Prediction method and device for residual life of cutting tool and storage medium
By matrix construction and autoencoder processing of the historical current and vibration data of the cutting tool, a weighted matrix is generated, and combined with the life prediction model, the problem of inaccurate life prediction of the cutting tool is solved, and more efficient life prediction is achieved.
Patent Information
- Application Number
- CN202411146218.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-20
- Publication Date
- 2024-12-20
AI Technical Summary
In the prior art, the remaining life prediction of cutting tools is inaccurate, resulting in large measurement errors, high cost, and long time, which affects production efficiency.
By obtaining the historical current data and vibration data of the cutting tool, building a current matrix and vibration matrix, inputting the autoencoder to reduce the dimensionality, generating a weighted matrix, and combining the life prediction model to output the remaining life prediction value to improve prediction accuracy.
Improve the accuracy and efficiency of the remaining life prediction of cutting tool, reduce test costs, and shorten prediction time.
Smart Images
Figure CN119167746A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of mechanical processing technology, and in particular to a method, device and storage medium for predicting the remaining life of a cutting tool. Background Art
[0002] In the field of industrial manufacturing, equipment life prediction management is an important and urgent issue. Equipment in different industries, such as automated machine tools and steel smelting blast furnaces, require strict control over the life of their components. The life of cutting tools used in automated production lines is not fixed, and exceeding the limit may lead to a decline in product processing quality or even damage to the machine tool spindle.
[0003] In the prior art, the remaining life of a cutting tool is generally predicted by the following methods: (1) Cumulative cutting time: estimating the remaining life of a cutting tool based on the total actual cutting time of the cutting tool; (2) Cumulative cutting length: estimating the remaining life of a tool based on the cumulative cutting length of the cutting tool; (3) Empirical method: estimating the life of a cutting tool based on on-site practice.
[0004] However, the cumulative cutting time and cumulative cutting length of the cutting tool need to be determined through field experiments. If the remaining life of the cutting tool is predicted by the above method, it will take a long time, the testing cost will be too high, and the measurement error will be large, resulting in low accuracy in the subsequent life prediction of the cutting tool. Summary of the invention
[0005] The purpose of the embodiments of the present application is to provide a method, device and storage medium for predicting the remaining life of a cutting tool, so as to solve the problem of inaccurate prediction of the remaining life of a cutting tool in the prior art.
[0006] In order to achieve the above-mentioned object, the present application provides, in a first aspect, a method for predicting the remaining life of a cutting tool, comprising:
[0007] Acquire historical current data and historical vibration data of the cutting tool to be predicted at multiple historical moments within a historical time period;
[0008] According to the historical current data and the historical vibration data at each historical moment, respectively determine the current matrix and the vibration matrix at each historical moment;
[0009] Inputting the current matrix and the vibration matrix into the trained first autoencoder and the second autoencoder respectively, so as to obtain the current matrix and the vibration matrix after dimensionality reduction respectively;
[0010] A weighting matrix is obtained based on each reduced-dimensional current matrix and each reduced-dimensional vibration matrix;
[0011] The weighting matrix is input into the trained life prediction model so that the trained life prediction model outputs a remaining life prediction value for the cutting tool.
[0012] In an embodiment of the present application, obtaining a weighted matrix based on each reduced-dimensional current matrix and each reduced-dimensional vibration matrix includes: determining a first weighting coefficient and a second weighting coefficient corresponding to each reduced-dimensional current matrix and each reduced-dimensional vibration matrix, wherein the second weighting coefficient is determined based on the first weighting coefficient; determining a first matrix between each reduced-dimensional current matrix and the corresponding first weighting coefficient, and determining a second matrix between each reduced-dimensional vibration matrix and the corresponding second weighting coefficient; and determining the sum of the first matrix and the second matrix as the weighted matrix.
[0013] In an embodiment of the present application, the prediction method also includes: after determining the remaining life prediction value for the cutting tool, obtaining a wear state comparison table corresponding to the type of cutting tool; determining the numerical range corresponding to the remaining life prediction value of the cutting tool in the wear state comparison table; and determining the preset wear state corresponding to the numerical range as the wear state of the cutting tool.
[0014] In an embodiment of the present application, determining the current matrix and vibration matrix at each historical moment based on the historical current data and the historical vibration data at each historical moment includes: preprocessing the historical current data and the historical vibration data at each historical moment to obtain preprocessed historical current data and preprocessed historical vibration data, respectively; determining the vibration matrix at each historical moment based on the component vibration data on multiple coordinate axes included in the preprocessed historical vibration data at each historical moment; and determining the current matrix at each historical moment based on the preprocessed historical current data at each historical moment.
[0015] In an embodiment of the present application, a preset coordinate system where the cutting tool is located includes an X-axis, a Y-axis, and a Z-axis, and component vibration data on multiple coordinate axes include first component vibration data on the X-axis, second component vibration data on the Y-axis, and third component vibration data on the Z-axis. Determining the vibration matrix at each historical moment based on the component vibration data on multiple coordinate axes included in the preprocessed historical vibration data at each historical moment includes: determining the first component vibration matrix, the second component vibration matrix, and the third component vibration matrix corresponding to the first component vibration data, the second component vibration data, and the third component vibration data, respectively; performing cross multiplication processing on the transposed matrices of the first component vibration matrix, the second component vibration matrix, and the second component vibration matrix to obtain an intermediate matrix; performing cross multiplication processing on the intermediate matrix, the third component vibration matrix, and the transposed matrix of the third component vibration matrix to obtain a vibration matrix.
