Tool wear condition monitoring method based on graph Fourier transform and key order

Abstract

A tool wear condition monitoring method based on graph Fourier transform and key orders comprises the following steps: step 1: collecting monitoring signals of the tool at different wear stages, and selecting signals of a given length as original signal samples; step 2: converting the one-dimensional signal in the original signal sample into a graph signal according to graph spectrum theory, calculating the graph Laplacian matrix of the graph signal, and extracting eigenvalues and eigenvectors; step 3: converting the original signal sample into a graph Fourier transform (GFT) value related to the order of the eigenvalue; step 4: arranging the GFT values obtained from the monitoring signals at different wear stages in ascending order, taking the eigenvalue orders corresponding to the first k GFT values as key orders, and obtaining h (h≥k) key orders; step 5: obtaining all order combination patterns based on the h key orders and a dimension parameter m, and obtaining a typical pattern library; step 6: matching the order combination pattern of the monitoring signal processed by steps 1 to 5 with the typical pattern library to identify the tool wear condition.

Description

Technical Field

[0001] The present invention relates to a tool wear state monitoring method, and more specifically to a method for monitoring the tool wear state by utilizing tool cutting process monitoring data in combination with graph Fourier transform and key order technology. Background Art

[0002] Cutting tools are crucial components of CNC machine tools. Tool wear directly impacts the surface integrity, dimensional accuracy, and machining quality of finished products, further impacting system production efficiency and costs. In severe cases, tool wear can cause severe vibration, leading to forced downtime and potentially impacting personnel safety. Therefore, real-time online tool monitoring is crucial to ensuring proper machine operation and product quality.

[0003] The tool wear process can be divided into the initial wear phase, the steady wear phase, and the rapid wear phase. The wear rates and wear characteristics vary significantly across these phases, resulting in significant differences in typical tool wear curves under different wear conditions. Traditional tool changing strategies based on experience or scheduled changes are prone to premature or late tool changes. Premature tool changes fail to fully utilize the tool's value, resulting in increased costs; late tool changes can severely impact product quality and increase the rate of defective products.

[0004] By extracting relevant fault features from monitoring signals and establishing an association model with the wear status, online monitoring of tool wear status can be achieved. Most existing methods use traditional signal processing technology, in which characteristic parameters in the time domain, frequency domain, and time-frequency domain of the signal are extracted to establish a model. However, most of these features are based on the assumption that the signal is stable and linear, and they themselves have various defects. In addition, since the wear value detection of the tool generally requires shutdown, it is impossible to obtain the wear value in real time and judge the wear stage of the tool while the tool is being used for processing. Therefore, how to use key features to adaptively identify the various wear stages of the tool during the cutting process, realize online wear status monitoring of the tool and reasonably change the tool is the first priority to ensure product quality. Summary of the Invention

[0005] In order to solve the above problems in this field, an embodiment of the present invention provides a tool wear state monitoring method based on graph Fourier transform and key order, which introduces graph spectrum theory into the field of tool wear state monitoring.

[0006] According to an embodiment of the present invention, a tool wear state monitoring method based on graph Fourier transform and key order is provided, which is characterized by comprising the following steps: Step 1: Monitoring signal acquisition: collecting monitoring signals of the tool at different wear stages during the cutting process, and selecting signals of a given length as original signal samples; Step 2: Graph signal conversion: according to graph spectrum theory, converting the one-dimensional signal in the original signal sample into a graph signal, calculating the graph Laplace matrix of the graph signal, and extracting the eigenvalues and eigenvectors of the graph Laplace matrix; Step 3: Graph Fourier transform (GFT): according to the graph Fourier transform and the eigenvectors of the Laplace matrix obtained in step 2, converting the original signal sample into a series of GFT values related to the order of the eigenvalues; Step 4: Key order Number extraction: Arrange the GFT values obtained from the monitoring signals of different wear stages in ascending order, take the eigenvalue orders corresponding to the first k GFT values as the key orders, and integrate the key orders of each wear stage to obtain h (h ≥ k) key orders in total; Step five: Typical order combination pattern division: According to the h key orders and the dimension parameter m, all order combination patterns are obtained, and a typical pattern library with a good mapping relationship with different wear stages is obtained; Step six: Tool wear state identification: The order combination pattern obtained by processing the collected monitoring signals through steps one to five is matched with the typical pattern library obtained by processing the training data through steps one to five to realize tool wear state identification. When the wear state is in the acute wear stage, it is indicated to replace the tool in time.

