Tool wear state monitoring method and system based on visual features

Abstract

The application discloses a tool wear state monitoring method and system based on visual features, and relates to the technical field of tool wear monitoring.The method comprises the following steps: acquiring a one-dimensional time domain vibration signal of a tool to be measured in a machining process; converting the one-dimensional time domain vibration signal into a two-dimensional vibration signal-symmetrical point graph based on a symmetrical point graph technology; adaptively extracting a deep visual feature map of the two-dimensional vibration signal-symmetrical point graph by using a wavelet scattering convolution network; and classifying by using a deep convolution neural network model based on the deep visual feature map, and outputting a wear state of the tool to be measured.The application realizes accurate monitoring of the tool wear state, avoids deviation caused by manual screening of features, has high adaptability to multiple working conditions, and has a wide application range.

Description

Technical Field

[0001] This invention relates to the field of tool wear monitoring technology, and in particular to a method and system for monitoring tool wear status based on visual features. Background Technology

[0002] As the part that directly contacts the workpiece during the manufacturing process, the cutting tool plays a crucial role. Tool wear and degradation will severely affect product quality and precision. Therefore, it is necessary to monitor the tool wear condition and take timely maintenance measures when severe wear occurs to ensure machining quality.

[0003] However, due to the harsh machining environment, it is often difficult to directly observe the tool wear condition. Traditional methods use other signals collected during machining to indirectly assess the tool wear condition, such as acoustic emission, cutting force, and cutting power. Among these signals, vibration signals, as characteristic signals of the machine tool, have shown excellent superiority in monitoring tool condition and reflecting machining quality.

[0004] Currently, vibration signal-based fault diagnosis methods are widely used in gears, bearings, motors, and other fields. The fault diagnosis process typically includes signal acquisition, feature extraction, and state recognition. Traditional feature extraction methods based on one-dimensional time, frequency, and time-frequency domains, such as Fourier transform, wavelet transform, Hilbert-Huang transform, variational mode decomposition, and high-frequency resonance techniques, are widely employed. However, these methods focus on extracting one-dimensional detailed features, often requiring the construction of health factors characterizing degradation trends. They are highly dependent on signal processing expertise, easily affected by noise, and the entire feature extraction process is complex and has relatively poor accuracy. Furthermore, after feature extraction, intelligent algorithms are applied for state recognition. Deep learning methods can automatically learn valuable information from raw data and, based on their powerful classification capabilities, separate fault signals from normal signals. However, traditional deep learning methods mostly operate under limited conditions, heavily relying on the selection of processing parameters. Typically, diagnostic performance is better only when processing parameters are constant or vary within a small range. Changes in processing parameters directly affect the data feature space distribution, severely impacting the recognition performance of deep learning methods and resulting in poor final monitoring results. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a method and system for monitoring tool wear status based on visual features. It uses easily measurable vibration signals during machining to indirectly assess tool wear status, which is difficult to measure directly. The acquired one-dimensional time-domain vibration signal is transformed into a two-dimensional vibration signal-symmetric point map using symmetric point mapping technology, avoiding complex data preprocessing. Deep features of the two-dimensional image are extracted using a wavelet scattering convolutional network, and then a deep convolutional neural network model is used to classify the two-dimensional vibration signal-symmetric point map, transforming the status monitoring task into an image classification task. This method not only exhibits good consistency under varying machining parameters but also demonstrates strong robustness even with the addition of Gaussian white noise, ensuring the consistency and accuracy of the final monitoring results.

[0006] In a first aspect, the present invention provides a method for monitoring tool wear status based on visual features.

[0007] A method for monitoring tool wear condition based on visual features, comprising:

[0008] Acquire one-dimensional time-domain vibration signal of the tool under test during machining;

[0009] Based on the symmetric dot plot technique, one-dimensional time-domain vibration signals are transformed into two-dimensional vibration signals-symmetric dot plots;

[0010] Using a wavelet scattering convolutional network, a depth visual feature map of a two-dimensional vibration signal-symmetric point map is adaptively extracted;

[0011] Based on deep visual feature maps, a deep convolutional neural network model is used for classification to output the wear status of the tool under test.

[0012] Secondly, the present invention provides a tool wear condition monitoring system based on visual features.

