Multi-dimensional state comprehensive detection method for numerical control machine tool

Abstract

The invention discloses a numerical control machine tool multi-dimensional state comprehensive detection method, and belongs to the field of numerical control machine tools, the method comprises the following steps: obtaining multi-dimensional data of a numerical control machine tool, the multi-dimensional data comprising vibration data and audio data; acquiring a vibration feature vector of each piece of vibration data; acquiring an audio feature vector of each piece of audio data; splicing the vibration feature vector and the audio feature vector in pairs to obtain a plurality of first feature vectors; performing dimension reduction on the plurality of first feature vectors to obtain a second feature vector; the second feature vector is input into a support vector machine, a predicted state detection result of the numerical control machine tool is obtained, and the state detection result of the numerical control machine tool comprises a normal state and a fault state. The method can detect the state of the numerical control machine tool and predict whether the numerical control machine tool has a fault or not.

Description

Technical Field

[0001] The present disclosure relates to the field of numerically controlled machine tools, and in particular to a method for comprehensive detection of multi-dimensional states of numerically controlled machine tools. Background Art

[0002] During long-term operation, CNC machine tools may malfunction due to wear of mechanical parts, system failure, environmental factors, etc., causing the CNC machine tools to be in a faulty state.

[0003] In the related art, the status detection of CNC machine tools is usually achieved by monitoring data of a single dimension, for example, separately monitoring the temperature, voltage, current, etc. of the CNC machine tool.

[0004] However, these single-dimensional detection methods cannot fully reflect the true status of CNC machine tools, nor can they identify potential faults, and it is easy for CNC machine tools to report faults only after they occur. Summary of the invention

[0005] The present disclosure provides a multi-dimensional state comprehensive detection method for a CNC machine tool, which can detect the state of the CNC machine tool and predict whether the CNC machine tool has a fault. The technical solution at least includes the following solutions: In a first aspect, a method for comprehensive detection of multi-dimensional states of a CNC machine tool is provided, comprising: acquiring multi-dimensional data of the CNC machine tool, the multi-dimensional data comprising vibration data and audio data; acquiring a vibration feature vector of each of the vibration data, the vibration feature vector comprising root mean square, peak value, pulse degree, kurtosis, and spectral characteristics of the vibration data; acquiring an audio feature vector of each of the audio data, the audio feature vector comprising Mel-frequency cepstrum coefficients and linear prediction cepstrum coefficients of the audio data; concatenating the vibration feature vectors and the audio feature vectors in pairs to obtain a plurality of first feature vectors; performing dimensionality reduction on the plurality of the first feature vectors to obtain a second feature vector; inputting the second feature vector into a support vector machine to obtain a predicted state detection result of the CNC machine tool, the state detection result of the CNC machine tool comprising a normal state and a fault state.

[0006] Optionally, reducing the dimension of the first eigenvector to obtain the second eigenvector includes: reducing the dimension of the first eigenvector using a principal component analysis method to obtain the second eigenvector.

[0007] Optionally, the second eigenvector is expressed by the following formula:

[0008] in, is the second eigenvector, is a characteristic matrix composed of a plurality of the first eigenvectors, is a principal component matrix, wherein the principal component matrix includes the first k principal components of the first eigenvector, and the first k principal components are obtained by performing eigenvalue decomposition on the covariance matrix.

[0009] Optionally, obtaining the audio feature vector of each of the audio data includes: preprocessing each of the audio data, the preprocessing including pre-emphasis processing, denoising processing and smoothing processing; performing Fourier transform on the preprocessed audio data to obtain first spectrum data; using a Mel filter to decompose the first spectrum data into multiple Mel frequency bands; logarithmically processing the energy of each of the Mel frequency bands to obtain multiple Mel energy spectra; performing discrete cosine transform on each of the Mel energy spectra to obtain multiple Mel-frequency cepstrum coefficients.