[0007] In an optional embodiment, the different wear stages described in step one refer to the different stages that the tool goes through during continuous use according to its wear characteristics, including three stages: initial wear stage, steady wear stage and rapid wear stage.

[0008] In an optional embodiment, for an undirected and connected weighted graph containing N vertices, it is represented as G = (V, E, A), where G is the graph signal of the weighted graph, V = (v1, v2, ..., v N ) represents the set of vertices in the graph, |V|=N means there are N vertices in total; E=(e1,e2,…,e M ) represents the set of edges in the graph, |E|=M means there are M edges in total. When two vertices v i With v j Adjacent, that is, vertex v i With v j When there is an edge connection between them, it is recorded as e ij , i=1,2,…,N and j=1,2,…,N; A is the weighted adjacency matrix, A=(a ij ) N×N , when two vertices v i With v j When adjacent, aij Represents edge e ij The weighted ij =1; when two vertices v i With v j When not adjacent, a ij = 0, the matrix A is a matrix whose sub-diagonal elements are all 1 and other elements are all 0; define the degree diagonal matrix D = diag(d1, d2, ..., d N ) is a diagonal matrix, the diagonal elements d i Let it be vertex v i The degree of matrix D is a diagonal matrix with the first and last elements of the main diagonal being 1 and the remaining diagonal elements being 2.

[0009]

[0010] The graph Laplacian matrix is defined as L = DA. The graph Laplacian matrix L is based on the weighted adjacency matrix A and the degree diagonal matrix D. It contains both the degree information of the vertices and the association information between the vertices. The graph Laplacian matrix L is a real symmetric matrix. After performing a complete orthogonal decomposition, a set of non-negative real eigenvalues σ(L) = {λ1,λ2,…,λ N} and the corresponding eigenvector Y={y1,y2,…,y N}, that is, the graph Laplace matrix L satisfies:

[0011] Ly i =λ i y i

[0012] Where i = 1, 2, ..., N represents the sequence number of the vertex in the graph, λ i is the eigenvalue, and the eigenvalue σ(L) satisfies 0=λ1≤λ 2≤ …≤λ N =λ max ,y i represents the eigenvalue λ i The corresponding eigenvector.

[0013] In an optional embodiment, the monitoring signal is a road map, wherein each sampling point in the monitoring signal corresponds to a vertex in the road map, the value of the sampling point corresponds to the vertex value of the road map, the sequential connection relationship of the sampling points obtained according to the sampling frequency in the monitoring signal corresponds to the edge connection relationship in the road map, and according to this conversion relationship, the one-dimensional monitoring signal is converted into a graph signal G = (V, E, A) in the form of a road map.

[0014] In an optional embodiment, the graph Fourier transform in step 3 includes decomposing the graph signal G according to the eigenvector Y of the graph Laplacian matrix L, specifically including:

[0015] The GFT of signal X is calculated as follows:

[0016]

[0017] Where, GFT l represents the eigenvalue λ l The corresponding GFT value, l = 1, 2, ..., N represents the eigenvalue order of the GFT of signal X, x n represents the amplitude of the nth sampling point in the signal X, represents the eigenvalue λ l The corresponding eigenvector The nth element value in .

[0018] In an optional embodiment, the critical order described in step 4 mainly refers to the eigenvalue order closely associated with each wear stage of the tool, reflecting the wear stage of the tool. For each type of wear stage, the N GFT values of the original signal are calculated, and the eigenvalue order corresponding to the first k GFT values is taken as the critical order, and a total of h (h ≥ k) critical orders are obtained.

[0019] In an optional embodiment, the order combination pattern described in step 5 is an order combination pattern determined according to h key orders and a dimension parameter m. The order combination pattern is calculated as follows:

[0020] Determine the dimension parameter m (m≤h), and randomly select m from the h key orders to obtain all possible combinations. There are C combinations in total, which are calculated as follows:

[0021]

[0022] Each order combination mode is recorded as c represents the order number of the combination mode, c = 1, 2, ..., C. The order of the order in the combination is arranged in descending order of the GFT amplitude corresponding to the order, indicating the importance of the order.

[0023] In an optional embodiment, the typical pattern library described in step five is to select typical order combination patterns that have a mapping relationship with the three tool wear stages respectively, and construct a correlation model between the arrangement pattern and the tool wear state.

