Boring quality inspection method and related hardware
By performing signal conditioning and feature analysis on the acceleration signal during the boring process, and using a pre-trained model to judge the boring quality, the problem of inaccuracy caused by measurement error and manual judgment in boring quality inspection is solved, and efficient and accurate quality inspection results are achieved.
Patent Information
- Application Number
- CN202511508354.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-10-21
- Publication Date
- 2026-01-09
Smart Images
Figure CN121290165A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of machining defect detection technology, and more particularly to a boring quality inspection method and related hardware. Background Technology
[0002] Currently, during the boring process, manual measurement of the machined hole is required to check whether the machining parameters meet the standards. This method is subject to measurement errors and uncertainties caused by subjective human judgment. Summary of the Invention
[0003] This invention provides a boring quality inspection method and related hardware to solve the problem in the prior art where the quality inspection results of the machined holes obtained by boring are easily inaccurate due to factors such as measurement errors.
[0004] In a first aspect, embodiments of the present invention provide a boring quality inspection method, comprising: Acquire the acceleration signal generated by the target object in the target direction during the boring process of the machine tool on the target workpiece; The acceleration signal is conditioned to obtain the target time-domain signal; The target time-domain signal is input into a pre-trained vibration feature analysis model to obtain the output result determined by the vibration feature analysis model; Based on the output results, determine whether the actual machining parameters of the target hole obtained by boring the target workpiece are abnormal.
[0005] Optionally, the step of conditioning the acceleration signal to obtain the target time-domain signal includes: The acceleration signal is transformed into a first frequency domain signal; The first frequency domain signal is filtered in the frequency domain to obtain the second frequency domain signal; The second frequency domain signal is transformed into a first time domain signal; The first time-domain signal is nonlinearly amplified to obtain the target time-domain signal.
[0006] As an optional implementation, the output result is the actual machining parameters of the target hole obtained by boring the target workpiece; The step of determining whether the machining parameters of the target hole obtained by boring the target workpiece are abnormal based on the output result includes: Based on the numerical relationship between the actual processing parameters and the target processing parameters of the target hole, determine whether the actual processing parameters are abnormal.
[0007] As another optional implementation, the output result is a classification result indicating whether the actual machining parameters of the target hole obtained by boring the target workpiece are abnormal.
[0008] Optionally, the vibration feature analysis model sequentially includes an input layer, a TimesNet layer, a pooling layer, a fully connected layer, and an output layer; the TimesNet layer includes multiple sequentially connected time blocks (TimeBlocks). The step of inputting the target features into a pre-trained vibration feature analysis model to obtain the classification result output by the vibration feature analysis model includes: The target features are embedded through the input layer to obtain an embedding vector; The embedded vector is processed by the TimesNet layer to obtain global temporal features. Specifically, each TimesBlock performs the following steps sequentially according to the connection order: an FFT is applied to the one-dimensional input vector to obtain frequency domain features; the target feature period and corresponding period weights are determined based on the frequency domain features; for any target feature period, the input vector is reshaped according to the target feature period to obtain a two-dimensional first intermediate feature tensor; for any first intermediate feature tensor, a convolutional neural network (CNN) is used to transform the first intermediate feature tensor to obtain a two-dimensional second intermediate feature tensor; for any second intermediate feature tensor, the second intermediate feature tensor is reshaped to obtain a one-dimensional intermediate temporal feature; the intermediate temporal features corresponding to each target feature period are fused according to the corresponding period weights to obtain output features; wherein the input vector of the first-level TimesBlock is the embedded vector, the input vector of non-first-level TimesBlocks is the output feature of the previous-level TimesBlock, and the global temporal feature is the output feature of the last-level TimesBlock. The global temporal features are compressed in terms of temporal dimension through the pooling layer to obtain globally compressed features. The global compressed features are reduced in dimensionality using the fully connected layer to obtain a global dimensionality-reduced vector. The output layer determines the output result based on the global dimensionality reduction vector.
[0009] Optionally, the target object includes at least one of the following: The machine tool body, the target workpiece, the target tooling for fixing the target workpiece, the target tool for boring, and the target motion mechanism for moving the target tool or the target workpiece.
[0010] Secondly, based on the same inventive concept, embodiments of the present invention also provide a boring quality inspection device, comprising: The data acquisition module is used to acquire the acceleration signal generated by the target object in the target direction during the boring process of the machine tool on the target workpiece; The signal processing module is used to perform signal conditioning on the acceleration signal to obtain the target time-domain signal; The analysis module is used to input the target time-domain signal into a pre-trained vibration feature analysis model to obtain the output result determined by the vibration feature analysis model; and to determine whether the actual machining parameters of the target hole obtained by boring the target workpiece are abnormal based on the output result.