[0013] A tool wear condition monitoring system based on visual features, comprising:

[0014] The signal acquisition module is used to acquire the one-dimensional time-domain vibration signal of the tool under test during the machining process;

[0015] The image conversion module is used to convert one-dimensional time-domain vibration signals into two-dimensional vibration signals-symmetric dot maps based on symmetric dot plot technology.

[0016] The feature extraction module is used to adaptively extract the depth visual feature map of the two-dimensional vibration signal-symmetric point map using a wavelet scattering convolutional network.

[0017] The condition monitoring module is used to classify the wear status of the tool under test based on the deep visual feature map through a deep convolutional neural network model.

[0018] Thirdly, the present invention also provides an electronic device, including a memory and a processor, and computer instructions stored in the memory and running on the processor, wherein the computer instructions, when executed by the processor, perform the steps of the method described in the first aspect.

[0019] Fourthly, the present invention also provides a computer-readable storage medium for storing computer instructions, which, when executed by a processor, perform the steps of the method described in the first aspect.

[0020] The above one or more technical solutions have the following beneficial effects:

[0021] 1. This invention provides a method and system for monitoring tool wear status based on visual features. It uses easily measurable vibration signals during machining to indirectly assess tool wear status, which is difficult to measure directly. By using symmetric dot plotting technology, the acquired one-dimensional time-domain vibration signal is transformed into a two-dimensional vibration signal-symmetric dot plot. Compared with directly extracting one-dimensional detailed features, this method does not require the construction of health factors representing degradation trends. The signal processing process is simple, the computational load is small, and complex data preprocessing is avoided. Moreover, the transformed two-dimensional image can reflect the relevant information between amplitude and phase in the original signal as well as the effective features hidden in the original signal. The transformed two-dimensional image has good consistency under changes in machining parameters and strong robustness to Gaussian white noise, ensuring the consistency and accuracy of the final monitoring results.

[0022] 2. This invention uses a wavelet scattering convolutional network to perform multi-layer convolution on images, adaptively extracting deep features from two-dimensional images. This reduces inter-class differences and maximizes intra-class features, ensuring that the extracted image features have the same physical meaning, which facilitates subsequent clustering processing of tool wear states. The extracted features satisfy deformation stability and translation invariance, characterizing the multi-scale and multi-directional frequency characteristics of symmetrical point images. The entire process eliminates the need for manually screening health factors sensitive to degradation trends, reducing reliance on prior knowledge. All convolutional kernels in the wavelet scattering convolutional network are pre-selected, so the target features to be extracted are determined before feature extraction from the image, omitting the training process. Features with target properties can be extracted based on theory, greatly reducing the computational load.

[0023] 3. This invention uses a deep convolutional neural network model to classify two-dimensional vibration signals-symmetric point maps. It uses a deep convolutional neural network model to process two-dimensional data, inputs the visual features of the vibration signal-symmetric point map extracted by a wavelet scattering convolutional network, and outputs the final diagnostic result of the tool wear state. By training the deep convolutional neural network model classifier, the state monitoring task is transformed into an image classification task, simplifying the monitoring. Attached Figure Description

[0024] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0025] Figure 1 This is an overall flowchart of the tool wear condition monitoring method based on visual features according to an embodiment of the present invention;

[0026] Figure 2 This is a schematic diagram of the symmetrical dot plot technique in an embodiment of the present invention;

[0027] Figure 3 This is a schematic diagram illustrating the relationship between the angle gain factor ζ, the time delay coefficient l, and the image similarity in an embodiment of the present invention;

[0028] Figure 4 This is a symmetrical point diagram of vibration signals for four tool wear states in an embodiment of the present invention;

[0029] Figure 5 This is a noise immunity diagram of the vibration signal-symmetry point diagram in an embodiment of the present invention;

[0030] Figure 6 This is a symmetrical point diagram of vibration signals under different wear states with different processing parameters in an embodiment of the present invention;

[0031] Figure 7 This is a schematic diagram illustrating the extraction of feature coefficients from a vibration signal-symmetric point image in an embodiment of the present invention;

[0032] Figure 8 This is a framework diagram of a deep convolutional neural network model in an embodiment of the present invention. Detailed Implementation

[0033] It should be noted that the following detailed descriptions are exemplary and are intended only to describe specific embodiments and to provide further explanation of the invention, and are not intended to limit the scope of exemplary embodiments of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0034] Example 1

[0035] This embodiment provides a method for monitoring tool wear status based on visual features, such as... Figure 1 As shown, the specific steps include:

[0036] Acquire one-dimensional time-domain vibration signal of the tool under test during machining;

[0037] Based on the symmetric dot plot technique, one-dimensional time-domain vibration signals are transformed into two-dimensional vibration signals-symmetric dot plots;

[0038] Using a wavelet scattering convolutional network, a depth visual feature map of a two-dimensional vibration signal-symmetric point map is adaptively extracted;

[0039] Based on deep visual feature maps, a deep convolutional neural network model is used for classification to output the wear status of the tool under test.

