model based on a support vector machine SVM model; and performing fault prediction on the numerical control machine tool, namely factory manufacturing equipment. Potential faults of the numerical control machine tool can be predicted, maintenance and overhaul can be arranged in advance, the equipment reliability is improved, the accidental shutdown time caused by the faults is shortened, and the production continuity is guaranteed.

Description

Technical Field

[0001] The present invention relates to the field of intelligent manufacturing technology, and in particular to a data-driven fault prediction method and system for factory manufacturing equipment. Background Art

[0002] With the development of Industry 4.0 and intelligent manufacturing, the fault prediction and health management of factory manufacturing equipment have received more and more attention. As key equipment in modern manufacturing, the reliability and stability of CNC machine tools directly affect production efficiency and product quality. Traditional CNC machine tool fault prediction mainly relies on manual experience and regular maintenance, which makes it difficult to achieve accurate prediction and timely maintenance.

[0003] The existing data-driven fault prediction methods still have limitations in feature extraction, data fusion, model optimization, etc. Some methods only use a single type of data, such as vibration signal data or tool wear data, which makes it difficult to fully characterize the health status of machine tools; some methods do not fully consider the heterogeneity and high dimensionality of data, resulting in poor feature extraction and fusion effects; the prediction models constructed by some methods have weak generalization capabilities and are difficult to adapt to the complex and changeable working conditions of industrial sites.

[0004] Most existing fault prediction systems run on PCs or industrial servers, and real-time performance and reliability are difficult to guarantee. Deploying prediction models to the embedded control systems of machine tools to achieve online real-time prediction and decision-making is the key to improving the practicality of the system, but there are few studies on this aspect.

[0005] There is relatively little research on fault prediction using multi-source data such as vibration signal data and tool wear of CNC machine tools; how to organically integrate these heterogeneous data and fully mine the fault-related information is an urgent problem to be solved. Summary of the invention

[0006] The purpose of the present invention is to address the deficiencies of the prior art and to propose a data-driven fault prediction system and method for factory manufacturing equipment.

[0007] To achieve the above object, the present invention provides the following technical solution: a data-driven method for predicting a fault of factory manufacturing equipment, characterized in that it comprises:

[0008] Collecting vibration signal data and tool wear data of factory manufacturing equipment data, and preprocessing the collected data, wherein the factory manufacturing equipment is a CNC machine tool, and the tool is a cutting tool on the CNC machine tool;

[0009] Extract the time-frequency domain features of the collected vibration signal data, calculate the time-domain features of the vibration signal data, and calculate the frequency-domain features of the vibration signal data through Fourier transform; extract the tool wear features;

[0010] The extracted tool wear features, vibration signal data features and cutting parameter features are integrated to form a high-dimensional feature vector. The first k feature vectors are selected through principal component analysis (PCA) to reduce the dimension of the high-dimensional feature vector.

[0011] The feature vector and the corresponding CNC machine tool working status constitute a training data set, and the CNC machine tool fault prediction model is constructed and trained based on the support vector machine SVM model;

[0012] The trained CNC machine tool fault prediction model is deployed to the embedded control system of the CNC machine tool to predict faults of the CNC machine tool, i.e., factory manufacturing equipment.

[0013] Preferably, a three-axis acceleration sensor is arranged on the CNC machine tool to collect vibration signal data of the CNC machine tool on the X, Y, and Z axes, and the collected X, Y, and Z axes are combined into comprehensive vibration signal data, which is recorded as vibration signal data x(t), where t is a time variable, indicating the sampling time;

[0014] The collected tool wear is expressed as: w(t)=w0+Δw(t); where w0 is the initial tool wear, Δw(t) is the wear increment relative to the initial state at time t, and represents the change process of tool wear over time. The initial tool wear is a constant, and the wear increment is a function of time.

[0015] The collected vibration signal data x(t) is filtered and a Butterworth low-pass filter and a high-pass filter are used to remove high-frequency noise and low-frequency interference;

[0016] Butterworth low-pass filter transfer function:

[0017] where ω c is the cut-off frequency, N l is the filter order, s is the complex frequency;

[0018] Butterworth high-pass filter transfer function:

[0019] where ω c is the cut-off frequency, N l is the filter order; s is the complex frequency, and the filtered CNC machine tool vibration signal data is recorded as

[0020] Preferably, extract the filtered CNC machine tool vibration signal data The time-frequency domain features and the tool wear features are extracted; the time-frequency domain features include time domain features and frequency domain features; the time domain features include mean μ, root mean square value RMS and peak value x p; The frequency domain feature is obtained by Perform fast Fourier transform to obtain the spectrum X(f) and calculate the energy E of each frequency band k , characterizes the energy distribution of vibration signal data in different frequency bands;

[0021] Extracting vibration signal data The frequency domain characteristics of Perform fast Fourier transform to get the spectrum X(f):

[0022]

[0023] Calculate the energy E of each frequency band k :

[0024]

[0025] Among them, f p is the starting frequency of the pth frequency band;

[0026] Extract tool wear characteristics, including tool wear amount w(t): w(t) = w0 + Δw(t); tool wear rate v w (t):v w =Δw(t) / Δt; tool wear acceleration a w (t): a w (t) = Δv w (t) / Δt; where Δt is the time interval, Δv w (t) is the tool wear rate increment.