[0011] Thirdly, based on the same inventive concept, embodiments of the present invention also provide an electronic device, including: a processor and a memory for storing processor-executable instructions; The processor is configured to execute the instructions to implement the boring quality inspection method as described in the first aspect.
[0012] Fourthly, based on the same inventive concept, embodiments of the present invention also provide a computer-readable storage medium storing computer program code, which, when executed on a computer, causes the computer to perform the boring quality inspection method as described in the first aspect.
[0013] Fifthly, based on the same inventive concept, embodiments of the present invention also provide a computer program product, the computer program product comprising: computer program code, which, when the computer program code is run on a computer, causes the computer to execute the boring quality inspection method described in the first aspect.
[0014] The beneficial effects of this invention are as follows: The boring quality inspection method and related hardware provided in this invention analyze the vibration signals of the machining equipment and / or the target workpiece during the boring process. This allows for the determination of whether the machining parameters of the target hole are abnormal without directly measuring them. This avoids measurement errors and subjective judgments by inspection personnel that can lead to inaccurate results, thus improving the accuracy of quality inspection. Furthermore, the method can be automated to provide classification results, reducing the workload of inspection personnel and enabling simultaneous production and inspection, thereby improving production efficiency. Attached Figure Description
[0015] Figure 1 A flowchart of a boring quality inspection method provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of signal transformation in the boring quality inspection method provided in an embodiment of the present invention; Figure 3This is one of the flowcharts of the boring quality inspection method provided in the embodiments of the present invention; Figure 4 These are signal diagrams corresponding to normal holes with some actual processing parameters in embodiments of the present invention; Figure 5 These are signal diagrams corresponding to holes with abnormal actual processing parameters in some embodiments of the present invention; Figure 6 This is a second partial flowchart of the boring quality inspection method provided in the embodiments of the present invention; Figure 7 This is the third partial flowchart of the boring quality inspection method provided in the embodiments of the present invention; Figure 8 This is a schematic diagram of the boring quality inspection device provided in an embodiment of the present invention; Figure 9 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation
[0016] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, the present invention will be further described below in conjunction with the accompanying drawings and embodiments. However, the exemplary embodiments can be implemented in many forms and should not be construed as limited to the embodiments set forth herein; rather, these embodiments are provided to make the present invention more comprehensive and complete, and to fully convey the concept of the exemplary embodiments to those skilled in the art. The same reference numerals in the figures denote the same or similar structures, and therefore repeated descriptions of them will be omitted. Terms describing position and direction in the present invention are illustrative based on the accompanying drawings, but changes can be made as needed, and all such changes are included within the scope of protection of the present invention. The accompanying drawings of the present invention are for illustrative purposes only and do not represent actual proportions.
[0017] It should be noted that specific details are set forth in the following description to provide a full understanding of the invention. However, the invention can be practiced in many ways other than 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. The following description is a preferred embodiment for carrying out the present application; however, the description is for the purpose of illustrating the general principles of the application and is not intended to limit the scope of the application. The scope of protection of this application shall be determined by the appended claims.
[0018] The boring quality inspection method and related hardware provided in the embodiments of the present invention will be described in detail below with reference to the accompanying drawings.
[0019] This invention provides a boring quality inspection method, such as... Figure 1 As shown, it includes: S110. Acquire the acceleration signal generated by the target object in the target direction during the boring process of the machine tool on the target workpiece.
[0020] In the specific implementation process, since the target workpiece and related parts of the machine tool will generate a certain vibration during the boring process of the machine tool, the acceleration signals of these components that will vibrate can be collected and analyzed in subsequent steps to determine whether abnormal vibration occurs during the boring process, resulting in abnormal machining parameters of the target hole (including the size and roughness of the target hole).
[0021] Optionally, the target object includes at least one of the following: The machine tool body, the target workpiece, the target tooling for fixing the target workpiece, the target tool for boring, and the target motion mechanism for moving the target tool or the target workpiece.
[0022] Accordingly, accelerometers can be installed on the target object to measure the acceleration signal generated by the vibration of the target object. For example, accelerometers can be installed at the front end of the spindle used to drive the target tool, the end of the target tool used for boring, and the target tooling fixture used to fix the target workpiece to collect acceleration signals.
[0023] In practical implementation, acceleration signals generated by the target object in at least one direction (i.e., the target direction) can be collected to obtain the vibration information of the target object. The target direction may include the Y direction (i.e., the axial direction) of the machine tool coordinate system, and may further include the X and Z directions of the machine tool coordinate system, etc., without further limitation in this embodiment of the invention. If the target direction simultaneously includes the X, Y, and Z directions of the machine tool coordinate system, then a three-axis accelerometer can be used to collect the acceleration signals of the target object in each target direction.