[0040] The following content provides a more detailed description of the visual feature-based tool wear condition monitoring method proposed in this embodiment.

[0041] Step S1: Acquire the one-dimensional time-domain vibration signal of the tool under test during machining. In this embodiment, triaxial vibration signals of tools in different wear states under different machining parameters are collected. The one-dimensional time-domain vibration signal, which is easy to measure during machining, is used to indirectly evaluate the tool wear state information, which is difficult to measure directly.

[0042] Step S2: Based on the symmetric dot plot technique, the one-dimensional time-domain vibration signal is transformed into a two-dimensional vibration signal-symmetric dot plot. The acquired one-dimensional time-domain vibration signal is a discrete signal, which can be represented as discrete time series data. For the i-th and i+l-th values ​​(l represents the time delay coefficient) in the discrete time series data, a mapping relationship between the Cartesian coordinate system and the polar coordinate system is established. Then, each value in the discrete time series data is traversed to transform the vibration signal into a vibration signal-symmetric dot plot.

[0043] Specifically, such as Figure 2 As shown, by using the symmetric dot plot technique, a mapping relationship between the rectangular coordinate system and the polar coordinate system is established based on the i-th and (i+l)-th values ​​in the discrete time series, representing the vibration signal in the form of a symmetric dot plot, that is, transforming the vibration signal into a vibration signal-symmetric dot plot. For the discrete signal x(i), the pole plot P(i) can be formulated from the original signal as:

[0044]

[0045]

[0046]

[0047]

[0048] Where, x i Let x represent the amplitude of the i-th signal in the time-domain vibration signal. i+l It is related to the time-domain vibration signal x i The corresponding signal amplitude with a time delay coefficient l; x min and xmax Represents the time-domain vibration signal x i and x i+l The minimum and maximum values ​​in the window; r(i) represents the radius of the polar coordinate space, θ is the rotation angle of the mirror symmetry plane, θ(i) and φ(i) are the counterclockwise and clockwise rotation angles of the mirror symmetry plane, l represents the time delay coefficient, ζ represents the angular gain factor (θ≥ζ), N is the total number of symmetry planes; θ is chosen as 60° to make the symmetry point image a snowflake-shaped six-fold symmetry pattern, which facilitates a clear description of the signal characteristics.

[0049] Furthermore, the subtle differences in the symmetrical point maps obtained from vibration signal conversion under different wear conditions mainly depend on the selection of the time delay coefficient l and the angular gain factor ζ. Typically, l ranges from 1 to 10, and ζ ranges from 0° to 30°. Based on the principle of maximizing image difference, the similarity between the symmetrical point maps obtained from vibration signal conversion under different wear conditions is calculated. Based on the calculated image similarity, the optimal parameter values ​​for the symmetrical point maps are determined to be l = 10 and ζ = 5°.

[0050] Specifically, the process of calculating image similarity and selecting the optimal parameters based on this similarity is as follows: According to the above formula for generating the pole graph P(i), under the same wear condition, different pole graphs P(i) can be obtained by adjusting the time delay coefficient l and the angular gain factor ζ. Based on the principle of maximizing image differences, that is, under the same time delay coefficient l and angular gain factor ζ, the similarity of the pole graphs P(i) under different wear conditions is minimized, thereby selecting the optimal parameter values. Figure 3 To determine the image similarity for different time delay coefficients l and angular gain factors ζ, the time delay coefficient l and angular gain factor ζ with the lowest similarity are selected. The image similarity calculation formula is as follows:

[0051]

[0052]

[0053]

[0054] In the above formula, A is the grayscale matrix of image 1, and B is the grayscale matrix of image 2. B' is the average value of the grayscale matrix of image 1, B' is the average value of the grayscale matrix of image 2, and r is the similarity between the two images.