[0010] Optionally, obtaining the audio feature vector of each of the audio data includes: preprocessing each of the audio data, the preprocessing including pre-emphasis processing, frame processing and smoothing processing; performing linear prediction analysis on each of the preprocessed audio data to obtain multiple prediction coefficients; performing discrete cosine transform on each of the prediction coefficients to obtain multiple linear prediction cepstrum coefficients.

[0011] In the second aspect, a multi-dimensional state comprehensive detection device for a CNC machine tool is also provided, including: a data acquisition module, used to acquire multi-dimensional data of the CNC machine tool, the multi-dimensional data including vibration data and audio data; a first vector acquisition module, used to acquire a vibration feature vector of each of the vibration data, the vibration feature vector including the root mean square, peak value, pulse degree, kurtosis, and spectrum characteristics of the vibration data; a second vector acquisition module, used to acquire an audio feature vector of each of the audio data, the audio feature vector including the Mel-frequency cepstral coefficients and the linear prediction cepstral coefficients of the audio data; a splicing module, used to splice the vibration feature vectors and the audio feature vectors in pairs to obtain a plurality of first feature vectors; a dimensionality reduction module, used to reduce the dimensionality of the plurality of first feature vectors to obtain a second feature vector; a state detection module, used to input the second feature vector into a support vector machine to obtain a predicted state detection result of the CNC machine tool, the state detection result of the CNC machine tool including a normal state and a fault state.

[0012] Optionally, the dimension reduction module is further used to: use a principal component analysis method to reduce the dimension of the first eigenvector to obtain the second eigenvector.

[0013] Optionally, the second acquisition module is also used to preprocess each of the audio data, and the preprocessing includes pre-emphasis processing, denoising processing and smoothing processing; performing Fourier transform on the preprocessed audio data to obtain first spectrum data; using a Mel filter to decompose the first spectrum data into multiple Mel frequency bands; logarithmically processing the energy of each Mel frequency band to obtain multiple Mel energy spectra; performing discrete cosine transform on each Mel energy spectrum to obtain multiple Mel-frequency cepstrum coefficients.

[0014] Optionally, the second acquisition module is also used to preprocess each of the audio data, and the preprocessing includes pre-emphasis processing, frame processing and smoothing processing; perform linear prediction analysis on each of the preprocessed audio data to obtain multiple prediction coefficients; perform discrete cosine transform on each of the prediction coefficients to obtain multiple linear prediction cepstrum coefficients.

[0015] In a third aspect, a computer device is also provided, comprising: a memory and a processor, wherein at least one computer program is stored in the memory, and the at least one computer program is loaded and executed by the processor, thereby executing the multi-dimensional state comprehensive detection method of CNC machine tools described in the above embodiment.

[0016] In a fourth aspect, a computer-readable storage medium is also provided, in which at least one computer program is stored. The at least one computer program is loaded and executed by a processor, thereby executing the multi-dimensional state comprehensive detection method of CNC machine tools described in the above embodiment.

[0017] In a fifth aspect, a computer program product is provided, comprising a computer program / instruction, wherein when the computer program / instruction is executed by a processor, the method described in the first aspect is implemented.

[0018] The beneficial effects brought by the technical solution provided by the embodiments of the present disclosure include at least: In the disclosed embodiment, by obtaining multi-dimensional data of a CNC machine tool, wherein the multi-dimensional data includes vibration data and audio data, and then respectively obtaining a vibration feature vector and an audio feature vector, and then splicing the vibration feature vector and the audio feature vector to obtain a first feature vector, the first feature vector can be used for comprehensive detection of the machine tool state after dimensionality reduction, which is equivalent to integrating data of multiple dimensions for machine tool fault diagnosis, and can fully reflect the working state of the machine tool, and identify potential faults, effectively avoiding the problem of being unable to identify potential faults when using a single-dimensional detection method. After accurately predicting the state of the CNC machine tool, the CNC machine tool can be timely managed and controlled based on the state of the CNC machine tool, thereby extending the service life of the CNC machine tool. In addition, the method in the disclosed embodiment is applicable to various types of CNC machine tools, including but not limited to lathes, milling machines, grinders, etc., and has strong versatility and adaptability. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the technical solutions in the embodiments of the present disclosure, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present disclosure. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0020] Figure 1 A flowchart of a multi-dimensional state comprehensive detection method for a CNC machine tool provided by an exemplary embodiment of the present disclosure is shown; Figure 2 A flowchart of a multi-dimensional state comprehensive detection method for a CNC machine tool provided by another exemplary embodiment of the present disclosure is shown; Figure 3 A schematic diagram of the structure of a multi-dimensional state comprehensive detection device for a CNC machine tool provided by an exemplary embodiment of the present disclosure is shown; Figure 4 It is a schematic diagram of the structure of a computer device provided in an embodiment of the present disclosure. DETAILED DESCRIPTION