[0024] In practice, during the machining process, the target tool only performs boring operations on the workpiece for a portion of the time. The remaining time involves operations such as tool feed or retraction (not directly boring) and other machining operations performed by the machine tool. Therefore, if... Figure 2As shown, the acceleration signal corresponding to the boring process can be extracted from the raw acceleration signal collected by the accelerometer and then further processed. For example, the target time period corresponding to the boring process of the target tool can be determined based on parameters such as the size of the machine tool, the motion parameters of the target tool (including motion speed, motion direction, motion time, etc.), and the size of the target workpiece. The acceleration signal corresponding to the target time period can then be extracted as the acceleration signal corresponding to the boring process. As another example, considering that the actual power consumed by the machine tool varies during different operations in the machining process, the target time period corresponding to the actual power of the target tool's boring process can be determined based on the actual power variation of the machine tool. The acceleration signal corresponding to the target time period can then be extracted as the acceleration signal corresponding to the boring process. For example, considering that the vibration amplitude generated when the tool is not in contact with the workpiece during machining is usually smaller than the vibration amplitude generated during machining when the tool is in contact with the workpiece (e.g., boring), the raw acceleration signal collected by the accelerometer can be divided into preset sampling periods, and the average value of the raw acceleration signal corresponding to each preset sampling period can be determined. The sampling period corresponding to the maximum average value is determined as the target sampling period. Starting from the target sampling period corresponding to the maximum average value, the adjacent sampling periods earlier than the current earliest target sampling period are sequentially determined as the first undetermined sampling periods. It is then determined whether the difference between the average value of the raw acceleration signal corresponding to the first undetermined sampling period and the maximum average value is within a preset fluctuation range. If the difference is within the preset fluctuation range, the first undetermined sampling period is then... The sampling period is determined as the target sampling period, and continues until the difference between the average value and the maximum average value of the original acceleration signal corresponding to the first undetermined sampling period is no longer within a preset fluctuation range. Starting from the target sampling period corresponding to the maximum average value, adjacent sampling periods later than the current latest target sampling period are sequentially determined as the second undetermined sampling periods. It is then determined whether the difference between the average value and the maximum average value of the original acceleration signal corresponding to the second undetermined sampling period is within a preset fluctuation range. If the difference is within the preset fluctuation range, the second undetermined sampling period is determined as the target sampling period, and this process continues until the difference between the average value and the maximum average value of the original acceleration signal corresponding to the second undetermined sampling period is no longer within the preset fluctuation range. All acceleration signals corresponding to the target sampling periods are then determined as the acceleration signals corresponding to the boring process. This embodiment of the invention does not impose further limitations.
[0025] S120. Perform signal conditioning on the acceleration signal to obtain the target time domain signal.
[0026] In practical implementation, signal conditioning of the acceleration signal may include time-domain filtering, frequency-domain filtering, amplitude amplification, etc., to enhance the representation of characteristic information in the acceleration signal regarding whether the actual machining parameters of the target hole obtained by boring are abnormal, thus facilitating subsequent processing. Optionally, such as Figure 3As shown, step S120 specifically includes: S121. Convert the acceleration signal into a first frequency domain signal.
[0027] In practical implementation, Fourier transform can be used to transform the acceleration signal into a vibration frequency domain signal. Specifically, Fast Fourier Transform (FFT) can be used to transform the acceleration signal into a vibration frequency domain signal.
[0028] S122. Perform frequency domain filtering on the first frequency domain signal to obtain the second frequency domain signal.
[0029] In practical implementation, the first frequency domain signal can be filtered using a preset window through methods such as mean filtering and root mean square (RMS) filtering to determine the second frequency domain signal. One possible implementation is to apply a 10Hz preset window to the first frequency domain signal using sliding RMS filtering to obtain the second frequency domain signal.
[0030] S123. Transform the second frequency domain signal into the first time domain signal.
[0031] In the specific implementation process, Fourier transform can be used to transform the second frequency domain signal into the first time domain signal, which can be achieved by FFT transform.
[0032] S124. Perform amplitude nonlinear amplification on the first time-domain signal to obtain the target time-domain signal. (e.g.) Figure 2 (As shown)
[0033] For example, the amplitude of the first time-domain signal can be amplified by a fixed factor (e.g., 200 times), and then the amplified amplitude can be squared again to obtain the target time-domain signal. By amplifying the amplitude of the first time-domain signal, the hidden vibration characteristics in the first time-domain signal can be effectively highlighted (manifested as signal spikes in the target time-domain signal), so as to facilitate the subsequent extraction of features from the target time-domain signal for further processing.
[0034] like Figure 4 and Figure 5 As shown, there are certain differences between the target time-domain signals of holes with normal actual processing parameters and holes with abnormal actual processing parameters. For example, the amplitude and abscissa of the peaks in the target time-domain signals are different. Based on this, the target time-domain signals can be input into the vibration characteristic analysis model for analysis to determine whether the processing parameters of the target holes are abnormal.