[0055] Furthermore, the different wear states of the cutting tools are divided into new tool (NT), initial wear VB = 0.1 mm (IW), intermediate wear VB = 0.2 mm (MW), and severe wear VB = 0.3 mm (SW). The vibration signal-symmetry point diagram of the four tool wear states is shown in the figure. Figure 4 As shown.

[0056] (1) With constant processing parameters, additive white Gaussian noise in the range of 20dB to 50dB is added to the original measurement signal at intervals of 10dB signal-to-noise ratio (SNR), and the mixed signal is converted into a symmetrical point image. Figure 5 Vibration signal-symmetry point images under different noise interferences are displayed. A comparison of the original signal image and the symmetry point image of the mixed noise signal shows that the images under different noise interferences in the same wear state have similar shapes with only minor differences. The main difference lies in the slight change in the concentrated area of ​​the central footprint in the symmetry point image as the noise increases. However, there are still significant differences in the symmetry point images under different wear states, mainly in the thickness and curvature of the image arms. It can be seen that the symmetry point images under noise interference still maintain significant intra-class similarity and inter-class inconsistency. Therefore, the vibration signal-symmetry point image method proposed in this embodiment not only has strong noise resistance but also more intuitively reveals the degradation characteristics of tools under different wear states.

[0057] (2) Changing the processing parameters under the same wear condition, such as Figure 6 As shown, images under different processing parameters for the same wear condition exhibit high similarity, with only slight variations in thickness and dispersion. However, significant separability remains between different wear conditions, and the differences between them are not disrupted. Therefore, this demonstrates that the vibration signal-symmetric point image method proposed in this embodiment has strong parameter independence.

[0058] Step S3: Adaptively extract the depth visual feature map of the two-dimensional vibration signal-symmetric point map using a wavelet scattering convolutional network. That is, input the two-dimensional vibration signal-symmetric point map into a three-layer wavelet scattering convolutional network, calculate the scattering coefficient of each layer, use the scattering coefficient as the extracted image feature, and then stitch the image features together to obtain the depth visual feature map of the vibration signal-symmetric point map.

[0059] Specifically, assume the wavelet scattering path is P = {λ1, λ2, λ3, ..., λ m}, where m is the path length of the scattering, then the output scattering coefficients of each layer in the network are the texture features extracted from the input vibration signal-symmetric point image. The scale map coefficients are obtained by iterative wavelet mode convolution, and are:

[0060]

[0061] Where f is the input vibration signal-symmetric point image information, t is time, and λ1 is a first-order wavelet. It is an m-order wavelet (0≤m≤J), where J is the maximum path length of the network.

[0062] The wavelet scattering coefficient is defined as:

[0063]

[0064] Where, φ J (t) is the scaling function, S0g(t) = f*φ J (t) are the 0th-order wavelet scattering coefficients. The final scattering vector consists of all scattering coefficients 0 ≤ m ≤ J, and its expression is: Sf = (S m f) 0≤m≤J .

[0065] In this embodiment, a three-layer wavelet scattering convolutional network is used to extract features from the transformed vibration signal-symmetry point image. Each layer directly calculates wavelet scattering coefficients of different orders on the input two-dimensional vibration signal-symmetry point image to extract image features. Specifically, layer 0 extracts the basic features of the vibration signal-symmetry point image but loses signal details; layer 1 captures detailed information, similar to a scale-invariant feature transform function; and layer 2 provides supplementary information to improve classification. Furthermore, since layer 0 uses a wavelet low-pass filter to average the input signal, high-frequency details are lost. Therefore, layers 1 and 2 continue to perform continuous wavelet transformation based on layer 0 to extract more detailed feature information. The visual feature maps of the vibration signal-symmetry point image extracted from different layers of the wavelet scattering convolutional network are shown below. Figure 7 As shown, the scattering coefficients of each layer are used as the extracted image features, and the final depth visual feature map is obtained by feature stitching.

[0066] Step S4: Based on the deep visual feature map, classify the data using a deep convolutional neural network model and output the wear status of the tool under test.

[0067] Specifically, a deep convolutional neural network model is used to classify vibration signals and symmetric point maps, such as... Figure 8 As shown, the model includes one input layer, two convolutional layers, two pooling layers, one fully connected layer, and one output layer. For the deep convolutional neural network model, the input data is the feature map of the vibration signal-symmetric point image extracted by the wavelet scattering convolutional network. The image is preprocessed in the input layer (this preprocessing is actually feature concatenation of the three layers of input image features to obtain a depth visual feature map). Then, in the convolutional layers, multiple feature maps are generated in one convolutional layer using different convolutional kernels. The feature map is output through a non-linear activation function, where the feature image C output by the k-th feature map is... k Represented as:

[0068]

[0069] In the above formula, x represents the input image, W k b represents the convolution kernel associated with the k-th feature map.k Indicates the paranoid aspect. represents a two-dimensional convolution operation, represents the inner product of the kernel and the input image, and f(·) represents the activation function that performs a nonlinear transformation on the generated features.

