Abrasive drill working state monitoring and early warning method and system
By combining feature extraction and cluster analysis of vibration and acoustic emission signals, the problem of insufficient feature information in the monitoring of grinding and drilling operation status is solved, and accurate monitoring and early warning of the grinding and drilling status throughout its entire life cycle are realized.
Patent Information
- Application Number
- CN202511784306.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-01
- Publication Date
- 2026-03-03
AI Technical Summary
Existing technologies rely on a single signal source (such as vibration or acoustic emission signals), which has limited characteristic information and cannot effectively characterize the working status of the grinding and drilling machine throughout its entire life cycle. They are also difficult to capture transient changes and are prone to missing key failure signs.
By combining vibration and acoustic emission signals, time-frequency and frequency-domain features are extracted through empirical mode decomposition, Hilbert transform, and fast Fourier transform. Gaussian kernel similarity matrix and normalized Laplace matrix are constructed, and eigenvalue decomposition and k-means clustering are performed. Kernel density estimation and Mahalanobis distance are combined to monitor the grinding and drilling status.
It improves the completeness and accuracy of the grinding and drilling operation status characteristic information, enhances the ability to capture transient characteristics of non-stationary signals, and improves the accuracy and reliability of grinding and drilling operation status monitoring.
Smart Images
Figure CN121598032A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of equipment monitoring technology, and in particular to a method and system for monitoring and early warning of the working status of a grinding and drilling machine. Background Technology
[0002] As industrial manufacturing equipment develops towards higher speeds, greater precision, and greater intelligence, drilling, as one of the most commonly used metal cutting processes, directly impacts processing efficiency and workpiece quality due to its stability and reliability. The grinding drill, a key tool in drilling, is closely related to factors such as tool wear, machine tool rigidity, cutting parameters, and workpiece material. With the development of sensing technology, signal processing technology, and intelligent algorithms, academia and industry are increasingly exploring the use of multi-source monitoring information, such as vibration signals and acoustic emission signals, to identify and provide early warnings of the grinding drill's working status during drilling, thereby enabling early diagnosis of tool failure.
[0003] Existing technologies still have the following problems: some methods rely on only a single signal source (such as vibration or acoustic emission signals), resulting in limited feature information and an inability to effectively characterize the working status of the grinding and drilling machine throughout its entire life cycle; traditional feature extraction is mostly based on fixed statistical indicators or spectral energy distribution, making it difficult to capture transient changes and easily missing key failure signs. Summary of the Invention
[0004] In view of the aforementioned existing problems, the inventors have proposed the present invention.
[0005] Therefore, this invention provides a method and system for monitoring and early warning of the working status of a grinding drill, which solves the problems of some methods relying only on a single signal source (such as vibration or acoustic emission signal), having limited feature information, being unable to effectively characterize the working status of the grinding drill throughout its entire life cycle, and traditional feature extraction being based on fixed statistical indicators or spectral energy distribution, making it difficult to capture transient changes and easily missing key failure signs.
[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: In a first aspect, the present invention provides a method for monitoring and early warning of the working status of a grinding drill, comprising, Collect and preprocess operational data, extract time-frequency and frequency domain features from the operational data, and concatenate them into a comprehensive feature vector; Calculate the Euclidean distance between eigenvectors, construct the Gaussian kernel similarity matrix, calculate the normalized Laplacian matrix, and perform eigenvalue decomposition to form an eigenvector matrix. Use k-means clustering to cluster the eigenvectors and generate a set of state clusters. Based on the state cluster set, the kernel density of the comprehensive feature vector is estimated, normal comprehensive feature vectors are selected, the Mahalanobis distance between the comprehensive feature vectors is calculated, and the working status of the grinding and drilling is monitored.
[0007] As a preferred embodiment of the grinding and drilling working status monitoring and early warning method of the present invention, wherein: the extraction of time-frequency and frequency domain features of the operating data, concatenated into a comprehensive feature vector, includes: Empirical mode decomposition is performed on vibration and acoustic emission signals to generate eigenmode functions; Perform a Hilbert transform on each intrinsic modulus function, and calculate the instantaneous amplitude and frequency based on the Hilbert transform results as time-frequency features; The vibration and acoustic emission signals are subjected to fast Fourier transform to generate frequency domain signals. The main frequency component and amplitude peak value are extracted from the frequency domain signals as frequency domain features. The time-frequency and frequency-domain features are concatenated into a comprehensive feature vector, and principal component analysis is used to reduce the dimensionality of the comprehensive feature vector.