[0035] S130. Input the target features into the pre-trained vibration feature analysis model to obtain the output results determined by the vibration feature analysis model.
[0036] S140. Determine whether the actual machining parameters of the target hole obtained by boring the target workpiece are abnormal based on the output results.
[0037] As an optional implementation, the vibration characteristic analysis model is a regression model. The output result determined by the vibration characteristic analysis model is the actual processing parameters of the target hole. Therefore, step S140 specifically includes: determining whether the actual processing parameters are abnormal based on the numerical relationship between the actual processing parameters and the target processing parameters of the target hole. For example, if the processing parameter is the diameter of the target hole, if the actual diameter meets the target processing accuracy corresponding to the target diameter, then the actual diameter is determined to be normal; if the actual diameter exceeds the target processing accuracy corresponding to the target diameter, then the actual diameter is determined to be abnormal.
[0038] As an alternative implementation, the vibration characteristic analysis model is a classification model. The output of the vibration characteristic analysis model is a classification result indicating whether the actual processing parameters of the target hole are abnormal. Accordingly, the conclusion of whether the actual processing parameters of the target hole are abnormal can be directly obtained from the classification result.
[0039] In practical implementation, the vibration feature analysis model can be implemented using Convolutional Neural Network (CNN), Support Vector Machine (SVM), Decision Tree, Random Forest, etc., and this embodiment of the invention does not impose any further limitations. For example, if the vibration feature analysis model is a classification model, then it can employ models such as Gated Recurrent Unit (GRU), Temporal Convolutional Networks (TCN), Spatio-Temporal Gated Network (STGN), Transformer, Inverted Transformer (iTransformer), Frequency Enhanced Decomposed Transformer (FEDFormer), and TimesNet.
[0040] In practical implementation, the vibration feature analysis model is trained using the target time-domain signal as the input (features of the training samples) and the output (labels of the training samples). Specifically, the raw data generated during the boring process of a machine tool can be processed to construct machine learning samples. The sample features are the target time-domain signals, which are obtained through the signal conditioning process described in step S120 above. If the vibration feature analysis model is a regression model, the training sample labels are the actual machining parameters of the hole obtained from the boring process; if the vibration feature analysis model is a classification model, the training sample labels are the classification categories indicating whether the actual machining parameters of the hole obtained from the boring process are abnormal. Subsequently, the machine learning samples are divided into training samples, test samples, and validation samples. The vibration feature analysis model is trained on the training samples for at least one round. After each round of training, the vibration feature analysis model is validated using validation samples to determine whether the training termination condition is met. If the model training termination condition is not met based on the validation results (e.g., the validation index does not meet the preset index value range, and / or the preset number of training rounds is not reached), the model parameters of the vibration feature analysis model are adjusted (e.g., the model parameters are adjusted using gradient descent), and another round of training is performed. If the model training condition is met based on the validation results, the training ends, and the vibration feature analysis model is tested using test samples to evaluate the generalization ability of the vibration feature analysis model. In the process of partitioning machine learning samples, the hold-out method can be used to partition the samples, dividing multiple machine learning samples into training sample sets, validation sample sets, and test sample sets according to a certain ratio (for example, selecting 3178 machine learning samples as training samples, 681 machine learning samples as validation samples, and 681 machine learning samples as test samples). Alternatively, cross-validation can be used to partition the machine learning samples, dividing the machine learning samples into multiple sample sets, selecting a fixed sample set as the test sample set, and selecting one sample set from the remaining sample sets as the test sample set in each round of training, with the remaining sample sets forming the training sample set. The test sample set is different in different rounds of training.
[0041] Thus, the boring quality inspection method provided in this embodiment of the invention analyzes the vibration signals of the machining equipment and / or the target workpiece during the boring process. This allows for the determination of whether the machining parameters of the target hole are abnormal without directly measuring them. This avoids measurement errors and subjective judgments by inspection personnel that can lead to inaccurate results, thereby improving the accuracy of quality inspection. Furthermore, the method can be automated by equipment to provide classification results, reducing the workload of inspection personnel and enabling simultaneous production and inspection, thus improving production efficiency.
[0042] Furthermore, alternatively, such as Figure 1 As shown, after step S130, the method includes: S200. Construct machine learning samples based on target features. The machine learning samples are training samples, validation samples, or test samples.