[0021] Unless otherwise defined, the technical or scientific terms used herein shall have the usual meanings understood by persons of ordinary skill in the field to which the present disclosure belongs. The words "first", "second", "third" and similar words used in the patent application specification and claims of the present disclosure do not indicate any order, quantity or importance, but are only used to distinguish different components. Similarly, words such as "one" or "one" do not indicate a quantitative limitation, but rather indicate the presence of at least one. Words such as "include" or "comprise" and similar words mean that the elements or objects appearing before "include" or "comprises" cover the elements or objects listed after "include" or "comprises" and their equivalents, and do not exclude other elements or objects.

[0022] In order to make the objectives, technical solutions and advantages of the present disclosure more clear, the embodiments of the present disclosure will be further described in detail below with reference to the accompanying drawings.

[0023] Figure 1 A flowchart of a method for comprehensive detection of multi-dimensional states of a CNC machine tool provided by an exemplary embodiment of the present disclosure is shown. The method can be executed by a computer device. Figure 1 , the method comprising: In step 101, multi-dimensional data of a CNC machine tool is obtained.

[0024] The multi-dimensional data includes vibration data and audio data, such as real-time vibration data and real-time audio data of a numerical control machine tool.

[0025] During implementation, the vibration data may be acquired through a vibration sensor disposed on a numerically controlled machine tool, and the audio data may be acquired through a sound collection device, such as a microphone or a pickup.

[0026] Here, the vibration data and the audio data are in the same time period, and each vibration data corresponds to one audio data. For example, the vibration data and audio data in the [A, B] time period can be obtained, and the vibration data at time t and the audio data at time t are corresponding. When the vibration feature vectors and the audio feature vectors are spliced ​​in pairs in the subsequent step 104, the vibration feature vectors corresponding to the vibration data at time t and the audio feature vectors corresponding to the audio data at time t are spliced. Where t is less than or equal to B and t is greater than or equal to A.

[0027] In step 102, a vibration feature vector of each vibration data is obtained.

[0028] The vibration feature vector includes the root mean square (RMS), peak value, impulse degree, kurtosis, and spectrum characteristics of the vibration data.

[0029] In step 103, an audio feature vector of each audio data is obtained.

[0030] The audio feature vector includes the Mel-frequency cepstral coefficients and the linear prediction cepstral coefficients of the audio data.

[0031] In step 104, the vibration feature vectors and the audio feature vectors are concatenated in pairs to obtain a plurality of first feature vectors.

[0032] In step 105, dimensionality reduction is performed on the plurality of first eigenvectors to obtain second eigenvectors.

[0033] In step 106, the second feature vector is input into a support vector machine to obtain a predicted state detection result of the CNC machine tool.

[0034] The status detection results of CNC machine tools include normal status and fault status.

[0035] The support vector machine in step 106 is a trained support vector machine that can be used for CNC machine tool state detection. If the predicted state detection result of the CNC machine tool output by the support vector machine is a fault state, it means that the CNC machine tool is about to fail (or has failed), and the staff needs to troubleshoot and repair the CNC machine tool to ensure that the CNC machine tool can run smoothly.