[0008] As a preferred embodiment of the grinding and drilling working status monitoring and early warning method of the present invention, the step of constructing a Gaussian kernel similarity matrix, calculating a normalized Laplacian matrix, and performing eigenvalue decomposition to form an eigenvector matrix includes: Based on the dimensionality-reduced comprehensive feature vectors, the Euclidean distance between the feature vectors is calculated, and a Gaussian kernel similarity matrix is constructed from the Euclidean distance between the feature vectors. Based on the similarity matrix, a normalized Laplacian matrix is calculated. Eigenvalue decomposition is then performed on the Laplacian matrix, and the eigenvalues are sorted from largest to smallest. The eigenvalue gap method is used to determine the number of eigenvectors to be selected. Before choosing The eigenvectors corresponding to the largest eigenvalues form an eigenvector matrix.
[0009] As a preferred embodiment of the grinding and drilling working status monitoring and early warning method of the present invention, the step of clustering feature vectors using k-means clustering to generate a set of state clusters includes: Normalize each row of the eigenvector matrix, randomly select initial cluster centers from the normalized eigenvector matrix, calculate the Euclidean distance between each row of eigenvectors in the normalized eigenvector matrix and the initial cluster centers, assign each eigenvector to the cluster center with the smallest distance to form a preliminary state cluster, and based on the state cluster, traverse all normalized eigenvectors, calculate the average value, and use it as the new cluster center. The k-means iterative optimization method is used to set a maximum number of iterations. When the maximum number of iterations is reached, the update stops and a set of state clusters is generated.
[0010] As a preferred embodiment of the grinding and drilling working status monitoring and early warning method of the present invention, the step of estimating the kernel density of the comprehensive feature vector based on the state cluster set and filtering the normal comprehensive feature vector includes: Based on the set of state clusters, the kernel density of the comprehensive feature vector after dimensionality reduction of each state cluster is estimated to obtain the probability density of the state cluster. Using statistical analysis to set a classification threshold, the comprehensive feature vectors corresponding to probability densities greater than the classification threshold are selected as normal comprehensive feature vectors.
[0011] As a preferred embodiment of the grinding and drilling working status monitoring and early warning method of the present invention, the step of calculating the Mahalanobis distance between comprehensive feature vectors to monitor the grinding and drilling working status includes: Calculate the Mahalanobis distance between the mean of the normal composite eigenvectors and the dimensionality-reduced composite eigenvectors; The monitoring threshold is set using statistical analysis. The Mahalanobis distance is compared with the monitoring threshold. When the Mahalanobis distance is less than the monitoring threshold, it is judged as a normal state; otherwise, it is marked as an abnormal state.
[0012] As a preferred embodiment of the grinding and drilling working status monitoring and early warning method of the present invention, the step of collecting and preprocessing the operating data includes: The operation data of the grinding and drilling equipment is collected using smart sensors and then processed for noise reduction and normalization. The intelligent sensor includes vibration and acoustic emission sensors; The operational data includes vibration and acoustic emission signals.
[0013] Secondly, the present invention provides a grinding and drilling working status monitoring and early warning system, comprising, The data collection and processing module is used to collect the operating data of the grinding and drilling equipment and perform preprocessing. The comprehensive feature module is used to extract the instantaneous amplitude and instantaneous frequency of vibration and acoustic emission signals as time-frequency features, extract the main frequency component and amplitude peak value as frequency domain features, and concatenate the time-frequency and frequency domain features into a comprehensive feature vector; The decomposition and clustering module is used to construct the Gaussian kernel similarity matrix, obtain the normalized Laplacian matrix and perform eigenvalue decomposition, and perform k-means clustering on the eigenvector matrix to form a cluster set of grinding and drilling working states. The screening and monitoring module is used to perform kernel density estimation on the comprehensive feature vector based on the state cluster, screen out normal feature vectors with probability density greater than the threshold, calculate the Mahalanobis distance between the comprehensive feature vectors, and monitor the working status of the grinding and drilling.