[0027] Preferably, the extracted time-frequency domain features of the vibration signal data and the tool wear features constitute a feature vector X:

[0028] X = [μ, RMS, x p , X(f), E K ,w(t),v w (t), a w (t)] T

[0029] Normalize the eigenvector X and scale the eigenvalues ​​to the range [0, 1]:

[0030]

[0031] Among them, x i is the i-th original eigenvalue in the eigenvector X, x max and x min are the maximum and minimum values ​​of the current feature respectively;

[0032] Principal component analysis (PCA) is used to reduce the dimension of the normalized feature vector. The steps are as follows:

[0033] Compute the mean vector of the eigenvectors:

[0034] Compute the covariance matrix of the eigenvectors:

[0035] Where n represents the number of training samples, and the covariance matrix CX is decomposed to obtain the eigenvalues ​​λ1, λ2, ..., λ m and the corresponding eigenvectors u1, u2, ..., u m , m represents the dimension of the original high-dimensional feature space;

[0036] Select the eigenvectors corresponding to the first k largest eigenvalues ​​to form the transformation matrix: W = [u1, u2, ..., u k ], k<8;

[0037] Use the transformation matrix to perform linear transformation on the normalized eigenvector to map it from high-dimensional space to low-dimensional space; reduce the dimension of the normalized eigenvector to obtain the reduced-dimensional eigenvector Y: Y = W T (X-μ X ), W T Represents the transpose of a transformation matrix.

[0038] Preferably, the reduced-dimensional feature vector Y and the corresponding machine tool working state label L constitute a training data set: where Y i represents the i-th eigenvector, L i represents the label corresponding to the i-th eigenvector, L i ∈{-1, +1}, +1 indicates the normal state of the machine tool, and -1 indicates the fault state of the machine tool;

[0039] The fault prediction model is constructed based on the support vector machine (SVM). The soft margin SVM model is adopted, and the slack variable ξi is introduced. The optimization problem of SVM is written as:

[0040]

[0041] sL i (w T y i +b)≥1-ξ i ,ξ i ≥0, i=1,2,…,n

[0042] Among them, C represents the penalty factor, b represents the bias term, w represents the weight vector, and w T represents the transpose of the weight vector, ξ irepresents the i-th slack variable;

[0043] The original optimization problem is transformed into a dual problem using the Lagrange multiplier method, and the Lagrange multiplier α is introduced. i :

[0044]

[0045] Among them, K(Y i , Y j ) represents the kernel function, L i and L j Represent the true labels of the i-th and j-th samples respectively, α i represents the i-th inequality constraint, α j represents the jth inequality constraint;

[0046] The optimization goal of the dual problem is to maximize α i The difference between the sum of and a quadratic term, subject to the constraint α i The value range and α i With label L i The inner product of is 0.

[0047] Preferably, solve the transformed dual problem and obtain α i The optimal solution Calculate the weight vector of the SVM model and bias

[0048]

[0049]

[0050] Among them, S is the subscript set of support vectors, N s is the number of support vectors, is the optimal solution to the j-th dual problem, is the weight of the i-th sample when constructing the optimal weight vector;

[0051] Retrain the SVM model using the optimal parameters to obtain the final fault prediction model:

[0052]

[0053] Among them, f(y) represents the prediction function of the SVM model, sign(·) is a binary function, and the output result is -1 or +1, indicating the normal or faulty state of the machine tool.

[0054] Preferably, the trained SVM model is converted into a format suitable for running on an embedded system, including C / C++ code, and the CNC machine tool fault prediction model is deployed into an embedded control system of the CNC machine tool, including a PLC and an ARM system;

[0055] The vibration signal data and tool wear data of the machine tool are collected in real time and input into the prediction model for fault prediction. According to the prediction results, when the output result of the CNC machine tool fault prediction model is +1, the CNC machine tool will not fail in the future time period; when the output result of the CNC machine tool fault prediction model is -1, the CNC machine tool will fail in the future time period and the machine tool needs to be repaired.

[0056] A data-driven factory manufacturing equipment fault prediction system, which is used to implement a data-driven factory manufacturing equipment fault prediction method, including: a data acquisition module, a feature extraction module, a feature fusion module, a fault prediction module and a model deployment module;

[0057] The data acquisition module is used to collect vibration signal data and tool wear data of factory manufacturing equipment data and pre-process the collected data. The factory manufacturing equipment is a CNC machine tool, and the tool is a cutting tool on the CNC machine tool;

[0058] The feature extraction module is used to extract the time-frequency domain features of the collected vibration signal data, calculate the time-domain features of the vibration signal data, and calculate the frequency-domain features of the vibration signal data through Fourier transform; extract the tool wear features;

[0059] The feature fusion module is used to fuse the extracted tool wear features, vibration signal data features and cutting parameter features to form a high-dimensional feature vector, and select the first k feature vectors through principal component analysis PCA to reduce the dimension of the high-dimensional feature vector;

[0060] The fault prediction module is used to form a training data set with the feature vector and the corresponding CNC machine tool working state, and to build and train a CNC machine tool fault prediction model based on a support vector machine (SVM) model;

[0061] The model deployment module is used to deploy the trained CNC machine tool fault prediction model into the embedded control system of the CNC machine tool to perform fault prediction on the CNC machine tool, i.e., factory manufacturing equipment.