[0070] Subsequently, max pooling is used in the pooling layer to fuse similar local features, reduce the dimensionality of the feature mapping, and improve the robustness of feature learning.

[0071] Repeat the convolutional and pooling layer operations again, and flatten the output feature map before inputting it into the fully connected layer; in the fully connected layer, according to y=σ(Wa+b c The algorithm classifies the data and applies a softmax function to the output layer to normalize the output of the fully connected layer, resulting in the final diagnostic result of the tool wear state. Here, y represents the prediction result, σ(·) represents the softmax activation function commonly used in classification tasks, a represents a neuron in the fully connected layer connected to the output layer, W is the weight matrix between neurons a, and b... c This indicates a paranoid term, where each cell in the layer is connected to all cells from the previous layer.

[0072] This embodiment uses the above method to monitor the wear condition of the cutting tool, avoiding the deviation caused by manual selection of features. It has high adaptability to multiple working conditions and a wide range of applications.

[0073] Example 2

[0074] This embodiment provides a tool wear condition monitoring system based on visual features, including:

[0075] The signal acquisition module is used to acquire the one-dimensional time-domain vibration signal of the tool under test during the machining process;

[0076] The image conversion module is used to convert one-dimensional time-domain vibration signals into two-dimensional vibration signals-symmetric dot maps based on symmetric dot plot technology.

[0077] The feature extraction module is used to adaptively extract the depth visual feature map of the two-dimensional vibration signal-symmetric point map using a wavelet scattering convolutional network.

[0078] The condition monitoring module is used to classify the wear status of the tool under test based on the deep visual feature map through a deep convolutional neural network model.

[0079] Example 3

[0080] This embodiment provides an electronic device, including a memory and a processor, as well as computer instructions stored in the memory and running on the processor. When the processor executes the computer instructions, it completes the steps in the visual feature-based tool wear condition monitoring method described above.

[0081] Example 4

[0082] This embodiment also provides a computer-readable storage medium for storing computer instructions, which, when executed by a processor, complete the steps in the visual feature-based tool wear condition monitoring method described above.

[0083] The steps and methods involved in Embodiments 2 to 4 above correspond to those in Embodiment 1. For specific implementation details, please refer to the relevant description section of Embodiment 1. The term "computer-readable storage medium" should be understood as a single medium or multiple media including one or more instruction sets; it should also be understood as including any medium capable of storing, encoding, or carrying an instruction set for execution by a processor and enabling the processor to perform any of the methods in this invention.

[0084] Those skilled in the art will understand that the modules or steps of the present invention described above can be implemented using general-purpose computer devices. Optionally, they can be implemented using computer-executable program code, thereby allowing them to be stored in a storage device for execution by a computer device, or they can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. The present invention is not limited to any particular combination of hardware and software.

[0085] The above description is only a preferred embodiment of the present invention. Although the specific implementation of the present invention has been described in conjunction with the accompanying drawings, it is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that, based on the technical solution of the present invention, various modifications or variations that can be made by those skilled in the art without creative effort are still within the scope of protection of the present invention.

Claims

1. A method for monitoring tool wear condition based on visual features, characterized in that, include: Acquire one-dimensional time-domain vibration signal of the tool under test during machining; Based on the symmetric dot plot technique, one-dimensional time-domain vibration signals are transformed into two-dimensional vibration signals-symmetric dot plots; Using a wavelet scattering convolutional network, a depth visual feature map of a two-dimensional vibration signal-symmetric point map is adaptively extracted; Based on deep visual feature maps, a deep convolutional neural network model is used for classification to output the wear status of the tool under test. Among them, based on the symmetric dot plot technique, one-dimensional time-domain vibration signals are transformed into two-dimensional vibration signals—symmetric dot plots—including: The acquired one-dimensional time-domain vibration signal is a discrete signal, which is represented as discrete time series data. Specifically, regarding the first... i and i + l We establish a mapping relationship between the rectangular coordinate system and the polar coordinate system using several values; among them... l Indicates the time delay coefficient; Iterate through each value in the discrete time series data to transform the vibration signal into a vibration signal-symmetric point plot; transform the discrete vibration signal... Convert to vibration signal - symmetric pole figure The formula is: ； ； ； ； In the above formula, Represents the first time-domain vibration signal. i The amplitude of the signal, It is related to time-domain vibration signals Correspondingly, it has a time delay coefficient l The signal amplitude; and Represents time-domain vibration signal and The minimum and maximum values ​​in the window; Represents the radius in polar coordinate space. For the rotation angle of the mirror-symmetric plane, and It refers to the counterclockwise and clockwise rotation angles of the mirror-symmetric plane. Represents the time delay coefficient. Represents the angular gain factor and , N This represents the total number of symmetrical planes.