[0036] In the disclosed embodiment, by obtaining multi-dimensional data of a CNC machine tool, wherein the multi-dimensional data includes vibration data and audio data, and then respectively obtaining a vibration feature vector and an audio feature vector, and then splicing the vibration feature vector and the audio feature vector to obtain a first feature vector, the first feature vector can be used for comprehensive detection of the machine tool state after dimensionality reduction, which is equivalent to integrating data of multiple dimensions for machine tool fault diagnosis, and can fully reflect the working state of the machine tool, and identify potential faults, effectively avoiding the problem of being unable to identify potential faults when using a single-dimensional detection method. After accurately predicting the state of the CNC machine tool, the CNC machine tool can be timely managed and controlled based on the state of the CNC machine tool, thereby extending the service life of the CNC machine tool. In addition, the method in the disclosed embodiment is applicable to various types of CNC machine tools, including but not limited to lathes, milling machines, grinders, etc., and has strong versatility and adaptability.

[0037] Figure 2 A flowchart of a multi-dimensional state comprehensive detection method for a CNC machine tool provided by another exemplary embodiment of the present disclosure is shown. The method can be executed by a computer device. Figure 2 , the method comprising: In step 201, multi-dimensional data of a CNC machine tool is obtained.

[0038] The relevant contents of step 201 refer to the aforementioned step 101, and the detailed description is omitted here.

[0039] In step 202, the multi-dimensional data is preprocessed.

[0040] Preprocessing the multi-dimensional data includes: preprocessing the vibration data and preprocessing the audio data.

[0041] Optionally, preprocessing the vibration data includes Kalman filtering, Gaussian filtering, etc.

[0042] Optionally, preprocessing the audio data includes: pre-emphasis processing, frame processing and smoothing processing.

[0043] Pre-emphasis processing refers to enhancing the high-frequency part of the audio data in order to highlight the high-frequency information and compensate for the attenuation of the high-frequency part of the signal. Usually a simple high-pass filter is used to implement pre-emphasis. The pre-emphasis process can be expressed by formula (1).

[0044] (1) In formula (1), Represents the audio data at time t after pre-emphasis, represents the audio data at time t, represents the audio data at time t-1, It is the pre-emphasis coefficient, usually set to 0.95, which enhances the high-frequency part of the audio signal.

[0045] Framing refers to cutting audio data into frames. Each frame usually contains 10 to 50 milliseconds of audio data, and 25 milliseconds is common. Each frame will contain a certain number of sampling points to ensure that there is a 50% overlap between frames.

[0046] The smoothing process is, for example, a moving average method or a Gaussian filter. The smoothing process can smooth the instantaneous fluctuation of the signal to obtain relatively stable audio data.

[0047] In step 203, a vibration feature vector of each vibration data is obtained.

[0048] The vibration feature vector includes the root mean square, peak value, pulse degree, kurtosis, and spectrum characteristics of the vibration data. The vibration feature vector of each vibration data can be obtained by respectively calculating the root mean square, peak value, pulse degree, kurtosis, and spectrum characteristics of the preprocessed vibration data.

[0049] There are many methods for calculating root mean square, peak value, pulse degree, kurtosis, and spectrum characteristics in the related technology, and detailed description is omitted here.

[0050] After calculating the root mean square, peak value, pulse degree, kurtosis, and spectrum characteristics of each vibration data, for any vibration data, these parameters can be concatenated to obtain the vibration feature vector.

[0051] In step 204, an audio feature vector of each audio data is obtained.

[0052] The audio feature vector includes the Mel Frequency Cepstral Coefficients (MFCC) and Linear Predictive Cepstral Coefficients (LPCC) of the audio data. MFCC extracts features from audio data by imitating the auditory mechanism of the human ear. LPCC models audio data based on the Linear Predictive Coding (LPC) model and extracts cepstral coefficients.

[0053] Optionally, the Mel-frequency cepstral coefficients are calculated using the following steps ad.

[0054] Step a: Perform Fourier transform on the preprocessed audio data to obtain first spectrum data.

[0055] The audio signal of each frame is subjected to Fourier transform, so that the audio data can be converted from the time domain to the frequency domain to obtain the first spectrum data.