[0014] Thirdly, the present invention provides a computer device including a memory and a processor, wherein the memory stores a computer program, wherein when the computer program is executed by the processor, it implements any step of the grinding and drilling working status monitoring and early warning method as described in the first aspect of the present invention.
[0015] Fourthly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the grinding and drilling working status monitoring and early warning method as described in the first aspect of the present invention.
[0016] The beneficial effects of this invention are as follows: By combining vibration signals with acoustic emission signals, this invention improves the completeness and accuracy of the characteristic information of the grinding and drilling operation status, and by combining empirical mode decomposition with Hilbert transform, it enhances the ability to capture the transient characteristics of non-stationary signals. Attached Figure Description
[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0018] Figure 1 This is a flowchart of the grinding and drilling working status monitoring and early warning method in Example 1.
[0019] Figure 2 This is a schematic diagram of the grinding and drilling working status monitoring and early warning system in Example 1. Detailed Implementation
[0020] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.
[0021] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.
[0022] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.
[0023] Example 1, referring to Figure 1 and Figure 2 This is the first embodiment of the present invention, which provides a method for monitoring and early warning of the working status of a grinding drill, comprising the following steps: S1. Collect and preprocess the running data, extract the time-frequency and frequency domain features of the running data, and concatenate them into a comprehensive feature vector; Specifically, collecting and preprocessing runtime data includes: The operation data of the grinding and drilling equipment is collected using smart sensors and then processed for noise reduction and normalization. The intelligent sensor includes vibration and acoustic emission sensors; The operational data includes vibration and acoustic emission signals.
[0024] Denoising effectively filters out environmental noise, random interference, and electromagnetic interference, ensuring the accuracy of subsequent feature extraction. Normalization ensures that the signal amplitudes output by different sensors are on the same order of magnitude, avoiding the impact of excessive differences in signal scale on feature fusion, thereby improving the robustness of the model.
[0025] Furthermore, the time-frequency and frequency-domain features of the running data are extracted and concatenated into a comprehensive feature vector, including: Empirical mode decomposition is performed on vibration and acoustic emission signals to generate eigenmode functions; Perform a Hilbert transform on each intrinsic modulus function, and calculate the instantaneous amplitude and frequency based on the Hilbert transform results as time-frequency features, as shown in the formula:
[0026] in The result of the Hilbert transform. For the first Each intrinsic modulus function For integration variables, For the first The instantaneous amplitude of each eigenmode function. For the first The instantaneous frequency of each eigenmode function. For time, It is a differential operator; The vibration and acoustic emission signals are subjected to fast Fourier transform to generate frequency domain signals. The main frequency component and amplitude peak value are extracted from the frequency domain signals as frequency domain features. The time-frequency and frequency-domain features are concatenated into a comprehensive feature vector, as shown in the formula: , in For the first A comprehensive feature vector, and These are the time-frequency characteristics of vibration and acoustic emission signals, respectively. and These are the frequency domain characteristics of vibration and acoustic emission signals, respectively; Principal component analysis is used to reduce the dimensionality of the composite eigenvectors.
[0027] The introduction of Hilbert transform can obtain instantaneous amplitude and instantaneous frequency, thereby revealing the dynamic energy distribution and frequency evolution of the signal on the time axis. Empirical mode decomposition decomposes complex waveforms into eigenmode functions of different scales. Fast Fourier transform can efficiently convert time-domain signals into frequency-domain representations, revealing the distribution law of signal energy at different frequencies. Through multi-source feature splicing, feature-level information fusion is realized, which considers both local instantaneous changes and overall spectral features, making the comprehensive features more fully reflect the operating status of the equipment. PCA maps high-dimensional features to low-dimensional space through linear transformation, maximizing the retention of variance information of the original data. The dimensionality-reduced feature vectors are more suitable for real-time processing, reducing computational resource consumption and improving the feasibility of online monitoring and diagnosis.