[0016] In an embodiment of the present application, the prediction method also includes: obtaining historical current data and historical vibration data of another cutting tool of the same type as the cutting tool throughout its life cycle; determining multiple current sample matrices and multiple vibration sample matrices throughout its life cycle based on the historical current data and historical vibration data of the other cutting tool throughout its life cycle; inputting each current sample matrix into the first autoencoder so that the first autoencoder outputs the corresponding reduced-dimensional current sample matrix to train the first autoencoder; inputting each vibration sample matrix into the second autoencoder so that the second autoencoder outputs the corresponding reduced-dimensional vibration sample matrix to train the second autoencoder; iteratively adjusting the model parameters of the first autoencoder and the second autoencoder by the stochastic gradient descent method, and obtaining the trained first autoencoder and the second autoencoder respectively when the loss values of the loss functions of the first autoencoder and the second autoencoder converge.
[0017] In the embodiment of the present application, the loss function used by the first autoencoder and the second autoencoder during the training process is defined by formula (1):
[0018]
[0019] Where Loss refers to the loss value of the first autoencoder or the second autoencoder, n refers to the number of current sample matrices or vibration sample matrices, P i refers to the i-th current sample matrix or vibration sample matrix input to the first autoencoder or the second autoencoder, P′ i It refers to the reduced-dimensional sample matrix output by the first autoencoder or the second autoencoder corresponding to the i-th current sample matrix or vibration sample matrix.
[0020] In an embodiment of the present application, the prediction method also includes: inputting each current sample matrix and each vibration sample matrix into the trained first autoencoder and the second autoencoder respectively to obtain the corresponding reduced-dimensional current sample matrix and vibration sample matrix; performing weighted fusion processing on the reduced-dimensional current sample matrix and vibration sample matrix to obtain a weighted sample matrix; inputting the weighted sample matrix into a preset model so that the preset model outputs a remaining life simulation value for another cutting tool, until the preset model meets the preset training completion condition, and the trained preset model is determined as the life prediction model.
[0021] A second aspect of the present application provides a device for predicting the remaining life of a cutting tool, comprising:
[0022] a memory configured to store instructions;
[0023] The processor is configured to call instructions from the memory and implement the above-mentioned method for predicting the remaining life of the cutting tool when executing the instructions.
[0024] A third aspect of the present application provides a machine-readable storage medium having instructions stored thereon, which, when executed by a processor, configure the processor to execute the above-mentioned method for predicting the remaining life of a cutting tool.
[0025] Through the above technical scheme, the historical current data and historical vibration data of the cutting tool to be predicted at multiple historical moments within the historical time length are obtained; the current matrix and vibration matrix of each historical moment are determined respectively according to the historical current data and historical vibration data of each historical moment; the current matrix and the vibration matrix are respectively input into the trained first autoencoder and the second autoencoder to obtain the reduced-dimensional current matrix and the reduced-dimensional vibration matrix respectively; a weighted matrix is obtained based on each reduced-dimensional current matrix and each reduced-dimensional vibration matrix; the weighted matrix is input into the trained life prediction model, so that the trained life prediction model outputs the remaining life prediction value for the cutting tool, which can better combine the detailed information of the current data and the vibration data, and improve the accuracy of the remaining life prediction of the cutting tool.
[0026] Other features and advantages of the embodiments of the present application will be described in detail in the subsequent specific implementation section. BRIEF DESCRIPTION OF THE DRAWINGS
[0027] The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the following specific implementations, they are used to explain the embodiments of the present application, but do not constitute a limitation on the embodiments of the present application. In the accompanying drawings:
[0028] Figure 1 A schematic diagram of a process flow of a method for predicting the remaining life of a cutting tool according to an embodiment of the present application is schematically shown;
[0029] Figure 2 A schematic diagram of a process flow of a method for predicting the remaining life of a cutting tool according to another embodiment of the present application is schematically shown;
[0030] Figure 3 The internal structure diagram of a computer device according to an embodiment of the present application is schematically shown. DETAILED DESCRIPTION
[0031] In order to make the purpose, technical scheme and advantages of the embodiments of the present application clearer, the technical scheme in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. It should be understood that the specific implementation methods described herein are only used to illustrate and explain the embodiments of the present application, and are not used to limit the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.
[0032] It should be noted that if there are descriptions involving "first", "second", etc. in the embodiments of the present application, the descriptions of "first", "second", etc. are only used for descriptive purposes and cannot be understood as indicating or suggesting their relative importance or implicitly indicating the number of technical features indicated. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In addition, the technical solutions between the various embodiments can be combined with each other, but they must be based on the ability of ordinary technicians in the field to implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be deemed that such combination of technical solutions does not exist and is not within the scope of protection required by this application.
[0033] Figure 1 The flowchart of the method for predicting the remaining life of a cutting tool according to an embodiment of the present application is schematically shown. Figure 1 As shown, in one embodiment of the present application, a method for predicting the remaining life of a cutting tool is provided, comprising the following steps:
[0034] Step 101: Obtain historical current data and historical vibration data of a cutting tool to be predicted at multiple historical moments within a historical time period.