[0062] An electronic device comprises: a processor and a memory, wherein the memory stores a computer program that can be called by the processor; the processor executes the data-driven factory manufacturing equipment fault prediction method by calling the computer program stored in the memory.

[0063] A computer-readable storage medium stores instructions, which, when executed on a computer, enable the computer to execute the data-driven method for predicting faults of factory manufacturing equipment.

[0064] Beneficial effects of the present invention: The present invention comprehensively utilizes multi-source data such as vibration signal data of CNC machine tools and tool wear, and through time-frequency domain feature extraction and data fusion, comprehensively characterizes the health status of machine tools and improves the accuracy of fault prediction.

[0065] The present invention uses principal component analysis to reduce the dimensionality of high-dimensional feature vectors, selects the first k feature subsets, reduces the data dimension while retaining the main information, and reduces the computational overhead of subsequent model training and prediction.

[0066] The present invention constructs a fault prediction model based on a support vector machine, obtains optimal model parameters through training, and forms a high-precision and highly robust prediction model that can adapt to complex and changeable industrial scenarios.

[0067] The present invention deploys the trained fault prediction model into the embedded control system of the CNC machine tool, realizes the real-time collection, analysis and prediction of the machine tool operation status, timely discovers and warns of potential faults, and provides a basis for maintenance decisions.

[0068] The present invention can predict potential failures of CNC machine tools, arrange maintenance and overhaul in advance, improve equipment reliability, reduce unexpected downtime caused by failures, and ensure production continuity. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] Figure 1 This is a flow chart of a method for predicting a fault of factory manufacturing equipment based on data-driven according to the present invention;

[0070] Figure 2 This is a structural diagram of a fault prediction system for factory manufacturing equipment based on data-driven according to the present invention;

[0071] Figure 3 It is a schematic diagram of the structure of an electronic device provided by the present invention;

[0072] Figure 4 It is a schematic diagram of the structure of the computer-readable storage medium provided by the present invention. DETAILED DESCRIPTION

[0073] In order to better understand the present application, a more detailed description will be made of various aspects of the present application with reference to the accompanying drawings. It should be understood that these detailed descriptions are only descriptions of exemplary embodiments of the present application, and are not intended to limit the scope of the present application in any way. Throughout the specification, the same reference numerals refer to the same elements. The expression "and / or" includes any and all combinations of one or more of the associated listed items.

[0074] In the accompanying drawings, the size, dimensions and shapes of the elements have been slightly adjusted for ease of illustration. The drawings are for illustration only and are not strictly drawn to scale. As used herein, the terms "substantially", "approximately" and similar terms are used as terms of approximation, not as terms of degree, and are intended to illustrate the inherent deviations in measurements or calculations that will be recognized by those of ordinary skill in the art. In addition, in this application, the order in which the steps are described does not necessarily represent the order in which these processes occur in actual operation, unless otherwise specified or can be derived from the context.

[0075] It should also be understood that expressions such as "include", "including", "have", "contain" and / or "comprising" are open rather than closed expressions in this specification, which indicate the presence of the stated features, elements and / or components, but do not exclude the presence of one or more other features, elements, components and / or combinations thereof. In addition, when expressions such as "at least one of..." appear after a list of listed features, they modify the entire list of features rather than just the individual elements in the list. In addition, when describing embodiments of the present application, "may" is used to mean "one or more embodiments of the present application". And, the term "exemplary" is intended to refer to an example or illustration.

[0076] Unless otherwise specified, all words (including engineering terms and scientific and technological terms) used in this article have the same meaning as those commonly understood by ordinary technicians in the field to which this application belongs. It should also be understood that, unless clearly stated in this application, words defined in common dictionaries should be interpreted as having the same meaning as their meaning in the context of the relevant technology, and should not be interpreted in an idealized or overly formal sense.

[0077] It should be noted that, in the absence of conflict, the embodiments and features of the embodiments in the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0078] Example 1

[0079] Reference Figure 1 The first embodiment of the present invention provides a data-driven method for predicting faults in factory manufacturing equipment.

[0080] Step 1: Collect vibration signal data and tool wear data of factory manufacturing equipment data, and pre-process the collected data. The factory manufacturing equipment is a CNC machine tool, and the tool is a cutting tool on the CNC machine tool.