[0056] Step b: using a Mel filter to decompose the first spectrum data into a plurality of Mel frequency bands.

[0057] Mel frequency is a frequency scale that simulates the human auditory system, with denser low-frequency parts and sparser high-frequency parts. Mel filter banks convert the spectrum into Mel frequency scale. Mel filter banks decompose the spectrum into multiple bands, each corresponding to a different frequency range. The common Mel frequency range is 0~8000Hz, which is suitable for feature extraction of most audio signals.

[0058] Step c: logarithmically process the energy of each Mel frequency band to obtain multiple Mel energy spectra.

[0059] Here, logarithmic transformation can better simulate the way the human ear perceives volume.

[0060] Step d: performing discrete cosine transform on each Mel energy spectrum to obtain a plurality of Mel frequency cepstrum coefficients.

[0061] The Mel energy spectrum after logarithmic processing is transformed into cepstral coefficients by discrete cosine transform (DCT). Cepstral coefficients help remove redundant information in the spectrum and extract important features.

[0062] The MFCC feature will retain the first m coefficients as the main features of the audio signal. Exemplarily, the value range of m is 10-14, such as 10, 12, 14, etc.

[0063] Optionally, the linear prediction cepstral coefficients are calculated using the following steps ef.

[0064] Step e: performing linear prediction analysis on each preprocessed audio data to obtain multiple prediction coefficients.

[0065] Linear Prediction Analysis, or LPC, predicts the current sample value of the audio signal through a linear model to fit the spectral envelope of the signal. The goal of LPC is to minimize the prediction error of the signal and use these prediction coefficients to represent the audio signal.

[0066] Step f: performing discrete cosine transform on each prediction coefficient to obtain a plurality of linear prediction cepstral coefficients.

[0067] By performing DCT on each prediction coefficient, the redundancy of features is reduced and multiple linear prediction cepstral coefficients are obtained.

[0068] After calculating the MPCC and LPCC of each audio data, for any audio data, these parameters can be concatenated to obtain the audio feature vector.

[0069] In step 205, the vibration feature vectors and the audio feature vectors are concatenated in pairs to obtain a plurality of first feature vectors.

[0070] Before executing step 205, the vibration feature vector and the audio feature vector need to be normalized to eliminate the influence caused by different dimensions.

[0071] For example, for the vibration feature vector, the Z-Score normalization method can be used. Vibration data usually has large numerical fluctuations and may contain high-frequency noise, so normalization can smooth it and eliminate the differences between different vibration features at different scales. For audio feature vectors, the minimum-maximum normalization method can be used to scale each audio feature vector to the interval [0, 1]. This is because the features of audio signals usually have a small numerical range, and the numerical ranges of multiple audio features vary greatly. By normalizing the data, the unfair influence caused by the differences in numerical ranges and dimensions between the feature vectors can be effectively eliminated, ensuring that the subsequent support vector machine can reasonably use various features for learning and classification.

[0072] Exemplarily, the i-th vibration eigenvector can be expressed as , the i-th audio feature vector can be expressed as , then the concatenated first eigenvector can be expressed as , whose dimensions are When there are N first eigenvectors in total, the characteristic matrix formed by these N first eigenvectors can be expressed as , where the dimension of the feature matrix is .

[0073] In step 206, the dimensions of the multiple first eigenvectors are reduced to obtain second eigenvectors.

[0074] Optionally, principal component analysis (PCA) is used to reduce the dimensions of the plurality of first eigenvectors to obtain second eigenvectors.

[0075] The core of the principal component analysis method is to calculate the covariance matrix. In the embodiment of the present disclosure, the covariance matrix can be calculated using formula (2).

[0076] (2) In formula (2), is the covariance matrix, N is the number of vibration eigenvectors (also equal to the number of audio eigenvectors, or the number of first eigenvectors), N and i are positive integers, and the value range of i is 1 to N. is the i-th first eigenvector, is the mean of the N first eigenvectors.