[0035] Cutting tools refer to tools used for cutting in mechanical manufacturing. Cutting tools may include cutting tools, specifically metal cutting tools. Cutting tools may be installed on machine tools, specifically on the spindle of the machine tool, on which vibration sensors and current sensors may be installed. The current sensor may be used to collect current data of the cutting tool. The vibration sensor may be used to collect vibration data of the cutting tool.
[0036] The processor may obtain historical current data of a cutting tool to be predicted at multiple historical moments within a historical duration. The historical duration may be set based on demand. The multiple historical moments within the historical duration may be obtained by dividing the historical duration according to a preset time interval. The preset time interval may be customized in advance.
[0037] Step 102: Determine the current matrix and the vibration matrix at each historical moment according to the historical current data and the historical vibration data at each historical moment.
[0038] The processor may determine the current matrix and the vibration matrix at each historical moment respectively according to the historical current data and the historical vibration data at each historical moment.
[0039] In an embodiment of the present application, determining the current matrix and vibration matrix at each historical moment based on the historical current data and the historical vibration data at each historical moment includes: preprocessing the historical current data and the historical vibration data at each historical moment to obtain preprocessed historical current data and preprocessed historical vibration data, respectively; determining the vibration matrix at each historical moment based on the component vibration data on multiple coordinate axes included in the preprocessed historical vibration data at each historical moment; and determining the current matrix at each historical moment based on the preprocessed historical current data at each historical moment.
[0040] The processor can preprocess the historical current data and historical vibration data at each historical moment to obtain the preprocessed historical current data and preprocessed historical vibration data. The preprocessing may include wavelet analysis, data elimination processing, and data filling processing. Wavelet analysis can better analyze the changing characteristics of tool wear in the collected data, so as to better obtain the trend and mutation of tool wear.
[0041] By eliminating the non-compliant data through data elimination processing, it is possible to prevent it from interfering with the prediction results of subsequent cutting tools. For example, historical current data and historical vibration data that are obviously too large or too small can be eliminated, and a data range can be set in advance for the historical current data and historical vibration data respectively. If the historical current data and historical vibration data at a certain historical moment fall into the corresponding data range, the historical current data and historical vibration data at that historical moment can be retained. If they do not fall into the corresponding data range, it can be determined that the historical current data and historical vibration data at that historical moment do not meet the conditions, that is, an abnormality may occur. At this time, the historical current data and historical vibration data at that historical moment can be eliminated.
[0042] The integrity of the data can be ensured through data filling processing. For example, for historical current data at different times, if the historical current data at multiple historical times is missing, and the proportion of the missing part to the total historical current data is within a certain range, the missing part of the historical current data can be filled based on the average value of the non-missing historical current data. If the proportion of the missing part to the total historical current data is not within a certain range, for example, the number of missing parts exceeds more than half of the total historical current data, at this time, in order to ensure the accuracy of the subsequent prediction of the remaining life of the cutting tool, the data of this part can be eliminated, and the current data of multiple historical times can be re-acquired.
[0043] The preprocessed historical vibration data at each historical moment includes component vibration data on multiple coordinate axes. The coordinate axes refer to the X-axis, Y-axis, and Z-axis included in the coordinate system where the cutting tool is located. The processor can determine the vibration matrix at each historical moment based on the component vibration data on multiple coordinate axes at each historical moment. The processor can determine the current matrix at each historical moment based on the preprocessed historical current data at each historical moment.
[0044] In an embodiment of the present application, a preset coordinate system where the cutting tool is located includes an X-axis, a Y-axis, and a Z-axis, and component vibration data on multiple coordinate axes include first component vibration data on the X-axis, second component vibration data on the Y-axis, and third component vibration data on the Z-axis. Determining the vibration matrix at each historical moment based on the component vibration data on multiple coordinate axes included in the preprocessed historical vibration data at each historical moment includes: determining the first component vibration matrix, the second component vibration matrix, and the third component vibration matrix corresponding to the first component vibration data, the second component vibration data, and the third component vibration data, respectively; performing cross multiplication processing on the transposed matrices of the first component vibration matrix, the second component vibration matrix, and the second component vibration matrix to obtain an intermediate matrix; performing cross multiplication processing on the intermediate matrix, the third component vibration matrix, and the transposed matrix of the third component vibration matrix to obtain a vibration matrix.
[0045] The preset coordinate system where the cutting tool is located includes an X-axis, a Y-axis, and a Z-axis, and the component vibration data on the multiple coordinate axes include first component vibration data on the X-axis, second component vibration data on the Y-axis, and third component vibration data on the Z-axis. For each historical moment, the processor can determine the first component vibration matrix, the second component vibration matrix, and the third component vibration matrix corresponding to the first component vibration data, the second component vibration data, and the third component vibration data at the historical moment, respectively.
[0046] The processor may perform a cross multiplication process on the first component vibration matrix, the second component vibration matrix, and the transposed matrix of the second component vibration matrix to obtain an intermediate matrix. The processor may perform a cross multiplication process on the intermediate matrix, the third component vibration matrix, and the transposed matrix of the third component vibration matrix to obtain a vibration matrix.