[0043] In practical implementation, the vibration feature analysis model requires significant computing power during inference. However, the electronic equipment deployed on the production site (including machine tools, industrial controllers, etc.) may be limited by hardware computing performance and unable to run the vibration feature analysis model. Therefore, a cloud-edge architecture control system can be set up to implement the method provided in this embodiment of the invention. The industrial controller located on the production site acts as the edge side to collect acceleration signals, perform preprocessing (e.g., extracting acceleration signals corresponding to the boring process), and then report them to the cloud. The cloud then deploys the vibration feature analysis model. The workpiece number can be used to establish a correspondence between the workpiece data (e.g., the specific values of the actual processing parameters of the target workpiece, and the classification results of whether the actual processing parameters of the target workpiece are abnormal) stored in the offline database of the industrial controller or other equipment and the target time-domain signal of the target workpiece stored online in the cloud. Then, based on the target time-domain signal corresponding to the same workpiece, the sample features of the machine learning sample are used, and the workpiece data stored in the offline database (e.g., the specific values of the actual processing parameters of the target workpiece, and the classification results of whether the actual processing parameters of the target workpiece are abnormal) are used as the sample labels of the machine learning sample to construct machine learning samples.
[0044] S300. When the model training conditions are met, machine learning samples are used to perform machine learning on the vibration feature analysis model.
[0045] In specific implementation, model training conditions may include receiving a model training instruction triggered by the user, the number of machine learning samples reaching a preset number of samples, etc. This embodiment of the invention does not impose too many limitations here.
[0046] In this way, by continuously collecting real data to build machine learning samples during the operation of the scheme, and using these machine learning samples to retrain the vibration feature analysis model, the trained vibration feature analysis model can better fit the real situation.
[0047] One possible implementation is that the vibration feature analysis model is implemented using a machine learning model based on the TimesNet model. Specifically, the vibration feature analysis model sequentially includes an input layer, a TimesNet layer, a pooling layer, a fully connected layer, and an output layer. The TimesNet layer includes multiple sequentially connected time blocks (TimeBlocks).
[0048] Accordingly, such as Figure 6As shown, step S130 specifically includes: S131. The target features are embedded through the input layer to obtain the embedding vector.
[0049] S132. The embedding vector is processed through the TimesNet layer to obtain global temporal features.
[0050] Specifically, such as Figure 7 As shown, step S132 specifically includes: S1321. Select the current TimesBlock in the order in which they are connected.
[0051] If the current TimesBlock is successfully selected, proceed to step S1322; if all TimesBlocks in the TimesNet layer have been selected, proceed to step S1326.
[0052] S1322. Perform FFT transformation on the one-dimensional input vector to obtain frequency domain features, and determine the target feature period and the period weight corresponding to each target feature period based on the frequency domain features.
[0053] The input vector of the first-level TimesBlock is the embedding vector, while the input vector of the non-first-level TimesBlock is the output feature of the previous-level TimesBlock.
[0054] In practical implementation, a preset number of frequencies can be selected as target feature frequencies according to the amplitudes corresponding to each frequency in the frequency domain characteristics, in descending order of amplitude. The period corresponding to each target feature frequency is then taken as the target feature period. Accordingly, for any target feature period, the corresponding period weight is the ratio of the amplitude of the target feature frequency corresponding to the target feature period to the sum of the amplitudes of all target feature frequencies.
[0055] S1323. For any target feature period, reshape the input vector according to the target feature period to obtain the first intermediate feature tensor with a two-dimensional data structure.
[0056] Fourier transform can decompose the complex time variations contained in the input vector into variations within multiple periods and during the period, thereby enabling the extraction of feature information contained in the input vector based on periodicity.
[0057] S1324. For any first intermediate feature tensor, perform feature transformation on the first intermediate feature tensor using CNN to obtain a second intermediate feature tensor with a two-dimensional data structure.
[0058] In practical implementation, the CNN in TimesBlock can be implemented using parameter-efficient Inception blocks, ResNet, ConvNeXt, etc. By performing two-dimensional convolution on the first intermediate feature tensor using a CNN to obtain the second intermediate feature tensor, it is possible to capture the interaction information between features at different time scales.
[0059] S1325. For any second intermediate feature tensor, reshape the second intermediate feature tensor to obtain an intermediate temporal feature with a one-dimensional data structure.
[0060] By reshaping the second intermediate feature tensor, which has a two-dimensional data structure, and converting it back into an intermediate time-series feature with a one-dimensional data structure, TimesNet can capture the periodic patterns at different time scales in time series data.
[0061] S1326. The intermediate time-series features corresponding to each target feature period are fused according to the period weights corresponding to each target feature period to obtain the output features. Return to step S1321.
[0062] By using the periodic weights determined in step S1322 to perform a weighted summation of the intermediate time-series features corresponding to each target feature period, the feature information corresponding to each target feature period can be effectively fused.
[0063] S133. The global temporal features are compressed in temporal dimension by using a pooling layer to obtain the global compressed features.
[0064] Among them, the global time series features are the output features of the final-level TimesBlock.
[0065] In the specific implementation process, global average pooling, global max pooling, and other methods can be used to compress the time-series dimension of global time-series features to obtain globally compressed features.
[0066] S134. Dimensionally reduce the global compressed features by using a fully connected layer to obtain the global dimensionality-reduced features.