[0077] Perform eigenvalue decomposition on the covariance matrix to obtain the eigenvalues ​​and eigenvectors of the covariance matrix, where the eigenvalues ​​correspond to the eigenvectors one by one. Sort the eigenvalues ​​and use the eigenvectors corresponding to the first k eigenvalues ​​as the first k principal components. These first k principal components can form a principal component matrix , the principal component matrix The dimension is .

[0078] The second eigenvector obtained by principal component analysis can be expressed by formula (3).

[0079] (3) In formula (3), is the second eigenvector, is a characteristic matrix consisting of multiple first eigenvectors, is a principal component matrix, which contains the first k principal components of the first eigenvector, and the first k principal components are obtained by performing eigenvalue decomposition on the covariance matrix.

[0080] Features after dimensionality reduction It is a low-dimensional representation that contains most of the information in the original data.

[0081] In step 207, the second feature vector is input into a support vector machine to obtain a predicted state detection result of the CNC machine tool.

[0082] The status detection results of CNC machine tools include normal status and fault status.

[0083] The support vector machine in step 207 is a trained support vector machine that can be used for CNC machine tool state detection. When training the support vector machine, the historical multi-dimensional data of the CNC machine tool can be used for training. For example, after the historical multi-dimensional data is processed in the manner in steps 202-206, the support vector machine is trained using the historical multi-dimensional data. After the support vector machine training is completed, the support vector machine can be used in step 207.

[0084] If after the second eigenvector is input into the support vector machine, the predicted state detection result of the CNC machine tool output by the support vector machine is a fault state, it means that the CNC machine tool is about to fail (or has already failed). At this time, the staff is required to troubleshoot and repair the CNC machine tool to ensure that the CNC machine tool can run smoothly.

[0085] The following are device embodiments of the present application. For details not described in detail in the device embodiments, reference may be made to the above method embodiments.

[0086] Figure 3 FIG. 1 is a schematic diagram showing a structure of a multi-dimensional state comprehensive detection device for a CNC machine tool provided by an exemplary embodiment of the present disclosure. Figure 3 The multi-dimensional state comprehensive detection device 300 for a numerically controlled machine tool includes: a data acquisition module 301 , a first vector acquisition module 302 , a second vector acquisition module 303 , a splicing module 304 , a dimensionality reduction module 305 and a state detection module 306 .

[0087] The data acquisition module 301 is used to acquire multi-dimensional data of the CNC machine tool, and the multi-dimensional data includes vibration data and audio data.

[0088] The first vector acquisition module 302 is used to acquire a vibration feature vector of each vibration data. The vibration feature vector includes a root mean square, a peak value, a pulse degree, a kurtosis, and a spectrum feature of the vibration data.

[0089] The second vector acquisition module 303 is used to acquire an audio feature vector for each audio data, where the audio feature vector includes Mel-frequency cepstral coefficients and linear prediction cepstral coefficients of the audio data.

[0090] The concatenation module 304 is used to concatenate the vibration feature vectors and the audio feature vectors in pairs to obtain a plurality of first feature vectors.

[0091] The dimension reduction module 305 is used to reduce the dimensions of the multiple first eigenvectors to obtain second eigenvectors.

[0092] The state detection module 306 is used to input the second feature vector into the support vector machine to obtain a predicted state detection result of the CNC machine tool. The state detection result of the CNC machine tool includes a normal state and a fault state.

[0093] Optionally, the dimension reduction module 305 is further used to: use a principal component analysis method to reduce the dimension of the first eigenvector to obtain a second eigenvector.

[0094] Optionally, the second acquisition module 303 is also used to preprocess each audio data, and the preprocessing includes pre-emphasis processing, denoising processing and smoothing processing; performing Fourier transform on the preprocessed audio data to obtain first spectrum data; using a Mel filter to decompose the first spectrum data into multiple Mel frequency bands; logarithmically processing the energy of each Mel frequency band to obtain multiple Mel energy spectra; performing discrete cosine transform on each Mel energy spectrum to obtain multiple Mel-frequency cepstrum coefficients.