[0081] A three-axis acceleration sensor is arranged on the CNC machine tool to collect the vibration signal data of the CNC machine tool on the X, Y, and Z axes. The collected X, Y, and Z axes are combined into comprehensive vibration signal data, which is recorded as vibration signal data x(t), where t is a time variable, indicating the sampling time; according to the Nyquist sampling theorem, the acceleration sensor sampling frequency f s Should meet: f s ≥2f max ; where f max is the highest frequency component of the vibration signal data; that is, in order to avoid frequency aliasing, the sampling frequency should be at least twice the highest frequency of the vibration signal data.

[0082] A micron-level displacement sensor is used to measure tool wear on CNC machine tools. After each cutting of the CNC machine tool, the tool wear is measured at a fixed position on the workpiece surface. When measuring tool wear, the average value of multiple results is statistically calculated to reduce random errors.

[0083] The collected tool wear is expressed as: w(t)=w0+Δw(t); wherein w0 is the initial tool wear, Δw(t) is the wear increment relative to the initial state at time t, indicating the change process of tool wear over time, the initial tool wear is a constant, and the wear increment is a function of time.

[0084] Random errors can be reduced by taking multiple measurements and taking the average: where n c is the number of measurements, w i is the wear amount measured for the i-th time.

[0085] The collected vibration signal data x(t) is filtered and Butterworth low-pass filter and high-pass filter are used to remove high-frequency noise and low-frequency interference.

[0086] Butterworth low-pass filter transfer function:

[0087] where ω c is the cut-off frequency, N l is the filter order, s is the complex frequency; the transfer function of the Butterworth low-pass filter describes the frequency response characteristics of the Butterworth low-pass filter, which can effectively filter out noise components higher than the cut-off frequency.

[0088] Butterworth high-pass filter transfer function:

[0089] where ω c is the cut-off frequency, N l is the filter order; s is the complex frequency, and the filtered CNC machine tool vibration signal data is recorded as

[0090] The transfer function of the Butterworth high-pass filter describes the frequency response characteristics of the Butterworth high-pass filter, which can effectively filter out interference components below the cutoff frequency.

[0091] Step 2: Extract the time-frequency domain features of the collected vibration signal data, calculate the time-domain features of the vibration signal data, and calculate the frequency-domain features of the vibration signal data through Fourier transform; extract the tool wear features.

[0092] Extract filtered CNC machine tool vibration signal data The time-frequency domain features and the tool wear features are extracted; the time-frequency domain features include time domain features and frequency domain features; the time domain features include mean μ, root mean square value RMS and peak value x p ; The frequency domain feature is obtained by Perform fast Fourier transform to obtain the spectrum X(f) and calculate the energy E of each frequency band k , characterizing the energy distribution of vibration signal data in different frequency bands.

[0093] Mean: RMS value: Peak:

[0094] Among them, N1 is Signal time length, where is the amplitude of the vibration signal data at the tth sampling time point.

[0095] Extracting vibration signal data The frequency domain characteristics of Perform fast Fourier transform to get the spectrum X(f):

[0096]

[0097] Calculate the energy E of each frequency band k :

[0098]

[0099] Among them, f p is the starting frequency of the pth frequency band.

[0100] Extract tool wear characteristics, including tool wear amount w(t): w(t) = w0 + Δw(t); tool wear rate v w (t):v w =Δw(t) / Δt; tool wear acceleration a w (t): a w (t) = Δv w (t) / Δt.

[0101] Where Δw(t) is the wear change, Δt is the time interval, and Δv w (t) is the tool wear rate increment; the wear rate reflects the rate of change of the wear amount, and the wear acceleration reflects the rate of change of the wear rate, reflecting the dynamic characteristics of the wear process.

[0102] The extracted vibration signal data time-frequency domain features and tool wear features constitute the feature vector X:

[0103] X = [μ, RMS, x p ,X(f),E K ,w(t),v w (t), a w (t)] T

[0104] Normalize the eigenvector X and scale the eigenvalues ​​to the range [0, 1]:

[0105]

[0106] Among them, x i is the i-th original eigenvalue in the eigenvector X, x max and x min are the maximum and minimum values ​​of the current feature respectively.

[0107] Step 3: The extracted tool wear features, vibration signal data features and cutting parameter features are integrated to form a high-dimensional feature vector. The first k feature vectors are selected through principal component analysis (PCA) to reduce the dimension of the high-dimensional feature vector.

[0108] Principal component analysis (PCA) is used to reduce the dimension of the normalized feature vector. The steps are as follows:

[0109] Compute the mean vector of the eigenvectors:

[0110] Compute the covariance matrix of the eigenvectors:

[0111] Where n represents the number of training samples, and the covariance matrix C X Perform eigendecomposition and obtain eigenvalues ​​λ1, λ2, ..., λ m and the corresponding eigenvectors u1, u2, ..., u m , m represents the dimension of the original high-dimensional feature space;

[0112] Select the eigenvectors corresponding to the first k largest eigenvalues ​​to form the transformation matrix: W = [u1, u2, ..., u k ], k<8;

[0113] Use the transformation matrix to perform linear transformation on the normalized eigenvector to map it from high-dimensional space to low-dimensional space; reduce the dimension of the normalized eigenvector to obtain the reduced-dimensional eigenvector Y: Y = W T (X-μ X ), where W T Represents the transpose of a transformation matrix.