[0047] In one embodiment, the intermediate matrix at each historical moment is determined by the following formula:
[0048]
[0049] Among them, P (XY) refers to the intermediate matrix, P X refers to the first component vibration matrix, P Y is the second component vibration matrix, It refers to the transposed matrix of the second component vibration matrix.
[0050] In one embodiment, the vibration matrix at each historical moment is determined by the following formula:
[0051]
[0052] Among them, P D is the vibration matrix, P (XY) refers to the intermediate matrix, P Z refers to the third component vibration matrix, It refers to the transposed matrix of the third component vibration matrix.
[0053] The above scheme takes into account that feature fusion of the current matrix and the vibration matrix will lead to the neglect of some detailed information. Therefore, feature fusion is performed on the component vibration matrices on multiple coordinate axes included in the preprocessed historical vibration data of each historical moment to avoid the generation of too much invalid information. This can better utilize the information in the vibration data and improve the prediction accuracy of the remaining life of subsequent cutting tools.
[0054] Step 103: inputting the current matrix and the vibration matrix into the trained first autoencoder and the second autoencoder respectively, so as to obtain the current matrix and the vibration matrix after dimensionality reduction respectively.
[0055] After determining the current matrix and the vibration matrix at each historical moment, the first autoencoder and the second autoencoder can be trained respectively until the first autoencoder and the second autoencoder respectively meet the training completion conditions, thereby obtaining the trained first autoencoder and the trained second autoencoder.
[0056] In an embodiment of the present application, the prediction method also includes: obtaining historical current data and historical vibration data of another cutting tool of the same type as the cutting tool throughout its life cycle; determining multiple current sample matrices and multiple vibration sample matrices throughout its life cycle based on the historical current data and historical vibration data of the other cutting tool throughout its life cycle; inputting each current sample matrix into the first autoencoder so that the first autoencoder outputs the corresponding reduced-dimensional current sample matrix to train the first autoencoder; inputting each vibration sample matrix into the second autoencoder so that the second autoencoder outputs the corresponding reduced-dimensional vibration sample matrix to train the second autoencoder; iteratively adjusting the model parameters of the first autoencoder and the second autoencoder by the stochastic gradient descent method, and obtaining the trained first autoencoder and the second autoencoder respectively when the loss values of the loss functions of the first autoencoder and the second autoencoder converge.
[0057] The processor can obtain historical current data and historical vibration data of another cutting tool of the same type as the cutting tool during its entire life cycle. The entire life cycle refers to the period from the beginning of use of the other cutting tool to the time when it cannot be used. If the remaining life is less than a preset value, the other cutting tool may be severely worn, and it can be determined that the other cutting tool cannot be used.
[0058] The processor can determine multiple current sample matrices and multiple vibration sample matrices in the entire life cycle according to the historical current data and historical vibration data of another cutting tool in the entire life cycle. Specifically, the processor can preprocess the historical current data and historical vibration data at each moment in the entire life cycle to obtain preprocessed historical current data and preprocessed historical vibration data. The preprocessing can include wavelet analysis, data elimination processing, and data filling processing.
[0059] The processor can determine the vibration sample matrix at each moment in the life cycle according to the component vibration data on multiple coordinate axes included in the preprocessed historical vibration data at each moment in the life cycle. Specifically, the component vibration data at each moment in the life cycle can be subjected to feature fusion processing to obtain the vibration sample matrix at each moment in the life cycle. The processor can determine the current sample matrix at each moment in the life cycle according to the preprocessed historical current data at each moment in the life cycle.
[0060] The autoencoder consists of two parts: the encoding process and the decoding process. The encoding process will reduce the dimension of the input features. The decoding process is the inverse process of encoding, which reconstructs the reduced-dimensional features into the original features. The processor can input each current sample matrix into the first autoencoder so that the first autoencoder outputs the corresponding reduced-dimensional current sample matrix to train the first autoencoder. The processor can input each vibration sample matrix into the second autoencoder so that the second autoencoder outputs the corresponding reduced-dimensional vibration sample matrix to train the second autoencoder.
[0061] The processor can iteratively adjust the model parameters of the first autoencoder and the second autoencoder by the stochastic gradient descent method. If the loss value of the loss function of the first autoencoder and the second autoencoder converges, it can be said that the decoded features and the input features are less than the set error value. At this time, the first autoencoder and the second autoencoder meet the training completion conditions, and the first autoencoder and the second autoencoder that have completed the training are obtained respectively.
[0062] In the embodiment of the present application, the loss function used by the first autoencoder and the second autoencoder during the training process is defined by formula (1):
[0063]
[0064] Where Loss refers to the loss value of the first autoencoder or the second autoencoder, n refers to the number of current sample matrices or vibration sample matrices, P i refers to the i-th current sample matrix or vibration sample matrix input to the first autoencoder or the second autoencoder, P i ′ It refers to the reduced-dimensional sample matrix output by the first autoencoder or the second autoencoder corresponding to the i-th current sample matrix or vibration sample matrix.
[0065] After obtaining the trained first autoencoder and the trained second autoencoder, the processor can input the current matrix and the vibration matrix into the trained first autoencoder and the second autoencoder respectively to obtain the reduced-dimensional current matrix and the reduced-dimensional vibration matrix respectively.
[0066] Step 104: Obtain a weighting matrix based on each current matrix after dimensionality reduction and each vibration matrix after dimensionality reduction.