[0024] In an optional embodiment, the pattern matching with the typical pattern library described in step six refers to obtaining the order combination pattern of the signal sample to be tested according to steps one to five, and matching the order combination pattern of the signal sample to be tested with the typical order combination pattern library. If the order combination pattern of the signal sample to be tested exists in the typical pattern of a certain wear stage, the signal sample to be tested is identified as the wear stage.

[0025] Compared with traditional signal processing techniques, the graph signal processing used in the embodiments of the present invention focuses more on the structural relationship between signals rather than the signal amplitude. From the perspective of graph spectrum theory, the Graph Fourier Transform (GFT) converts a one-dimensional signal into a graph (i.e., a graph signal) and uses the Laplace matrix to extract the key orders in the signal to achieve signal decomposition and obtain graph-based signal features. The key order is a method that uses the Laplace matrix eigenvalues of the graph signal and GFT to extract the key order combination pattern in the signal of different tool wear stages, which is used for tool wear status identification.

[0026] Therefore, based on the monitoring data of the tool during the cutting process, the GFT method is used to convert the monitoring signals at different times into a Laplace matrix based on the graph signal. On this basis, key orders are extracted, and key order combination patterns for different wear stages are obtained, which is of great significance for achieving adaptive identification of each wear stage of the tool. Based on this, an embodiment of the present invention provides a tool wear state monitoring method based on graph Fourier transform and key orders. According to the tool wear state monitoring method based on graph Fourier transform and key orders provided by the embodiment of the present invention, the original monitoring signals under different cutting paths are converted into graph signals through graph spectrum theory. On this basis, the GFT method is used to obtain the weighted adjacency matrix and Laplace matrix of the corresponding signal. The eigenvalues and eigenvectors of the Laplace matrix are obtained through orthogonal decomposition. The ordinal number of the matrix eigenvalues after ascending order is called the GFT order. The key orders of the monitoring signals of the tool at different wear stages are obtained, and a library of typical key order combination patterns for each wear stage is obtained. Based on the obtained typical pattern library, adaptive identification of the tool wear state can be achieved, providing a reference for tool replacement during the cutting process.

[0027] Through the method provided by the embodiment of the present invention, by utilizing the monitoring signals of some cutting tools at various wear stages during the cutting process, and combining with the relevant knowledge of graph theory, a tool wear state monitoring model can be established. In this model, the road map in graph theory is introduced into the tool monitoring signal, and the structural information of the monitoring signal is studied using the graph Fourier transform method. The model according to the embodiment of the present invention has at least the advantages of simple calculation and strong noise resistance, and can timely reflect the changes in the tool wear stage, realize online wear state monitoring of the tool cutting process, provide reasonable guidance solutions for measures such as tool changing during the processing process, make full use of the tool value and ensure product qualification rate. The method provided by the embodiment of the present invention has a certain openness in practical application, and is suitable for real-time monitoring of tool wear state and wear stage identification during cutting processing. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] The foregoing features of the present invention may be more readily understood with reference to the accompanying drawings and the following detailed description, in which:

[0029] Figure 1 A flow chart of a tool wear condition monitoring method based on graph Fourier transform and key order according to one embodiment of the present invention is shown;

[0030] Figure 2 A curve diagram showing a monitoring change of the wear value of a tool throughout its life cycle in an example of a tool wear condition monitoring method based on graph Fourier transform and key order according to an embodiment of the present invention is shown;

[0031] Figure 3 A diagram showing cutting force signals of a tool at different wear stages during a cutting process in an example of a tool wear state monitoring method based on graph Fourier transform and key order according to an embodiment of the present invention is shown;

[0032] Figure 4 A GFT diagram of a cutting force signal of a tool at different wear stages during a cutting process is shown in an example of a tool wear state monitoring method based on graph Fourier transform and key order according to an embodiment of the present invention;

[0033] Figure 5 The figure shows a typical pattern library (key order combination) of different wear stages of a tool during a cutting process in an example of a tool wear condition monitoring method based on graph Fourier transform and key order according to an embodiment of the present invention;

[0034] Figure 6 The tool wear state monitoring result in an example of applying the tool wear state monitoring method based on graph Fourier transform and key order according to an embodiment of the present invention is shown.

[0035] The serial numbers, symbols and codes in the figure are explained as follows:

[0036] L: Laplace matrix obtained by Fourier transform of the graph;

[0037] VB: tool flank wear, unit: mm;

[0038] Z: indicates the Zth pass of the tool;

[0039] GFT: stands for Graph Fourier Transform. DETAILED DESCRIPTION

[0040] The exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided in order to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art. It should be noted that, unless otherwise specified, the technical terms or scientific terms used in the present invention should have the common meanings understood by those skilled in the art to which the present invention belongs. In conjunction with the accompanying drawings, in one embodiment of the present invention, a specific implementation method is described in detail, but the present invention is not limited by the specific implementation method.