[0067] S135. The output result is determined by the output layer based on the global dimensionality reduction features.
[0068] In practical implementation, if the above vibration characteristic analysis model is a regression model, then the output layer can linearly map the global dimensionality reduction features to obtain the actual processing parameters of the target hole.
[0069] In practical implementation, if the vibration characteristic analysis model described above is a classification model, since the classification results include two categories: normal and abnormal processing parameters of the target hole, the global dimensionality reduction feature can be a two-dimensional classification probability vector. The two feature dimensions are used to indicate the probability feature value of normal actual processing parameters of the target hole and the probability feature value of abnormal actual processing parameters of the target hole, respectively. The output layer can use the Softmax or Sigmoid activation function to process the classification probability vector to obtain the final classification result.
[0070] Thus, the vibration feature analysis model designed based on the TimesNet model can capture frequency and time domain features from the target features for analysis, thereby obtaining accurate regression prediction or classification results. The vibration feature analysis models in the form of classification models trained for each type of workpiece were tested, and as shown in the table below, the vibration feature analysis model in the form of the TimesNet model has the best fit to the real situation compared to the classification model in the form of other machine learning models.
[0071] Table 1. Evaluation Indicators for Vibration Characteristic Analysis Models in Classification Model Forms
[0072] Furthermore, alternatively, such as Figure 1 As shown, after step S140, the method further includes: If the result of step S140 is yes, proceed to step S150.
[0073] S150, Control the machine tool to stop working.
[0074] In this way, by immediately stopping the machine tool when it is determined that the machining parameters of the target hole obtained by the machine tool boring the target workpiece are abnormal, the waste of production materials can be avoided due to the machine tool failure causing problems with the machining parameters of the target holes obtained by boring subsequent workpieces.
[0075] Based on the same inventive concept, embodiments of the present invention also provide a boring quality inspection device, such as... Figure 8 As shown, it includes: Data acquisition module M1 is used to acquire the acceleration signal generated by the target object in the target direction during the boring process of the machine tool on the target workpiece; Signal processing module M2 is used to condition the acceleration signal to obtain the target time domain signal; The analysis module M3 is used to input the target time-domain signal into a pre-trained vibration feature analysis model to obtain the output result determined by the vibration feature analysis model; and to determine whether the actual machining parameters of the target hole obtained by boring the target workpiece are abnormal based on the output result.
[0076] Optionally, the step of conditioning the acceleration signal to obtain the target time-domain signal includes: The acceleration signal is transformed into a first frequency domain signal; The first frequency domain signal is filtered in the frequency domain to obtain the second frequency domain signal; The second frequency domain signal is transformed into a first time domain signal; The first time-domain signal is nonlinearly amplified to obtain the target time-domain signal.
[0077] As an optional implementation, the output result is the actual machining parameters of the target hole obtained by boring the target workpiece; The step of determining whether the machining parameters of the target hole obtained by boring the target workpiece are abnormal based on the output result includes: Based on the numerical relationship between the actual processing parameters and the target processing parameters of the target hole, determine whether the actual processing parameters are abnormal.
[0078] As another optional implementation, the output result is a classification result indicating whether the actual machining parameters of the target hole obtained by boring the target workpiece are abnormal.
[0079] Optionally, the vibration feature analysis model sequentially includes an input layer, a TimesNet layer, a pooling layer, a fully connected layer, and an output layer; the TimesNet layer includes multiple sequentially connected time blocks (TimeBlocks). The step of inputting the target features into a pre-trained vibration feature analysis model to obtain the classification result output by the vibration feature analysis model includes: The target features are embedded through the input layer to obtain an embedding vector; The embedded vector is processed by the TimesNet layer to obtain global temporal features. Specifically, each TimesBlock performs the following steps sequentially according to the connection order: an FFT is applied to the one-dimensional input vector to obtain frequency domain features; the target feature period and corresponding period weights are determined based on the frequency domain features; for any target feature period, the input vector is reshaped according to the target feature period to obtain a two-dimensional first intermediate feature tensor; for any first intermediate feature tensor, a convolutional neural network (CNN) is used to transform the first intermediate feature tensor to obtain a two-dimensional second intermediate feature tensor; for any second intermediate feature tensor, the second intermediate feature tensor is reshaped to obtain a one-dimensional intermediate temporal feature; the intermediate temporal features corresponding to each target feature period are fused according to the corresponding period weights to obtain output features; wherein the input vector of the first-level TimesBlock is the embedded vector, the input vector of non-first-level TimesBlocks is the output feature of the previous-level TimesBlock, and the global temporal feature is the output feature of the last-level TimesBlock. The global temporal features are compressed in terms of temporal dimension through the pooling layer to obtain globally compressed features. The global compressed features are reduced in dimensionality using the fully connected layer to obtain a global dimensionality-reduced vector. The output layer determines the output result based on the global dimensionality reduction vector.