[0114] Step 4: The feature vector and the corresponding CNC machine tool working status constitute a training data set, and the CNC machine tool fault prediction model is constructed and trained based on the support vector machine SVM model.

[0115] The reduced-dimensional feature vector Y and the corresponding machine tool working status label L constitute the training data set: Among them, Y i represents the i-th eigenvector, L i represents the label corresponding to the i-th eigenvector, L i ∈{-1, +1}, +1 indicates the normal state of the machine tool, and -1 indicates the fault state of the machine tool.

[0116] The reduced feature vector Y i The corresponding machine tool working status label L i Paired to form a training data set for training fault prediction models.

[0117] The fault prediction model is constructed based on the support vector machine (SVM). The soft margin SVM model is adopted, and the slack variable ξi is introduced. The optimization problem of SVM is written as:

[0118]

[0119] sL i (w T Y i +b)≥1-ξ i ,ξ i ≥0, i=1,2,…,n

[0120] Among them, C represents the penalty factor, b represents the bias term, w represents the weight vector, and w T represents the transpose of the weight vector, ξ i represents the i-th slack variable.

[0121] The soft margin SVM model allows a certain amount of classification error by introducing a slack variable ξ i To measure the degree of classification error.

[0122] The optimization goal is to minimize the L2 norm of the weight vector w and the weighted sum of the classification errors, where C is the penalty factor used to control the penalty intensity of the classification error.

[0123] The constraint condition requires that for each training sample, the difference between the model's prediction value and the true label does not exceed 1-ξ i .

[0124] The original optimization problem is transformed into a dual problem using the Lagrange multiplier method, and the Lagrange multiplier α is introduced. i :

[0125]

[0126] Among them, K(Y i , Y j ) represents the kernel function, L i and L j Represent the true labels of the i-th and j-th samples respectively, α i represents the i-th inequality constraint, α j represents the jth inequality constraint.

[0127] The optimization goal of the dual problem is to maximize α i The difference between the sum of and a quadratic term, subject to the constraint α i The value range and α i With label L i The inner product of is 0.

[0128] K(Y i , Y j ) is a kernel function used to map the feature vector to a high-dimensional space; the kernel function selected is RBF: K(Y i , Y j )=exp(-γ||Y i -Y j || 2 ).

[0129] Kernel function K(Y i , Y j ) measures the two eigenvectors Y i and Y j Similarity in high-dimensional space. Commonly used kernel functions include linear kernel, polynomial kernel and radial basis kernel function (RBF). Among them, the RBF kernel function controls the width of the Gaussian function through the parameter γ and is a commonly used kernel function in SVM.

[0130] Solve the transformed dual problem and get α i The optimal solution Calculate the weight vector of the SVM model and bias

[0131]

[0132] Where S is the subscript set of support vectors, Ns is the number of support vectors, is the optimal solution to the j-th dual problem, is the weight of the i-th sample when constructing the optimal weight vector.

[0133] The grid search and cross-validation methods are used to optimize the hyperparameters of SVM, including the penalty factor C and the kernel function parameter γ: the search range and step size of C and γ are set to construct a hyperparameter grid; the k′-fold cross-validation method is used to divide the training data set into k′ subsets.

[0134] For each set of hyperparameter combinations, each subset was used as a validation set in turn, and the remaining subsets were used as training sets. The SVM model was trained and the performance was evaluated on the validation set.

[0135] Retrain the SVM model using the optimal parameters to obtain the final fault prediction model:

[0136]

[0137] Among them, f(y) represents the prediction function of the SVM model, sign(·) is a binary function, and the output result is -1 or +1, indicating the normal or faulty state of the machine tool.

[0138] Evaluate the model performance on the test set, calculate indicators such as precision, recall, and F1 value, and evaluate the generalization ability of the model.

[0139] Use the trained SVM model to predict the test set, compare the prediction results with the true labels, calculate evaluation indicators such as precision, recall rate, F1 value, and evaluate the generalization ability of the model on unknown data.

[0140] Based on the evaluation results, the strategies of feature extraction, feature selection and model training are further adjusted to optimize the fault prediction performance.

[0141] Step 5: Deploy the trained CNC machine tool fault prediction model to the embedded control system of the CNC machine tool to predict faults of the CNC machine tool, i.e., factory manufacturing equipment.

[0142] Convert the trained SVM model into a format suitable for running on embedded systems, including C / C++ code, and deploy the CNC machine tool fault prediction model to the embedded control system of the CNC machine tool, including PLC and ARM systems.

[0143] The vibration signal data and tool wear data of the machine tool are collected in real time and input into the prediction model for fault prediction. According to the prediction results, when the output result of the CNC machine tool fault prediction model is +1, the CNC machine tool will not fail in the future time period; when the output result of the CNC machine tool fault prediction model is -1, the CNC machine tool will fail in the future time period and the machine tool needs to be repaired.