[0041] A tool wear condition monitoring method based on Fourier transform and key order provided by one embodiment of the present invention will be described below with reference to the accompanying drawings. Figure 1 A flow chart of a tool wear condition monitoring method based on graph Fourier transform and key order according to an embodiment of the present invention is shown.

[0042] refer to Figure 1 According to an embodiment of the present invention, a tool wear condition monitoring method based on graph Fourier transform and key order is proposed, comprising the following steps:

[0043] Step 1: Monitoring signal acquisition: Collect monitoring signals of the tool at different wear stages during the cutting process, and select signals of a given length as the original signal samples;

[0044] Step 2: Graph signal conversion: According to graph spectrum theory, the one-dimensional signal in the original signal sample is converted into a graph signal, the graph Laplacian matrix of the graph signal is calculated, and the eigenvalues and eigenvectors of the graph Laplacian matrix are extracted;

[0045] Step 3: Graph Fourier Transform (GFT): Based on the Graph Fourier Transform and the eigenvectors of the graph Laplacian matrix obtained in step 2, the original signal samples are converted into a series of GFT values related to the eigenvalue order;

[0046] Step 4: Extract key orders: Arrange the GFT values obtained from the monitoring signals at different wear stages in ascending order, take the eigenvalue orders corresponding to the first k GFT values as key orders, and combine the key orders of each wear stage to obtain a total of h (h ≥ k) key orders;

[0047] Step 5: Classification of typical order combination patterns: Based on the h key orders and the dimension parameter m, all order combination patterns are obtained, and a typical pattern library that can establish a good mapping relationship with different wear stages is obtained;

[0048] Step 6: Identification of tool wear status: The order combination pattern obtained by processing the collected monitoring signals through steps 1 to 5 is matched with the typical pattern library to realize tool wear status identification. When the wear status is in the rapid wear stage, it is instructed to replace the tool in time.

[0049] Optionally, the monitoring signal mentioned in step 1 mainly refers to the time series signal collected during the cutting process of the tool, including but not limited to force signal, vibration signal, sound signal, current signal, etc. The monitoring signal with a given length N can be expressed as X = {x1, x2, ..., x N};

[0050] Optionally, the different wear stages described in step 1 mainly refer to the different stages that the tool goes through according to wear characteristics during continuous use, including three stages: initial wear stage, steady wear stage and rapid wear stage.

[0051] Optionally, the graph theory described in step 2 mainly refers to a method of obtaining the structural properties of a graph through algebraic methods such as the adjacency matrix and Laplace matrix of the graph. The specific process is as follows:

[0052] For an undirected and connected weighted graph containing N vertices, it can be expressed as G = (V, E, A), where G is a weighted graph, V = (v1, v2, ..., v N ) represents the set of vertices in the graph, |V|=N means there are N vertices in total; E=(e1,e2,…,e M ) represents the set of edges in the graph, |E|=M means there are M edges in total. When two vertices v i With v j Adjacent, that is, vertex v i With v j When there is an edge connection between them, it can be recorded as e ij , i=1,2,…,N and j=1,2,…,N; A is the weighted adjacency matrix, A=(a ij ) N×N , when two vertices v i With v j When adjacent, a ij Represents edge e ij The weighted ij =1; when two vertices v i With v j When not adjacent, a ij = 0. It can be seen that the matrix A is a matrix in which all the sub-diagonal elements are 1 and all other elements are 0.

[0053] Define the degree diagonal matrix D = diag(d1,d2,...,d N ) is a diagonal matrix, the diagonal elements di Let it be vertex v i It can be seen that the matrix D represents a diagonal matrix with the first and last elements of the main diagonal being 1 and the remaining diagonal elements being 2.

[0054]

[0055] The graph Laplacian matrix is defined as L = DA. The graph Laplacian matrix L is based on the weighted adjacency matrix A and the degree diagonal matrix D. It contains both the degree information of the vertices and the association information between the vertices. The graph Laplacian matrix L is a real symmetric matrix, so it can be completely orthogonal decomposition to obtain a set of non-negative real eigenvalues σ(L) = {λ1,λ2,…,λ N} and the corresponding eigenvector Y={y1,y2,…,y N}. That is, the graph Laplace matrix L satisfies:

[0056] Ly i =λ i y i

[0057] Where i=1,2,…,N represents the sequence number of the vertex in the graph. i is the eigenvalue, and the eigenvalue σ(L) satisfies 0=λ1≤λ 2≤ …≤λ N =λ max .y i represents the eigenvalue λ i The corresponding eigenvector.