[0028] S2. Calculate the Euclidean distance between eigenvectors, construct the Gaussian kernel similarity matrix, calculate the normalized Laplacian matrix, and perform eigenvalue decomposition to form an eigenvector matrix. Use k-means clustering to cluster the eigenvectors and generate a set of state clusters. Specifically, a Gaussian kernel similarity matrix is constructed, the normalized Laplacian matrix is calculated, and eigenvalue decomposition is performed to form an eigenvector matrix, including: Based on the dimensionality-reduced composite feature vectors, the Euclidean distance between the feature vectors is calculated using the following formula: , in Let be the Euclidean distance between the m-th and n-th eigenvectors, where m and n are the m-th and n-th time windows, respectively. The dimension of the feature vector. and These are the k-th eigenvalues for the m-th and n-th time windows, respectively; The Gaussian kernel similarity matrix is constructed from the Euclidean distance between the eigenvectors, using the following formula: , in For elements of the similarity matrix, The width of the Gaussian kernel; Based on the similarity matrix S, the normalized Laplacian matrix is calculated using the following formula: , , Where L is the normalized Laplacian matrix, D is the degree matrix, and S is the similarity matrix. Elements of the degree matrix; Perform eigenvalue decomposition on the Laplacian matrix, sort the eigenvalues in descending order, and use the eigenvalue gap method to determine the number of eigenvectors to select. Before choosing The eigenvectors corresponding to the largest eigenvalues form an eigenvector matrix, as shown in the formula: , Where U is the eigenvector matrix. No. 1 eigenvector.
[0029] Euclidean distance can quantify the differences between features in different time windows, intuitively reflecting the magnitude of changes in equipment operating status. Compared with cosine similarity, Euclidean distance is more sensitive to amplitude differences and is more suitable for capturing changes in energy-related features in operating data. The combination of Euclidean distance and Gaussian kernel improves the expressive power of nonlinear feature spaces. By introducing Gaussian kernel, the ability to identify local clustering structures is enhanced, avoiding the failure problem of global linear methods under non-convex data distributions. Laplace matrix can characterize the relationship between data points from a graph theory perspective, transforming the similarity between feature vectors into a graph structure problem. Eigenvalue gap method is used to determine the number of feature vectors, avoiding blindly selecting dimensions.
[0030] Furthermore, k-means clustering is used to cluster the feature vectors, generating a set of state clusters, including: Normalize each row of the eigenvector matrix, randomly select initial cluster centers from the normalized eigenvector matrix, calculate the Euclidean distance between each row of eigenvectors in the normalized eigenvector matrix and the initial cluster centers, assign each eigenvector to the cluster center with the smallest distance to form a preliminary state cluster, and based on the state cluster, traverse all normalized eigenvectors, calculate the average value, and use it as the new cluster center. The k-means iterative optimization method is used to set a maximum number of iterations. When the maximum number of iterations is reached, the update stops and a set of state clusters is generated.
[0031] By performing K-means clustering in the low-dimensional space generated by spectral clustering, the convergence speed and accuracy of clustering are improved. Normalization avoids the problem of unbalanced feature weights in different dimensions, making the clustering results more discriminative. The final set of state clusters can reflect various typical modes in the equipment operation process, such as normal operation, slight wear, severe wear, or abnormal vibration.
[0032] S3. Based on the state cluster set, perform kernel density estimation on the comprehensive feature vector, filter out normal comprehensive feature vectors, calculate the Mahalanobis distance between comprehensive feature vectors, and monitor the working status of the grinding and drilling. Specifically, based on the set of state clusters, kernel density estimation is performed on the comprehensive feature vectors to filter out normal comprehensive feature vectors, including: Based on the set of state clusters, the kernel density of the comprehensive feature vector after dimensionality reduction of each state cluster is estimated to obtain the probability density of the state cluster, as shown in the formula: , , in Let be the probability density of the i-th state cluster at point x, where x is the target point in the feature space. Let F be the i-th state cluster, and F be the comprehensive feature vector after dimensionality reduction. Let h be the dimension of the feature vector and h be the bandwidth, calculated using the Silverman rule. Let u be the Gaussian kernel function, and u be the standardized distance. Using statistical analysis to set a classification threshold, the comprehensive feature vectors corresponding to probability densities greater than the classification threshold are selected as normal comprehensive feature vectors.
[0033] Kernel density estimation combined with Gaussian kernel function improves the ability to characterize the continuity of state distribution. Silverman bandwidth selection combined with Gaussian kernel function improves estimation accuracy and stability. Kernel density estimation combined with statistical threshold improves the reliability of normal data screening. The screening process is equivalent to "denoising" the feature space, which enhances the accuracy of subsequent Mahalanobis distance calculation.