[0095] Optionally, the second acquisition module 303 is also used to preprocess each audio data, and the preprocessing includes pre-emphasis processing, frame processing and smoothing processing; perform linear prediction analysis on each preprocessed audio data to obtain multiple prediction coefficients; perform discrete cosine transform on each prediction coefficient to obtain multiple linear prediction cepstrum coefficients.

[0096] It should be noted that: when the CNC machine tool multi-dimensional state comprehensive detection device provided in the above embodiment performs CNC machine tool state detection, only the division of the above functional modules is used as an example. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device is divided into different functional modules to complete all or part of the functions described above. In addition, the CNC machine tool multi-dimensional state comprehensive detection device provided in the above embodiment and the CNC machine tool multi-dimensional state comprehensive detection method embodiment belong to the same concept. The specific implementation process is detailed in the method embodiment, which will not be repeated here.

[0097] The division of modules in the embodiments of the present disclosure is schematic and is only a logical function division. There may be other division methods in actual implementation. In addition, each functional module in each embodiment of the present disclosure may be integrated into a processor, or may exist physically separately, or two or more modules may be integrated into one module. The above-mentioned integrated modules may be implemented in the form of hardware or in the form of software functional modules.

[0098] If the integrated module is implemented in the form of a software function module and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present disclosure is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions to enable a terminal device (which can be a personal computer, mobile phone, or communication device, etc.) or a processor (processor) to execute all or part of the steps of the method of each embodiment of the present disclosure. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc., various media that can store program codes.

[0099] Figure 4 Schematic diagram of the structure of the computer device provided by the embodiment of the present disclosure. Figure 4 As shown, the computer device 400 includes: a processor 401 and a memory 402 .

[0100] The processor 401 may include one or more processing cores, such as a 4-core processor, an 8-core processor, etc. The processor 401 may be implemented in at least one hardware form of DSP (Digital Signal Processing), FPGA (Field-Programmable Gate Array), and PLA (Programmable Logic Array). The processor 401 may also include a main processor and a coprocessor. The main processor is a processor for processing data in the awake state, also known as a CPU (Central Processing Unit); the coprocessor is a low-power processor for processing data in the standby state. In some embodiments, the processor 401 may be integrated with a GPU (Graphics Processing Unit), which is responsible for rendering and drawing the content to be displayed on the display screen. In some embodiments, the processor 401 may also include an AI (Artificial Intelligence) processor, which is used to process computing operations related to machine learning.

[0101] The memory 402 may include one or more computer-readable storage media, which may be non-transitory. The memory 402 may also include a high-speed random access memory, and a non-volatile memory, such as one or more disk storage devices, flash memory storage devices. In some embodiments, the non-transitory computer-readable storage medium in the memory 402 is used to store at least one instruction, which is used to be executed by the processor 401 to implement the multi-dimensional state comprehensive detection method of the CNC machine tool provided in the embodiment of the present disclosure.

[0102] Those skilled in the art will understand that Figure 4 The structure shown in the figure does not constitute a limitation on the computer device 400, and the computer device 400 may include more or less components than those shown in the figure, or combine some components, or adopt a different arrangement of components.

[0103] The embodiment of the present disclosure also provides a non-temporary computer-readable storage medium. When the instructions in the storage medium are executed by the processor of a computer device, the computer device can execute the multi-dimensional state comprehensive detection method for CNC machine tools provided in the embodiment of the present disclosure.

[0104] The embodiments of the present disclosure also provide a computer program product, including a computer program / instruction, which, when executed by a processor, implements the multi-dimensional state comprehensive detection method for a CNC machine tool provided in the embodiments of the present disclosure.

[0105] The above description is only an optional embodiment of the present disclosure and is not intended to limit the present disclosure. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present disclosure shall be included in the protection scope of the present disclosure.