[0058] Optionally, the graph signal described in step 2 mainly refers to the method of converting the monitoring signal into a path graph. A path graph is a typical graph in graph theory, representing an undirected graph from one vertex to another, connected sequentially by edges. The unique points and edges traversed in the middle constitute the vertex set V and the edge set E.

[0059] The monitoring signal, as a discrete time series that varies over time and is sequentially connected, can be viewed as a special kind of road graph: each sampling point in the monitoring signal corresponds to a vertex in the road graph, the value of the sampling point corresponds to the vertex value of the road graph, and the sequential connection relationship of the sampling points in the monitoring signal obtained according to the sampling frequency corresponds to the edge connection relationship in the road graph (i.e., the structure of the road graph). Based on the above conversion relationship, the one-dimensional monitoring signal can be converted into a road graph-like graph signal G = (V, E, A). This graph signal G is also the weighted graph mentioned above.

[0060] Optionally, the graph Fourier transform (GFT) described in step 3 mainly refers to a method of decomposing the graph signal G according to the eigenvector Y of the graph Laplacian matrix L. The calculation method of the GFT of the signal X (from the original signal sample) is as follows:

[0061]

[0062] Where, GFT l represents the eigenvalue λ l The corresponding GFT value, l = 1, 2, ..., N represents the eigenvalue order of the GFT of signal X, x n represents the amplitude of the nth sampling point in the signal X, represents the eigenvalue λ l The corresponding eigenvector The nth element value in .

[0063] Optionally, the key order described in step 4 primarily refers to the eigenvalue order closely associated with each wear stage of the tool, reflecting the wear stage of the tool and can be considered a sensitive feature. For each wear stage, N GFT values of the original signal are calculated, and the eigenvalue orders corresponding to the first k GFT values are taken as the key order. Since tool wear is divided into three stages in the present invention, the k key orders for each stage may be different, resulting in a total of h (h ≥ k) key orders.

[0064] Optionally, the order combination pattern described in step 5 mainly refers to an order combination pattern determined according to h key orders and a dimension parameter m. The calculation method of the order combination pattern is as follows:

[0065] Determine the dimension parameter m (m≤h), and randomly select m from the h key orders to obtain all possible combinations. There are C combinations in total, which are calculated as follows:

[0066]

[0067] Each order combination mode is recorded as c represents the order combination mode number, c = 1, 2, ..., C. Taking h = 4, m = 3 as an example, there are C = 4 combinations. Assuming that the key orders are 1, 21, 44 and 86 respectively, the four order combinations are The order of the orders in the combination is arranged in descending order of the GFT amplitude corresponding to the order, which indicates the importance of the order.

[0068] Optionally, the typical pattern library described in step five mainly refers to selecting typical order combination patterns that have mapping relationships with the three tool wear stages, and constructing a correlation model between the arrangement pattern and the tool wear state.

[0069] Optionally, the pattern matching with the typical pattern library described in step 6 mainly refers to obtaining the order combination pattern of the signal sample to be tested according to steps 1 to 5, matching the order combination pattern of the signal sample to be tested with the typical pattern library, and if the order combination pattern of the signal sample to be tested exists in the typical pattern of a certain wear stage, then identifying the signal sample to be tested as the wear stage. Specifically, the detection signal in step 1 includes training data and signal samples to be tested (i.e., test data), wherein the training data is used as the original signal sample to obtain the typical pattern library (also referred to as the typical key order pattern library) through the processing and training of steps 1 to 5 above, and the signal sample to be tested is used as the original signal sample to obtain the order combination pattern of the signal sample to be tested through the processing of steps 1 to 5. In step 6, the order combination pattern of the signal sample to be tested is matched with the typical pattern library, thereby identifying the wear stage corresponding to the signal sample to be tested. Then, a judgment instruction is made as to whether to replace the tool, etc. Optionally, the typical pattern library can also be referred to as the typical order pattern library.