[0067] The processor may obtain a weighted matrix based on each reduced-dimensional current matrix and each reduced-dimensional vibration matrix. Specifically, the processor may perform weighted fusion on each reduced-dimensional current matrix and each reduced-dimensional vibration matrix to determine the matrix obtained after weighted fusion as the weighted matrix.
[0068] In an embodiment of the present application, obtaining a weighted matrix based on each reduced-dimensional current matrix and each reduced-dimensional vibration matrix includes: determining a first weighting coefficient and a second weighting coefficient corresponding to each reduced-dimensional current matrix and each reduced-dimensional vibration matrix, wherein the second weighting coefficient is determined based on the first weighting coefficient; determining a first matrix between each reduced-dimensional current matrix and the corresponding first weighting coefficient, and determining a second matrix between each reduced-dimensional vibration matrix and the corresponding second weighting coefficient; and determining the sum of the first matrix and the second matrix as the weighted matrix.
[0069] The processor may determine a first weighting coefficient and a second weighting coefficient corresponding to each reduced current matrix and each reduced vibration matrix. The first weighting coefficient may be any value within a preset data range, and the preset data range may be 0 to 1. Specifically, the first weighting coefficient may be set to 0.3 based on five-fold cross validation and ten-fold cross validation. The second weighting coefficient is determined based on the first weighting coefficient. Specifically, the second weighting coefficient is the difference between the preset value and the first weighting coefficient. The preset value may be set to 1.
[0070] The processor may determine a first matrix between each reduced-dimensional current matrix and the corresponding first weighting coefficient, and may determine a second matrix between each reduced-dimensional vibration matrix and the corresponding second weighting coefficient. The processor may determine the sum of the first matrix and the second matrix as a weighting matrix. In one embodiment, the weighting matrix may be determined based on the following formula:
[0071] W=γW I +(1-γ)W D
[0072] Where W is the weighting matrix, γ is the first weighting coefficient, and W I refers to the current matrix after dimension reduction, 1-γ refers to the second weighting coefficient, W D It refers to the vibration matrix after dimensionality reduction.
[0073] Step 105: Input the weighted matrix into the trained life prediction model, so that the trained life prediction model outputs a predicted value of the remaining life of the cutting tool.
[0074] After obtaining the weighted matrix of each historical moment, the preset model can be trained based on the weighted matrix of each historical moment until the training completion condition of the preset model is met to obtain a trained life prediction model. The preset model can be a GRU model, which can use the information of the previous time series in the input sequence to predict the value in the future time series.
[0075] In an embodiment of the present application, the prediction method also includes: inputting each current sample matrix and each vibration sample matrix into the trained first autoencoder and the second autoencoder respectively to obtain the corresponding reduced-dimensional current sample matrix and vibration sample matrix; performing weighted fusion processing on the reduced-dimensional current sample matrix and vibration sample matrix to obtain a weighted sample matrix; inputting the weighted sample matrix into a preset model so that the preset model outputs a remaining life simulation value for another cutting tool, until the preset model meets the preset training completion condition, and the trained preset model is determined as the life prediction model.
[0076] When training the life prediction model, the processor can input the current sample matrix and the vibration sample matrix at each moment into the trained first autoencoder and the second autoencoder, respectively, to obtain the corresponding current sample matrix and vibration sample matrix after dimensionality reduction. The processor can perform weighted fusion processing on the current sample matrix and the vibration sample matrix after dimensionality reduction to obtain a weighted sample matrix.
[0077] The processor can input the weighted sample matrix into the preset model so that the preset model outputs a remaining life simulation value for another cutting tool until the preset model meets the preset training completion condition, and the trained preset model is determined as the life prediction model.
[0078] The preset training completion condition may include that the number of training iterations of the preset model reaches a preset number, and / or the prediction accuracy of the preset model reaches a preset range, and / or the loss function of the preset model converges. The processor may input the weighted matrix into the trained life prediction model so that the trained life prediction model outputs a remaining life prediction value for the cutting tool.
[0079] In an embodiment of the present application, the prediction method also includes: after determining the remaining life prediction value for the cutting tool, obtaining a wear state comparison table corresponding to the type of cutting tool; determining the numerical range corresponding to the remaining life prediction value of the cutting tool in the wear state comparison table; and determining the preset wear state corresponding to the numerical range as the wear state of the cutting tool.
[0080] After determining the remaining life prediction value for the cutting tool, the processor may obtain a wear state comparison table corresponding to the type of the cutting tool. Each type of cutting tool may correspond to a wear state comparison table, and the wear state comparison table includes multiple value ranges of the corresponding type of cutting tool and a preset wear state corresponding to each value range. The value range can be used to define the remaining life value of the corresponding type of cutting tool when the corresponding type of cutting tool is in the corresponding preset wear state.
[0081] The processor may determine the numerical range corresponding to the remaining life prediction value of the cutting tool in the wear state comparison table. The processor may determine the preset wear state corresponding to the numerical range as the wear state of the cutting tool. Among them, the wear state may include the tool running-in state, the normal wear state and the rapid wear state. If the wear state of the cutting tool is the rapid wear state, a notification may be sent to the user terminal to prompt the user to replace the cutting tool in time to ensure the normal production of the subsequent products on the machine tool.