[0144] Example 2

[0145] Reference Figure 2 The second embodiment of the present invention provides a data-driven fault prediction system for factory manufacturing equipment.

[0146] The system includes: a data acquisition module, a feature extraction module, a feature fusion module, a fault prediction module and a model deployment module;

[0147] The data acquisition module is used to collect vibration signal data and tool wear data of factory manufacturing equipment data and pre-process the collected data. The factory manufacturing equipment is a CNC machine tool, and the tool is a cutting tool on the CNC machine tool.

[0148] The feature extraction module is used to extract time-frequency domain features from the collected vibration signal data, calculate the time-domain features of the vibration signal data, and calculate the frequency-domain features of the vibration signal data through Fourier transform; and extract tool wear features.

[0149] The feature fusion module is used to fuse the extracted tool wear features, vibration signal data features and cutting parameter features to form a high-dimensional feature vector, select the first k feature vectors through principal component analysis PCA, and reduce the dimension of the high-dimensional feature vector.

[0150] The fault prediction module is used to form a training data set by combining the feature vector and the corresponding CNC machine tool working state, and to build and train a CNC machine tool fault prediction model based on a support vector machine (SVM) model.

[0151] The model deployment module is used to deploy the trained CNC machine tool fault prediction model into the embedded control system of the CNC machine tool to perform fault prediction on the CNC machine tool, i.e., factory manufacturing equipment.

[0152] Example 3

[0153] Figure 3 FIG. 1 is a schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. Figure 3As shown, according to another aspect of the present invention, an electronic device 500 is also provided. The electronic device 500 may include one or more processors and one or more memories. The memories store computer readable codes, and when the computer readable codes are executed by one or more processors, the above-mentioned data-driven factory manufacturing equipment fault prediction method can be executed.

[0154] The method or system according to the embodiment of the present invention can also be used by Figure 3 The architecture of the electronic device shown is implemented.

[0155] like Figure 3 As shown, the electronic device 500 may include a bus 501, one or more CPUs 502, a read-only memory (ROM) 503, a random access memory (RAM) 504, a communication port 505 connected to a network, an input / output component 506, a hard disk 507, and the like.

[0156] The storage device in the electronic device 500, such as the ROM 503 or the hard disk 507, can store the data-driven factory manufacturing equipment fault prediction method provided by the present invention.

[0157] The data-driven fault prediction method for factory manufacturing equipment includes: collecting vibration signal data and tool wear data of factory manufacturing equipment data, preprocessing the collected data, wherein the factory manufacturing equipment is a CNC machine tool, and the tool is a cutting tool on the CNC machine tool; extracting time-frequency domain features of the collected vibration signal data, calculating the time domain features of the vibration signal data, and calculating the frequency domain features of the vibration signal data through Fourier transformation; extracting tool wear features; fusing the extracted tool wear features, vibration signal data features and cutting parameter features to form a high-dimensional feature vector, selecting the first k feature vectors through principal component analysis PCA to reduce the dimension of the high-dimensional feature vector; forming a training data set with the feature vector and the corresponding CNC machine tool working state, constructing and training a CNC machine tool fault prediction model based on a support vector machine SVM model; deploying the trained CNC machine tool fault prediction model to the embedded control system of the CNC machine tool, and performing fault prediction on the CNC machine tool, i.e., factory manufacturing equipment.

[0158] Example 4

[0159] Figure 4 It is a schematic diagram of the structure of a computer-readable storage medium provided by an embodiment of the present invention.

[0160] like Figure 4 As shown, a computer readable storage medium 600 according to one embodiment of the present invention.

[0161] Computer readable storage medium 600 has computer readable instructions stored thereon.

[0162] When the computer-readable instructions are executed by a processor, the data-driven fault prediction method for factory manufacturing equipment according to the embodiments of the present invention described with reference to the above drawings may be executed.

[0163] The storage medium 600 includes, but is not limited to, for example, volatile memory and / or non-volatile memory. Volatile memory may include, for example, random access memory (RAM) and cache memory (cache), etc. Non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, etc. In addition, according to an embodiment of the present invention, the process described above with reference to the flowchart may be implemented as a computer software program.

[0164] For example, the present invention provides a non-temporary machine-readable storage medium, which stores machine-readable instructions, and the machine-readable instructions can be run by a processor to execute instructions corresponding to the method steps provided by the present invention, for example: collecting vibration signal data and tool wear data of factory manufacturing equipment data, and preprocessing the collected data, wherein the factory manufacturing equipment is a CNC machine tool, and the tool is a cutting tool on the CNC machine tool: extracting time-frequency domain features of the collected vibration signal data, calculating the time domain features of the vibration signal data, and calculating the frequency domain features of the vibration signal data through Fourier transform; extracting tool wear features; fusing the extracted tool wear features, vibration signal data features and cutting parameter features to form a high-dimensional feature vector, selecting the top k feature vectors through principal component analysis PCA to reduce the dimension of the high-dimensional feature vector; forming a training data set with the feature vector and the corresponding CNC machine tool working state, and constructing and training a CNC machine tool fault prediction model based on a support vector machine SVM model; deploying the trained CNC machine tool fault prediction model to the embedded control system of the CNC machine tool to predict faults of the CNC machine tool, i.e., the factory manufacturing equipment.