[0034] Furthermore, the Mahalanobis distance between the composite feature vectors is calculated to monitor the working status of the grinding drill, including: The Mahalanobis distance between the mean of the normal composite eigenvectors and the dimensionality-reduced composite eigenvectors is calculated using the following formula: , , in The Mahalanobis distance, The mean of the normal comprehensive feature vector. For transpose operation, Let covariance matrix be the variance matrix. The number of feature vectors of the normal state cluster; The monitoring threshold is set using statistical analysis. The Mahalanobis distance is compared with the monitoring threshold. When the Mahalanobis distance is less than the monitoring threshold, it is judged as a normal state; otherwise, it is marked as an abnormal state.
[0035] Mahalanobis distance combined with covariance matrix improves the ability to characterize the correlation of multidimensional features; mean vector combined with covariance matrix improves the sensitivity of anomaly identification; Mahalanobis distance combined with monitoring threshold improves the accuracy of anomaly identification. By setting the threshold through statistical analysis, it can be dynamically adjusted according to actual operating data, making the method more adaptable and universal.
[0036] This embodiment also provides a grinding and drilling working status monitoring and early warning system, including: The data collection and processing module is used to collect the operating data of the grinding and drilling equipment and perform preprocessing. The comprehensive feature module is used to extract the instantaneous amplitude and instantaneous frequency of vibration and acoustic emission signals as time-frequency features, extract the main frequency component and amplitude peak value as frequency domain features, and concatenate the time-frequency and frequency domain features into a comprehensive feature vector; The decomposition and clustering module is used to construct the Gaussian kernel similarity matrix, obtain the normalized Laplacian matrix and perform eigenvalue decomposition, and perform k-means clustering on the eigenvector matrix to form a cluster set of grinding and drilling working states. The screening and monitoring module is used to perform kernel density estimation on the comprehensive feature vector based on the state cluster, screen out normal feature vectors with probability density greater than the threshold, calculate the Mahalanobis distance between the comprehensive feature vectors, and monitor the working status of the grinding and drilling.
[0037] This embodiment also provides a computer device applicable to the grinding and drilling working status monitoring and early warning method, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to realize the grinding and drilling working status monitoring and early warning method proposed in the above embodiment.
[0038] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.
[0039] This embodiment also provides a storage medium storing a computer program, which, when executed by a processor, implements the method for monitoring and early warning of the working status of a grinding mill as proposed in the above embodiment. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.
[0040] In summary, this invention improves the completeness and accuracy of information on the characteristics of the grinding and drilling operation by combining vibration signals with acoustic emission signals, and enhances the ability to capture transient characteristics of non-stationary signals by combining empirical mode decomposition with Hilbert transform.
[0041] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.
Claims
1. A method for monitoring and early warning of the working status of a grinding drill, characterized in that: include, Collect and preprocess operational data, extract time-frequency and frequency domain features from the operational data, and concatenate them into a comprehensive feature vector; Calculate the Euclidean distance between eigenvectors, construct the Gaussian kernel similarity matrix, calculate the normalized Laplacian matrix, and perform eigenvalue decomposition to form an eigenvector matrix. Use k-means clustering to cluster the eigenvectors and generate a set of state clusters. Based on the state cluster set, the kernel density of the comprehensive feature vector is estimated, normal comprehensive feature vectors are selected, the Mahalanobis distance between the comprehensive feature vectors is calculated, and the working status of the grinding and drilling is monitored.
2. The grinding and drilling working status monitoring and early warning method as described in claim 1, characterized in that: The extracted time-frequency and frequency-domain features of the running data are concatenated into a comprehensive feature vector, including: Empirical mode decomposition is performed on vibration and acoustic emission signals to generate eigenmode functions; Perform a Hilbert transform on each intrinsic modulus function, and calculate the instantaneous amplitude and frequency based on the Hilbert transform results as time-frequency features; The vibration and acoustic emission signals are subjected to fast Fourier transform to generate frequency domain signals. The main frequency component and amplitude peak value are extracted from the frequency domain signals as frequency domain features. The time-frequency and frequency-domain features are concatenated into a comprehensive feature vector, and principal component analysis is used to reduce the dimensionality of the comprehensive feature vector.