[0082] like Figure 2 As shown, a flow chart of another method for predicting the remaining life of a cutting tool is provided.
[0083] When predicting the remaining life of a cutting tool, a current signal and a vibration signal can be obtained. The current signal can be collected based on a current sensor, and the vibration signal can be collected based on a vibration sensor. Afterwards, feature extraction can be performed on the acquired current signal and vibration signal based on wavelet analysis, and denoising can be performed on the current signal and vibration signal after feature extraction to obtain the processed current signal and the processed vibration signal including the X-axis signal, the Y-axis signal, and the Z-axis signal.
[0084] The hybrid feature fusion method is used to fuse the features of the X-axis signal, the Y-axis signal, and the Z-axis signal to obtain the fused features. Among them, each axis signal will be updated according to the other axis signals to better integrate the information of other axis signals. After obtaining the fused features, the matrices corresponding to the fused features and the current signals can be used as the inputs of the corresponding autoencoders respectively, and the model parameters of the corresponding autoencoders are iteratively adjusted by the stochastic gradient descent method to minimize the loss function of the corresponding autoencoder and obtain the trained autoencoder. After the autoencoder training is completed, the features after dimensionality reduction can be decoded to be as close to the original features as possible. At this time, the reduced dimensionality current signal matrix and the reduced dimensionality vibration signal matrix can be obtained respectively.
[0085] The weighted fusion method can be used to perform weighted fusion on the reduced-dimensional current signal matrix and the reduced-dimensional vibration signal matrix to obtain a weighted fused matrix, which can better combine the detailed features of the current signal and the vibration signal. Afterwards, the weighted fused matrix can be used as the input of the GRU model to obtain the predicted life of the cutting tool. Based on the predicted life of the cutting tool, the wear state of the cutting tool can be determined. Among them, the wear state includes the tool running-in state, normal wear state and rapid wear state.
[0086] Through the above technical scheme, the historical current data and historical vibration data of the cutting tool to be predicted at multiple historical moments within the historical time length are obtained; the current matrix and vibration matrix of each historical moment are determined respectively according to the historical current data and historical vibration data of each historical moment; the current matrix and the vibration matrix are respectively input into the trained first autoencoder and the second autoencoder to obtain the reduced-dimensional current matrix and the reduced-dimensional vibration matrix respectively; a weighted matrix is obtained based on each reduced-dimensional current matrix and each reduced-dimensional vibration matrix; the weighted matrix is input into the trained life prediction model, so that the trained life prediction model outputs the remaining life prediction value for the cutting tool, which can better combine the detailed information of the current data and the vibration data, and improve the accuracy of the remaining life prediction of the cutting tool.
[0087] Figure 1 and 2 FIG. 1 is a flow chart of a method for predicting the remaining life of a cutting tool in one embodiment. Figure 1 and 2 The steps in the flowchart are shown in sequence as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified in this document, there is no strict order restriction for the execution of these steps, and these steps can be executed in other orders. Moreover, Figure 1 and 2At least part of the steps may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least part of the sub-steps or stages of other steps.
[0088] In one embodiment, a device for predicting the remaining life of a cutting tool is provided, comprising:
[0089] a memory configured to store instructions; and
[0090] The processor is configured to call instructions from the memory and implement the above-mentioned method for predicting the remaining life of the cutting tool when executing the instructions.
[0091] In one embodiment, a storage medium is provided, on which a program is stored, and when the program is executed by a processor, the above-mentioned method for predicting the remaining life of a cutting tool is implemented.
[0092] In one embodiment, a processor is provided, and the processor is used to run a program, wherein the program executes the above-mentioned method for predicting the remaining life of a cutting tool when it is run.
[0093] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Figure 3 As shown. The computer device includes a processor A01, a network interface A02, a memory (not shown in the figure) and a database (not shown in the figure) connected via a system bus. Among them, the processor A01 of the computer device is used to provide computing and control capabilities. The memory of the computer device includes an internal memory A03 and a non-volatile storage medium A04. The non-volatile storage medium A04 stores an operating system B01, a computer program B02 and a database (not shown in the figure). The internal memory A03 provides an environment for the operation of the operating system B01 and the computer program B02 in the non-volatile storage medium A04. The database of the computer device is used to store data such as the predicted value of the remaining life of the cutting tool. The network interface A02 of the computer device is used to communicate with an external terminal through a network connection. When the computer program B02 is executed by the processor A01, a method for predicting the remaining life of a cutting tool is implemented.
[0094] Those skilled in the art will understand that Figure 3 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.
[0095] An embodiment of the present application provides a device, which includes a processor, a memory, and a program stored in the memory and executable on the processor. When the processor executes the program, the following steps are implemented: obtaining historical current data and historical vibration data of a cutting tool to be predicted at multiple historical moments within a historical period; determining a current matrix and a vibration matrix at each historical moment according to the historical current data and the historical vibration data at each historical moment; inputting the current matrix and the vibration matrix into a trained first autoencoder and a trained second autoencoder, respectively, to obtain a reduced-dimensional current matrix and a reduced-dimensional vibration matrix, respectively; obtaining a weighted matrix based on each reduced-dimensional current matrix and each reduced-dimensional vibration matrix; and inputting the weighted matrix into a trained life prediction model, so that the trained life prediction model outputs a remaining life prediction value for the cutting tool.