2. The tool wear condition monitoring method based on visual features as described in claim 1, characterized in that, Based on the principle of maximizing image differences, the similarity between symmetrical point maps obtained by converting vibration signals under different wear conditions is calculated. Based on the calculated image similarity, the optimal parameter values ​​of the symmetrical point maps are determined and obtained.

3. The tool wear condition monitoring method based on visual features as described in claim 1, characterized in that, The two-dimensional vibration signal-symmetric point map is input into a three-layer wavelet scattering convolutional network. The scattering coefficient of each layer is calculated, and the scattering coefficient is used as the extracted image feature. The image features are then stitched together to obtain the depth visual feature map of the vibration signal-symmetric point map.

4. The tool wear condition monitoring method based on visual features as described in claim 1, characterized in that, The deep convolutional neural network model includes an input layer, two convolutional layers, two pooling layers, a fully connected layer, and an output layer. The feature map is input into the deep convolutional neural network model. First, image preprocessing is performed in the input layer. Then, multiple feature maps are generated in the convolutional layer using different convolutional kernels. The feature map is output through a non-linear activation function. Finally, similar local features are fused using the max pooling method in the pooling layer. The convolutional and pooling operations are repeated, and the output feature map is flattened before being input into the fully connected layer. Finally, the fully connected layer is used for classification, and the output layer is normalized to output the wear state of the tool under test.

5. A tool wear condition monitoring system based on visual features, characterized in that, include: The signal acquisition module is used to acquire the one-dimensional time-domain vibration signal of the tool under test during the machining process; The image conversion module is used to convert one-dimensional time-domain vibration signals into two-dimensional vibration signals-symmetric dot maps based on symmetric dot plot technology. The feature extraction module is used to adaptively extract the depth visual feature map of the two-dimensional vibration signal-symmetric point map using a wavelet scattering convolutional network. The condition monitoring module is used to classify the wear status of the tool under test based on the deep visual feature map through a deep convolutional neural network model. Among them, based on the symmetric dot plot technique, one-dimensional time-domain vibration signals are transformed into two-dimensional vibration signals—symmetric dot plots—including: The acquired one-dimensional time-domain vibration signal is a discrete signal, which is represented as discrete time series data. Specifically, regarding the first... i and i + l We establish a mapping relationship between the rectangular coordinate system and the polar coordinate system using several values; among them... l Indicates the time delay coefficient; Iterate through each value in the discrete time series data to transform the vibration signal into a vibration signal-symmetric point plot; transform the discrete vibration signal... Convert to vibration signal - symmetric pole figure The formula is: ； ； ； ； In the above formula, Represents the first time-domain vibration signal. i The amplitude of the signal, It is related to time-domain vibration signals Correspondingly, it has a time delay coefficient l The signal amplitude; and Represents time-domain vibration signal and The minimum and maximum values ​​in the window; Represents the radius in polar coordinate space. For the rotation angle of the mirror-symmetric plane, and It refers to the counterclockwise and clockwise rotation angles of the mirror-symmetric plane. Represents the time delay coefficient. Represents the angular gain factor and , N This represents the total number of symmetrical planes.

6. An electronic device, characterized in that, It includes a memory and a processor, as well as computer instructions stored in the memory and running on the processor, which, when executed by the processor, complete the steps of a tool wear condition monitoring method based on visual features as described in any one of claims 1-4.

7. A computer-readable storage medium, characterized in that, Used to store computer instructions, which, when executed by a processor, complete the steps of a tool wear condition monitoring method based on visual features as described in any one of claims 1-4.

Images