[0165] When the computer program is executed by the central processing unit (CPU), the above functions defined in the method of the present invention are performed. The method, apparatus, and device of the present invention may be implemented in many ways. For example, the method, apparatus, and device of the present invention may be implemented by software, hardware, firmware, or any combination of software, hardware, and firmware.

[0166] The above sequence for the steps of the method is for illustration only, and the steps of the method of the present invention are not limited to the sequence specifically described above unless otherwise specifically stated.

[0167] In addition, in some embodiments, the present invention can also be implemented as a program recorded in a recording medium, which includes machine-readable instructions for implementing the method according to the present invention. Therefore, the present invention also covers a recording medium storing a program for executing the method according to the present invention.

[0168] In addition, the parts of the above technical solutions provided in the embodiments of the present invention that are consistent with the implementation principles of the corresponding technical solutions in the prior art are not described in detail to avoid excessive redundancy.

[0169] The specific implementation modes described above further describe the purpose, technical solutions and beneficial effects of the present invention. It should be understood that the above description is only a specific implementation mode of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A data-driven method for predicting faults in factory manufacturing equipment, characterized in that: include: Collecting vibration signal data and tool wear data of factory manufacturing equipment data, and preprocessing the collected data, wherein the factory manufacturing equipment is a CNC machine tool, and the tool is a cutting tool on the CNC machine tool; Extract the time-frequency domain features of the collected vibration signal data, calculate the time-domain features of the vibration signal data, and calculate the frequency-domain features of the vibration signal data through Fourier transform; extract the tool wear features; The extracted tool wear features, vibration signal data features and cutting parameter features are integrated to form a high-dimensional feature vector. The first k feature vectors are selected through principal component analysis (PCA) to reduce the dimension of the high-dimensional feature vector. The feature vector and the corresponding CNC machine tool working status constitute a training data set, and the CNC machine tool fault prediction model is constructed and trained based on the support vector machine SVM model; The trained CNC machine tool fault prediction model is deployed to the embedded control system of the CNC machine tool to predict faults of the CNC machine tool, i.e., factory manufacturing equipment.

2. The data-driven fault prediction method for factory manufacturing equipment according to claim 1, characterized in that: A three-axis acceleration sensor is arranged on the CNC machine tool to collect vibration signal data of the CNC machine tool on the X, Y, and Z axes, and the collected X, Y, and Z axes are combined into comprehensive vibration signal data, which is recorded as vibration signal data x(t), where t is a time variable, indicating the sampling time; The collected tool wear is expressed as: w(t)=W0+Δw(t); where w0 is the initial tool wear, Δw(t) is the wear increment relative to the initial state at sampling time t, indicating the change process of tool wear over time, the initial tool wear is a constant, and the wear increment is a function of time; The collected vibration signal data x(t) is filtered and a Butterworth low-pass filter and a high-pass filter are used to remove high-frequency noise and low-frequency interference; Butterworth low-pass filter transfer function: where ω c is the cut-off frequency, N l is the filter order, s is the complex frequency; Butterworth high-pass filter transfer function: The filtered CNC machine tool vibration signal data is recorded as 3. The data-driven fault prediction method for factory manufacturing equipment according to claim 2, characterized in that: Extract filtered CNC machine tool vibration signal data The time-frequency domain features and the tool wear features are extracted; the time-frequency domain features include time domain features and frequency domain features; the time domain features include mean μ, root mean square value RMS and peak value x p ; The frequency domain feature is obtained by Perform fast Fourier transform to obtain the spectrum X(f) and calculate the energy E of each frequency band k , characterizes the energy distribution of vibration signal data in different frequency bands; Extracting vibration signal data The frequency domain characteristics of Perform fast Fourier transform to get the spectrum X(f): Calculate the energy E of each frequency band k : Among them, f p is the starting frequency of the pth frequency band; Extract tool wear characteristics, including tool wear amount w(t): w(t) = W0 + Δw(t); tool wear rate v w (t): vw = Δw(t) / Δt; tool wear acceleration a w (t): a w (t) = Δv w (t) / Δt; where Δt is the time interval, Δv w (t) is the tool wear rate increment.

4. The data-driven fault prediction method for factory manufacturing equipment according to claim 3 is characterized in that: The extracted vibration signal data time-frequency domain features and tool wear features constitute the feature vector X: X=[μ,RMS,x p ,X(f),E K ,w(t),v w (t),a w (t)] T Normalize the eigenvector X and scale the eigenvalues ​​to the range [0, 1]: Among them, x i is the i-th original eigenvalue in the eigenvector X, x max and x min are the maximum and minimum values ​​of the current feature respectively; Principal component analysis (PCA) is used to reduce the dimension of the normalized feature vector. The steps are as follows: Compute the mean vector of the eigenvectors: Compute the covariance matrix of the eigenvectors: Where n represents the number of training samples, and the covariance matrix C X Perform eigendecomposition and obtain eigenvalues ​​λ1, λ2, ..., λ m and the corresponding eigenvectors u1, u2, ..., u m , m represents the dimension of the original high-dimensional feature space; Select the eigenvectors corresponding to the first k largest eigenvalues ​​to form the transformation matrix: W = [u1, u2, ..., u k ], k<8; Use the transformation matrix to perform linear transformation on the normalized eigenvector to map it from high-dimensional space to low-dimensional space; reduce the dimension of the normalized eigenvector to obtain the reduced-dimensional eigenvector Y: Y = W T (X-μ X ), W T Represents the transpose of a transformation matrix.