[0096] In one embodiment, obtaining a weighted matrix based on each reduced-dimensional current matrix and each reduced-dimensional vibration matrix includes: determining a first weighting coefficient and a second weighting coefficient corresponding to each reduced-dimensional current matrix and each reduced-dimensional vibration matrix, wherein the second weighting coefficient is determined based on the first weighting coefficient; determining a first matrix between each reduced-dimensional current matrix and the corresponding first weighting coefficient, and determining a second matrix between each reduced-dimensional vibration matrix and the corresponding second weighting coefficient; and determining the sum of the first matrix and the second matrix as the weighted matrix.
[0097] In one embodiment, the prediction method also includes: after determining the remaining life prediction value for the cutting tool, obtaining a wear state comparison table corresponding to the type of the cutting tool; determining the numerical range corresponding to the remaining life prediction value of the cutting tool in the wear state comparison table; and determining the preset wear state corresponding to the numerical range as the wear state of the cutting tool.
[0098] In one embodiment, determining the current matrix and vibration matrix at each historical moment based on the historical current data and historical vibration data at each historical moment includes: preprocessing the historical current data and historical vibration data at each historical moment to obtain preprocessed historical current data and preprocessed historical vibration data respectively; determining the vibration matrix at each historical moment based on component vibration data on multiple coordinate axes included in the preprocessed historical vibration data at each historical moment; and determining the current matrix at each historical moment based on the preprocessed historical current data at each historical moment.
[0099] In one embodiment, the preset coordinate system where the cutting tool is located includes an X-axis, a Y-axis and a Z-axis, and the component vibration data on the multiple coordinate axes include first component vibration data on the X-axis, second component vibration data on the Y-axis and third component vibration data on the Z-axis. Determining the vibration matrix at each historical moment based on the component vibration data on the multiple coordinate axes included in the preprocessed historical vibration data at each historical moment includes: determining the first component vibration matrix, the second component vibration matrix and the third component vibration matrix corresponding to the first component vibration data, the second component vibration data and the third component vibration data respectively; cross-producting the first component vibration matrix, the second component vibration matrix and the transposed matrix of the second component vibration matrix to obtain an intermediate matrix; cross-producting the intermediate matrix, the third component vibration matrix and the transposed matrix of the third component vibration matrix to obtain a vibration matrix.
[0100] In one embodiment, the prediction method also includes: obtaining historical current data and historical vibration data of another cutting tool of the same type as the cutting tool throughout its life cycle; determining multiple current sample matrices and multiple vibration sample matrices throughout its life cycle based on the historical current data and historical vibration data of the other cutting tool throughout its life cycle; inputting each current sample matrix into the first autoencoder so that the first autoencoder outputs the corresponding reduced-dimensional current sample matrix to train the first autoencoder; inputting each vibration sample matrix into the second autoencoder so that the second autoencoder outputs the corresponding reduced-dimensional vibration sample matrix to train the second autoencoder; iteratively adjusting the model parameters of the first autoencoder and the second autoencoder by the stochastic gradient descent method, and obtaining the trained first autoencoder and the second autoencoder respectively when the loss values of the loss functions of the first autoencoder and the second autoencoder converge.
[0101] In one embodiment, the loss function used by the first autoencoder and the second autoencoder during the training process is defined by formula (1):
[0102]
[0103] Where Loss refers to the loss value of the first autoencoder or the second autoencoder, n refers to the number of current sample matrices or vibration sample matrices, P i refers to the i-th current sample matrix or vibration sample matrix input to the first autoencoder or the second autoencoder, P i ′ It refers to the reduced-dimensional sample matrix output by the first autoencoder or the second autoencoder corresponding to the i-th current sample matrix or vibration sample matrix.
[0104] In one embodiment, the prediction method also includes: inputting each current sample matrix and each vibration sample matrix into the trained first autoencoder and second autoencoder respectively to obtain the corresponding reduced-dimensional current sample matrix and vibration sample matrix; performing weighted fusion processing on the reduced-dimensional current sample matrix and vibration sample matrix to obtain a weighted sample matrix; inputting the weighted sample matrix into a preset model so that the preset model outputs a remaining life simulation value for another cutting tool, until the preset model meets the preset training completion condition, and the trained preset model is determined as the life prediction model.
[0105] The present application also provides a computer program product which, when executed on a data processing device, is suitable for executing a program for initiating the steps of a method for predicting the remaining life of a cutting tool.
[0106] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.
[0107] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.
[0108] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.
[0109] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.
[0110] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.
[0111] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.
[0112] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.
[0113] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.
[0114] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included within the scope of the claims of the present application.
Claims
1. A method for predicting the remaining life of a cutting tool, characterized in that: The prediction method comprises: Acquire historical current data and historical vibration data of the cutting tool to be predicted at multiple historical moments within a historical time period; According to the historical current data and the historical vibration data at each historical moment, respectively determine the current matrix and the vibration matrix at each historical moment; Inputting the current matrix and the vibration matrix into the trained first autoencoder and the second autoencoder respectively, so as to obtain the current matrix and the vibration matrix after dimensionality reduction respectively; A weighting matrix is obtained based on each reduced-dimensional current matrix and each reduced-dimensional vibration matrix; The weighting matrix is input into a trained life prediction model so that the trained life prediction model outputs a remaining life prediction value for the cutting tool.