3. The grinding and drilling working status monitoring and early warning method as described in claim 2, characterized in that: The process involves constructing a Gaussian kernel similarity matrix, calculating the normalized Laplacian matrix, and performing eigenvalue decomposition. The eigenvector matrix consists of: Based on the dimensionality-reduced comprehensive feature vectors, the Euclidean distance between the feature vectors is calculated, and a Gaussian kernel similarity matrix is constructed from the Euclidean distance between the feature vectors. Based on the similarity matrix, a normalized Laplacian matrix is calculated. Eigenvalue decomposition is then performed on the Laplacian matrix, and the eigenvalues are sorted from largest to smallest. The eigenvalue gap method is used to determine the number of eigenvectors to be selected. Before choosing The eigenvectors corresponding to the largest eigenvalues form an eigenvector matrix.
4. The grinding and drilling working status monitoring and early warning method as described in claim 3, characterized in that: The step of clustering feature vectors using k-means clustering to generate a set of state clusters includes: Normalize each row of the eigenvector matrix, randomly select initial cluster centers from the normalized eigenvector matrix, calculate the Euclidean distance between each row of eigenvectors in the normalized eigenvector matrix and the initial cluster centers, assign each eigenvector to the cluster center with the smallest distance to form a preliminary state cluster, and based on the state cluster, traverse all normalized eigenvectors, calculate the average value, and use it as the new cluster center. The k-means iterative optimization method is used to set a maximum number of iterations. When the maximum number of iterations is reached, the update stops and a set of state clusters is generated.
5. The grinding and drilling working status monitoring and early warning method as described in claim 4, characterized in that: The process of estimating the kernel density of the comprehensive feature vector based on the state cluster set and filtering out normal comprehensive feature vectors includes: Based on the set of state clusters, the kernel density of the comprehensive feature vector after dimensionality reduction of each state cluster is estimated to obtain the probability density of the state cluster. Using statistical analysis to set a classification threshold, the comprehensive feature vectors corresponding to probability densities greater than the classification threshold are selected as normal comprehensive feature vectors.
6. The grinding and drilling working status monitoring and early warning method as described in claim 5, characterized in that: The calculation of the Mahalanobis distance between the comprehensive feature vectors, and the monitoring of the grinding and drilling working status, includes: Calculate the Mahalanobis distance between the mean of the normal composite feature vector and the dimension-reduced composite feature vector; The monitoring threshold is set using statistical analysis. The Mahalanobis distance is compared with the monitoring threshold. When the Mahalanobis distance is less than the monitoring threshold, it is judged as a normal state; otherwise, it is marked as an abnormal state.
7. The grinding and drilling working status monitoring and early warning method as described in claim 6, characterized in that: The collection and preprocessing of operational data includes: The operation data of the grinding and drilling equipment is collected using smart sensors and then processed for noise reduction and normalization. The intelligent sensor includes vibration and acoustic emission sensors; The operational data includes vibration and acoustic emission signals.
8. A grinding drill working status monitoring and early warning system, based on the grinding drill working status monitoring and early warning method according to any one of claims 1 to 7, characterized in that: include, The data collection and processing module is used to collect the operating data of the grinding and drilling equipment and perform preprocessing. The comprehensive feature module is used to extract the instantaneous amplitude and instantaneous frequency of vibration and acoustic emission signals as time-frequency features, extract the main frequency component and amplitude peak value as frequency domain features, and concatenate the time-frequency and frequency domain features into a comprehensive feature vector; The decomposition and clustering module is used to construct the Gaussian kernel similarity matrix, obtain the normalized Laplacian matrix and perform eigenvalue decomposition, and perform k-means clustering on the eigenvector matrix to form a cluster set of grinding and drilling working states. The screening and monitoring module is used to perform kernel density estimation on the comprehensive feature vector based on the state cluster, screen out normal feature vectors with probability density greater than the threshold, calculate the Mahalanobis distance between the comprehensive feature vectors, and monitor the working status of the grinding and drilling.
9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the grinding and drilling working status monitoring and early warning method according to any one of claims 1 to 7.
10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the grinding and drilling working status monitoring and early warning method according to any one of claims 1 to 7.