Claims

1. A multi-dimensional state comprehensive detection method for a CNC machine tool, characterized in that: The method comprises: Acquiring multi-dimensional data of a numerically controlled machine tool, wherein the multi-dimensional data includes vibration data and audio data; Acquire a vibration feature vector of each vibration data, wherein the vibration feature vector includes a root mean square, a peak value, a pulse degree, a kurtosis, and a spectrum feature of the vibration data; Acquire an audio feature vector of each of the audio data, wherein the audio feature vector includes a Mel-frequency cepstral coefficient and a linear prediction cepstral coefficient of the audio data; splicing the vibration feature vectors and the audio feature vectors in pairs to obtain a plurality of first feature vectors; Performing dimensionality reduction on the plurality of first eigenvectors to obtain second eigenvectors; The second feature vector is input into a support vector machine to obtain a predicted state detection result of the numerically controlled machine tool, wherein the state detection result of the numerically controlled machine tool includes a normal state and a fault state.

2. The method according to claim 1, characterized in that The step of reducing the dimension of the first feature vector to obtain a second feature vector includes: The first eigenvector is reduced in dimension by using a principal component analysis method to obtain the second eigenvector.

3. The method according to claim 2, characterized in that The second eigenvector is expressed by the following formula: in, is the second eigenvector, is a characteristic matrix composed of multiple first eigenvectors, is a principal component matrix, wherein the principal component matrix includes the first k principal components of the first eigenvector, and the first k principal components are obtained by performing eigenvalue decomposition on the covariance matrix.

4. The method according to any one of claims 1 to 3, characterized in that: The obtaining of the audio feature vector of each audio data comprises: Preprocessing each of the audio data, wherein the preprocessing includes pre-emphasis processing, denoising processing and smoothing processing; Performing Fourier transform on the preprocessed audio data to obtain first spectrum data; Decomposing the first spectrum data into a plurality of Mel frequency bands by using a Mel filter; Performing logarithmic processing on the energy of each Mel frequency band to obtain multiple Mel energy spectra; Perform discrete cosine transform on each of the Mel energy spectra to obtain a plurality of Mel frequency cepstrum coefficients.

5. The method according to any one of claims 1 to 3, characterized in that: The obtaining of the audio feature vector of each audio data comprises: Preprocessing each of the audio data, wherein the preprocessing includes pre-emphasis processing, frame processing and smoothing processing; Performing linear prediction analysis on each of the preprocessed audio data to obtain a plurality of prediction coefficients; Perform discrete cosine transform on each prediction coefficient to obtain a plurality of linear prediction cepstral coefficients.

6. A multi-dimensional state comprehensive detection device for a CNC machine tool, characterized in that: The device comprises: A data acquisition module, used to acquire multi-dimensional data of the CNC machine tool, wherein the multi-dimensional data includes vibration data and audio data; A first vector acquisition module, used for acquiring a vibration feature vector of each vibration data, wherein the vibration feature vector includes a root mean square, a peak value, a pulse degree, a kurtosis, and a spectrum feature of the vibration data; A second vector acquisition module, used for acquiring an audio feature vector of each of the audio data, wherein the audio feature vector includes a Mel-frequency cepstral coefficient and a linear prediction cepstral coefficient of the audio data; A splicing module, used for splicing the vibration feature vectors and the audio feature vectors in pairs to obtain a plurality of first feature vectors; A dimension reduction module, used for reducing the dimension of the first feature vectors to obtain a second feature vector; The state detection module is used to input the second feature vector into a support vector machine to obtain a predicted state detection result of the CNC machine tool, wherein the state detection result of the CNC machine tool includes a normal state and a fault state.

7. The device according to claim 6, characterized in that The dimension reduction module is also used for: The first eigenvector is reduced in dimension by using a principal component analysis method to obtain the second eigenvector.

8. A computer device, characterized in that: The computer device comprises: a memory and a processor, wherein at least one computer program is stored in the memory, and the at least one computer program is loaded and executed by the processor to implement the method according to any one of claims 1 to 5.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores at least one computer program, and the at least one computer program is loaded and executed by a processor to implement the method according to any one of claims 1 to 5.

10. A computer program product comprising a computer program / instructions, characterized in that When the computer program / instructions are executed by a processor, the method according to any one of claims 1 to 5 is implemented.

Images