[0080] Optionally, the target object includes at least one of the following: The machine tool body, the target workpiece, the target tooling for fixing the target workpiece, the target tool for boring, and the target motion mechanism for moving the target tool or the target workpiece.
[0081] Optionally, the boring quality inspection device further includes: The shutdown module M4 is used to control the machine tool to stop working when the actual machining parameters of the target hole obtained by boring the target workpiece are abnormal, based on the output results.
[0082] Optionally, the boring quality inspection device further includes a learning module M5, used to: train the vibration feature analysis model based on machine learning samples, with the input as the target time-domain signal and the output as the training sample label.
[0083] Optionally, the learning module M5 is further configured to: construct machine learning samples based on target features; the machine learning samples are training samples, validation samples, or test samples; Once the model training conditions are met, the machine learning samples are used to perform machine learning on the vibration feature analysis model.
[0084] It should be understood that the device embodiments described above are merely illustrative. For example, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces, or indirect coupling or communication connection between devices or modules, and may be electrical, mechanical, or other forms.
[0085] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; that is, they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.
[0086] Furthermore, the functional modules in the various embodiments of the present invention can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module. The integrated module can be implemented in hardware or as a software functional module. If the integrated module is implemented as a software functional module and sold or used as an independent product, it can be stored in a computer-readable storage medium.
[0087] For example, since the vibration characteristic analysis model requires a large amount of computing power during the inference process, and the electronic equipment (including machine tools, industrial controllers, etc.) deployed on the production site may not be able to run the vibration characteristic analysis model, the boring quality inspection device can be set as a cloud-edge architecture control system. The industrial controller set on the production site acts as the edge side to implement the data acquisition module M1 and the shutdown module M4 of the boring quality inspection device. The acquired acceleration signal is preprocessed (e.g., the acceleration signal corresponding to the boring process is extracted) and then uploaded to the cloud. The vibration characteristic analysis model is deployed on the cloud to implement the signal processing module M2, analysis module M3, and learning module M5 of the boring quality inspection device.
[0088] Since the specific operation methods of each module of the boring quality inspection device have been described in detail in the corresponding boring quality inspection methods, they will not be repeated here.
[0089] Based on the same inventive concept, embodiments of this application also provide an electronic device, such as... Figure 9As shown, it includes: a processor 110 and a memory 120 for storing executable instructions of the processor 110; wherein the processor 110 is configured to execute the instructions to implement the boring quality inspection method.
[0090] In specific implementations, the device may vary significantly due to differences in configuration or performance. It may include one or more processors 110, memory 120, and computer-readable storage media 130. The memory 120 and / or computer-readable storage media 130 may contain one or more application programs 131 or data 132. The memory 120 and / or computer-readable storage media 130 may also contain one or more operating systems 133, such as Windows, Mac OS, Linux, iOS, Android, Unix, FreeBSD, etc. The memory 120 and computer-readable storage media 130 may be temporary or persistent storage. The application program 131 may include one or more of the aforementioned modules (…). Figure 9 (Not shown in the diagram), each module may include a series of instruction operations. Furthermore, the processor 110 may be configured to communicate with the computer-readable storage medium 130 and execute a series of instruction operations in the computer-readable storage medium 130 on the device. The device may also include one or more power supplies (…). Figure 9 (not shown in the image); one or more network interfaces 140, the network interface 140 including a wired network interface 141 and / or a wireless network interface 142; one or more input / output interfaces 143.
[0091] Based on the same inventive concept, embodiments of the present invention provide a computer-readable storage medium storing computer program code, which, when executed on a computer, enables the computer to implement the boring quality inspection method.
[0092] The computer-readable storage medium can be any available medium that a computer can store, or a data storage device such as a server or data center that integrates one or more available media. The available medium can be magnetic media (e.g., floppy disks, hard disks, USB flash drives, magnetic tapes, read-only memory (ROM), random access memory (RAM)), optical media (e.g., high-density digital video discs (DVDs), video compact discs (VCDs)), or semiconductor media (e.g., solid-state drives (SSDs)).
[0093] Since the principle of the computer-readable storage medium in solving the problem is the same as the boring quality inspection method described above, the implementation of the computer-readable storage medium can be found in the implementation of the method, and the repeated parts will not be described again.
[0094] Based on the same inventive concept, this application also provides a computer program product, which includes computer program code, which, when run on a computer, enables the computer to implement the boring quality inspection method.
[0095] The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions may be transmitted from a website site, computer, server, or data center to another website site, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, Digital Subscriber Line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means.
[0096] Since the principle of the above-mentioned computer program product in solving the problem is the same as the boring quality inspection method described above, the implementation of the above-mentioned computer program product can refer to the implementation of the method, and the repeated parts will not be repeated.