2. The method for predicting the remaining life of a cutting tool according to claim 1, characterized in that: The step of obtaining a weighted matrix based on each reduced-dimensional current matrix and each reduced-dimensional vibration matrix comprises: Determine a first weighting coefficient and a second weighting coefficient corresponding to each reduced-dimensional current matrix and each reduced-dimensional vibration matrix, wherein the second weighting coefficient is determined according to the first weighting coefficient; Determine a first matrix between each reduced-dimensional current matrix and the corresponding first weighting coefficient, and determine a second matrix between each reduced-dimensional vibration matrix and the corresponding second weighting coefficient; A sum between the first matrix and the second matrix is determined as the weighting matrix.
3. The method for predicting the remaining life of a cutting tool according to claim 1, characterized in that: The prediction method further comprises: After determining the remaining life prediction value for the cutting tool, obtaining a wear state comparison table corresponding to the type of the cutting tool; Determine the numerical range corresponding to the remaining life prediction value of the cutting tool in the wear state comparison table; The preset wear state corresponding to the numerical range is determined as the wear state of the cutting tool.
4. The method for predicting the remaining life of a cutting tool according to claim 1, characterized in that: The step of respectively determining the current matrix and the vibration matrix at each historical moment according to the historical current data and the historical vibration data at each historical moment comprises: Preprocessing the historical current data and the historical vibration data at each historical moment respectively to obtain preprocessed historical current data and preprocessed historical vibration data respectively; Determine a vibration matrix at each historical moment according to component vibration data on multiple coordinate axes included in the preprocessed historical vibration data at each historical moment; The current matrix at each historical moment is determined according to the preprocessed historical current data at each historical moment.
5. The method for predicting the remaining life of a cutting tool according to claim 4, characterized in that: The preset coordinate system where the cutting tool is located includes an X-axis, a Y-axis and a Z-axis, the component vibration data on the multiple coordinate axes include first component vibration data on the X-axis, second component vibration data on the Y-axis and third component vibration data on the Z-axis, and the vibration matrix at each historical moment is determined based on the component vibration data on the multiple coordinate axes included in the preprocessed historical vibration data at each historical moment, including: Determine a first component vibration matrix, a second component vibration matrix, and a third component vibration matrix corresponding to the first component vibration data, the second component vibration data, and the third component vibration data, respectively; Perform a cross multiplication process on the first component vibration matrix, the second component vibration matrix, and the transposed matrix of the second component vibration matrix to obtain an intermediate matrix; The intermediate matrix, the third component vibration matrix and the transposed matrix of the third component vibration matrix are cross-multiplied to obtain the vibration matrix.
6. The method for predicting the remaining life of a cutting tool according to claim 1, characterized in that: The prediction method further comprises: Acquire historical current data and historical vibration data of another cutting tool of the same type as the cutting tool over its entire life cycle; Determine a plurality of current sample matrices and a plurality of vibration sample matrices in the entire life cycle respectively according to the historical current data and the historical vibration data of the other cutting tool in the entire life cycle; Inputting each current sample matrix into a first autoencoder so that the first autoencoder outputs a corresponding current sample matrix after dimensionality reduction, so as to train the first autoencoder; Inputting each vibration sample matrix into a second autoencoder so that the second autoencoder outputs a corresponding vibration sample matrix after dimension reduction, so as to train the second autoencoder; The model parameters of the first autoencoder and the second autoencoder are iteratively adjusted by the stochastic gradient descent method, and when the loss values of the loss functions of the first autoencoder and the second autoencoder converge, the trained first autoencoder and the second autoencoder are respectively obtained.
7. The method for predicting the remaining life of a cutting tool according to claim 6, characterized in that: The loss function used by the first autoencoder and the second autoencoder during the training process is defined by formula (1): Wherein, Loss refers to the loss value of the first autoencoder or the second autoencoder, n refers to the number of current sample matrices or vibration sample matrices, P i refers to the i-th current sample matrix or vibration sample matrix input to the first autoencoder or the second autoencoder, P i ′ It refers to the reduced-dimensional sample matrix output by the first autoencoder or the second autoencoder corresponding to the i-th current sample matrix or vibration sample matrix.
8. The method for predicting the remaining life of a cutting tool according to claim 6, characterized in that: The prediction method further comprises: Input each current sample matrix and each vibration sample matrix into the trained first autoencoder and the second autoencoder respectively to obtain the corresponding current sample matrix and vibration sample matrix after dimensionality reduction; Performing weighted fusion processing on the current sample matrix and the vibration sample matrix after dimension reduction to obtain a weighted sample matrix; The weighted sample matrix is input into a preset model so that the preset model outputs a remaining life simulation value for the other cutting tool until the preset model meets a preset training completion condition, and the trained preset model is determined as the life prediction model.
9. A device for predicting the remaining life of a cutting tool, characterized in that: The device comprises: a memory configured to store instructions; A processor is configured to call the instructions from the memory and implement the method for predicting the remaining life of a cutting tool according to any one of claims 1 to 8 when executing the instructions.
10. A machine-readable storage medium, characterized in that: The machine-readable storage medium stores instructions for causing a machine to execute the method for predicting the remaining life of a cutting tool according to any one of claims 1 to 8.