5. The data-driven fault prediction method for factory manufacturing equipment according to claim 4, characterized in that: The reduced-dimensional feature vector Y and the corresponding machine tool working status label L constitute the training data set: Among them, Y i represents the i-th eigenvector, L i represents the label corresponding to the i-th eigenvector, L i ∈{-1, +1}, +1 indicates the normal state of the machine tool, and -1 indicates the fault state of the machine tool; The fault prediction model is constructed based on support vector machine (SVM), and the soft margin SVM model is adopted to introduce the relaxation variable ξ i , the optimization problem of SVM is written as: s.t.L i (w T Y i +b)≥1-ξ i ,ξ i ≥0,i=1,2,…,n Among them, C represents the penalty factor, b represents the bias term, w represents the weight vector, and w T represents the transpose of the weight vector, ξ i represents the i-th slack variable; The original optimization problem is transformed into a dual problem using the Lagrange multiplier method, and the Lagrange multiplier α is introduced. i : s.t.0≤α i ≤C, Among them, K(Y i , Y j ) represents the kernel function, L i and L j Represent the true labels of the i-th and j-th samples respectively, α i represents the i-th inequality constraint, α j represents the jth inequality constraint; The optimization goal of the dual problem is to maximize α i The difference between the sum of and a quadratic term, subject to the constraint α i The value range and α i With label L i The inner product of is 0.

6. The data-driven fault prediction method for factory manufacturing equipment according to claim 5, characterized in that: Solve the transformed dual problem and get α i The optimal solution Calculate the weight vector of the SVM model and bias Among them, S is the subscript set of support vectors, N s is the number of support vectors, is the optimal solution to the j-th dual problem, is the weight of the sample when constructing the optimal weight vector; Retrain the SVM model using the optimal parameters to obtain the final fault prediction model: Among them, f(y) represents the prediction function of the SVM model, sign(·) is a binary function, and the output result is -1 or +1, indicating the normal or faulty state of the machine tool.

7. The data-driven fault prediction method for factory manufacturing equipment according to claim 6, characterized in that: Convert the trained SVM model into a format suitable for running on embedded systems, including C / C++ code, and deploy the CNC machine tool fault prediction model into the embedded control system of the CNC machine tool, including PLC and ARM systems; Collect the vibration signal data and tool wear data of the machine tool in real time and input them into the prediction model for fault prediction; According to the prediction results, when the result output by the CNC machine tool fault prediction model is +1, the CNC machine tool will not fail in the future time period; When the result output by the CNC machine tool fault prediction model is -1, the CNC machine tool will fail in the future and needs to be repaired.

8. A data-driven factory manufacturing equipment fault prediction system, which is used to implement the data-driven factory manufacturing equipment fault prediction method according to any one of claims 1 to 7, characterized in that: include: Data acquisition module, feature extraction module, feature fusion module, fault prediction module and model deployment module; The data acquisition module is used to collect vibration signal data and tool wear data of factory manufacturing equipment data and pre-process the collected data. The factory manufacturing equipment is a CNC machine tool, and the tool is a cutting tool on the CNC machine tool; The feature extraction module is used to extract the time-frequency domain features of the collected vibration signal data, calculate the time-domain features of the vibration signal data, and calculate the frequency-domain features of the vibration signal data through Fourier transform; extract the tool wear features; The feature fusion module is used to fuse the extracted tool wear features, vibration signal data features and cutting parameter features to form a high-dimensional feature vector, and select the first k feature vectors through principal component analysis PCA to reduce the dimension of the high-dimensional feature vector; The fault prediction module is used to form a training data set with the feature vector and the corresponding CNC machine tool working state, and to build and train a CNC machine tool fault prediction model based on a support vector machine (SVM) model; The model deployment module is used to deploy the trained CNC machine tool fault prediction model into the embedded control system of the CNC machine tool to perform fault prediction on the CNC machine tool, i.e., factory manufacturing equipment.

9. An electronic device, characterized in that: include: A processor and a memory, wherein the memory stores a computer program that can be called by the processor; the processor executes the data-driven fault prediction method for factory manufacturing equipment as described in any one of claims 1 to 7 by calling the computer program stored in the memory.

10. A computer-readable storage medium, characterized in that: Instructions are stored, and when the instructions are executed on a computer, the computer is caused to execute the data-driven fault prediction method for factory manufacturing equipment according to any one of claims 1 to 7.

Images