[0097] The boring quality inspection method and related hardware provided in this invention analyze the vibration signals of the machining equipment and / or the target workpiece during the boring process. This allows for the determination of whether the machining parameters of the target hole are abnormal without directly measuring them. This avoids measurement errors and subjective judgments by inspection personnel that can lead to inaccurate results, thus improving the accuracy of quality inspection. Furthermore, the method can be automated to provide classification results, reducing the workload of inspection personnel and enabling simultaneous production and inspection, thereby improving production efficiency.
[0098] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0099] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to this application. It should be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.
[0100] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.
[0101] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.
[0102] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.
Claims
1. A method for quality inspection of boring holes, characterized in that, include: Acquire the acceleration signal generated by the target object in the target direction during the boring process of the machine tool on the target workpiece; The acceleration signal is conditioned to obtain the target time-domain signal; The target time-domain signal is input into a pre-trained vibration feature analysis model to obtain the output result determined by the vibration feature analysis model; Based on the output results, determine whether the actual machining parameters of the target hole obtained by boring the target workpiece are abnormal.
2. The method as described in claim 1, characterized in that, The step of conditioning the acceleration signal to obtain the target time-domain signal includes: The acceleration signal is transformed into a first frequency domain signal; The first frequency domain signal is filtered in the frequency domain to obtain the second frequency domain signal; The second frequency domain signal is transformed into a first time domain signal; The first time-domain signal is nonlinearly amplified to obtain the target time-domain signal.
3. The method as described in claim 1, characterized in that, The output result is the actual machining parameters of the target hole obtained by boring the target workpiece. The step of determining whether the machining parameters of the target hole obtained by boring the target workpiece are abnormal based on the output result includes: Based on the numerical relationship between the actual processing parameters and the target processing parameters of the target hole, determine whether the actual processing parameters are abnormal.
4. The method as described in claim 1, characterized in that, The output result is a classification result indicating whether the actual machining parameters of the target hole obtained by boring the target workpiece are abnormal.
5. The method as described in claim 3 or 4, characterized in that, The vibration feature analysis model comprises, in sequence, an input layer, a TimesNet layer, a pooling layer, a fully connected layer, and an output layer; the TimesNet layer includes multiple TimeBlocks connected in sequence. The step of inputting the target features into a pre-trained vibration feature analysis model to obtain the classification result output by the vibration feature analysis model includes: The target features are embedded through the input layer to obtain an embedding vector; The embedded vector is processed by the TimesNet layer to obtain global temporal features. Specifically, each TimesBlock performs the following steps sequentially according to the connection order: an FFT is applied to the one-dimensional input vector to obtain frequency domain features; the target feature period and corresponding period weights are determined based on the frequency domain features; for any target feature period, the input vector is reshaped according to the target feature period to obtain a two-dimensional first intermediate feature tensor; for any first intermediate feature tensor, a convolutional neural network (CNN) is used to transform the first intermediate feature tensor to obtain a two-dimensional second intermediate feature tensor; for any second intermediate feature tensor, the second intermediate feature tensor is reshaped to obtain a one-dimensional intermediate temporal feature; the intermediate temporal features corresponding to each target feature period are fused according to the corresponding period weights to obtain output features; wherein the input vector of the first-level TimesBlock is the embedded vector, the input vector of non-first-level TimesBlocks is the output feature of the previous-level TimesBlock, and the global temporal feature is the output feature of the last-level TimesBlock. The global temporal features are compressed in terms of temporal dimension through the pooling layer to obtain globally compressed features. The global compressed features are reduced in dimensionality using the fully connected layer to obtain a global dimensionality-reduced vector. The output layer determines the output result based on the global dimensionality reduction vector.
6. The method as described in claim 1, characterized in that, The target object includes at least one of the following: The machine tool body, the target workpiece, the target tooling for fixing the target workpiece, the target tool for boring, and the target motion mechanism for moving the target tool or the target workpiece.
7. A boring hole quality inspection device, characterized in that, include: The data acquisition module is used to acquire the acceleration signal generated by the target object in the target direction during the boring process of the machine tool on the target workpiece; The signal processing module is used to perform signal conditioning on the acceleration signal to obtain the target time-domain signal; The analysis module is used to input the target time-domain signal into a pre-trained vibration feature analysis model to obtain the output result determined by the vibration feature analysis model; and to determine whether the actual machining parameters of the target hole obtained by boring the target workpiece are abnormal based on the output result.
8. An electronic device, characterized in that, include: A processor and a memory for storing processor-executable instructions; The processor is configured to execute the instructions to implement the boring quality inspection method as described in any one of claims 1-6.
9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer program code, which, when executed on a computer, causes the computer to perform the boring quality inspection method as described in any one of claims 1-6.
10. A computer program product, characterized in that, The computer program product includes: computer program code, which, when run on a computer, causes the computer to perform the boring quality inspection method as described in any one of claims 1-6.