[0070] The following is a further detailed description of an exemplary embodiment of the application of the embodiment of the present invention. In this exemplary embodiment, the full life cycle data of the tool cutting of a high-speed CNC machine tool is used as the basis for test analysis, and the tool is a 3-tooth ball head carbide milling cutter. The full life cycle of the tool is divided into three stages, namely the initial wear stage, the steady wear stage and the rapid wear stage. The cutting force signal of the cutting process is collected using a three-axis dynamometer as a monitoring signal, and the sampling frequency is 50kHz. The spindle speed of the cutting test is 10400r / min, the feed speed is 1555mm / min, the axial cutting depth is 0.2mm, and the cutting amount per time is 0.001mm.

[0071] Figure 1 The following is a flow chart of a tool wear condition monitoring method based on graph Fourier transform and key order according to an embodiment of the present invention. Figure 1 , a tool wear condition monitoring method based on Fourier transform and key order provided by an embodiment of the present invention is described, and the method includes the following steps.

[0072] Step 1: Monitoring Signal Acquisition: A dynamometer was used to collect cutting force signals at different stages of tool wear during the cutting process as monitoring signals. During the experiment, each face milling pass was 108 mm long, with a total of 315 passes. After each pass, the machine was stopped for inspection, and the tool flank wear was measured using a microscope. Figure 2 The wear curves of the three cutting surfaces of the tool are shown. The data of the 10th to 30th, 100th to 120th, and 290th to 310th passes are selected as training data for the initial wear stage, the steady wear stage, and the rapid wear stage, respectively. In each pass, a data point of length N = 2048 is selected as a signal sample. Figure 3 The raw signal diagram of the cutting force at different wear stages of the tool is shown.

[0073] Step 2: Graph signal conversion: Each sampling point in the signal sample is regarded as a vertex in the graph signal, and the sampling point value is the vertex value. There are 2048 vertices in total. i With v i+1 There is an edge connection between them, that is, the edge weight a i,i+1 = 1. According to the relevant formula, we get the degree diagonal matrix D and weighted adjacency matrix A of the graph signal, and calculate the graph Laplacian matrix L = D–A. We perform orthogonal decomposition on the matrix L and get the eigenvalues of its graph Laplacian matrix σ(L) = {λ1,λ2,…,λ 2048} and the eigenvector Y={y1,y2,…,y 2048}.

[0074] Step 3: Graph Fourier transform: Use the eigenvector Y={y1,y2,…,y 2048} and related formulas, the monitoring signals of different wear stages are converted into GFT values related to the order of eigenvalues. The GFT spectra of the tool at different wear stages are as follows: Figure 4 shown.

[0075] Step 4: Extract key orders. Arrange the GFT values obtained from the training data samples at different wear stages in descending order. The eigenvalue orders corresponding to the first k GFT values are used as key orders. In this example, k = 5. The key orders obtained for each wear stage are shown in Table 1, for a total of h = 6 key orders.

[0076] Table 1 Description of vibration dataset

[0077]

[0078] Step 5: Classification of Typical Order Combination Patterns: Taking the dimension parameter m = 3, there are a total of C = 20 order combination patterns. The combination patterns that appear in the training data for different wear stages are counted, as shown in Table 1, to obtain a library of typical patterns for key orders and different wear stages. Figure 5 The typical order combination patterns of different wear stages are shown.

[0079] Step 6: Tool wear status identification: Use the remaining tool data except the training data as test data, and obtain its key order combination pattern according to the process of steps 1 to 5. Match the pattern with the typical pattern library to realize tool wear status identification. Figure 6 The wear stage identification results of the tool for 315 passes are shown. It can be seen that the method provided by the present invention can adaptively identify the three wear stages of the tool. The number of passes corresponding to each stage is shown in Table 1.

[0080] Based on the above tool wear status monitoring results, relevant staff can identify the tool wear stage in a timely manner, and according to the product's processing accuracy and other requirements, perform operations such as tool change when the tool enters the later wear stage, avoiding product quality degradation due to excessive tool wear, and ensuring the qualified rate requirements while fully utilizing the tool's use value.

[0081] The present invention proposes a tool wear state monitoring method based on graph Fourier transform and key orders, which can utilize the state monitoring data of the tool during the cutting process to realize the wear stage identification of the tool from the perspective of the graph. Compared with traditional signal processing methods, graph signal processing pays more attention to the structure and conversion relationship between data rather than the amplitude of the data, and has a strong anti-noise ability. The key order combination obtained according to the graph Fourier transform has a high sensitivity to the stage change of the tool, is simple to calculate, and can meet the real-time requirements of tool wear state monitoring. The model proposed in the present invention is not only applicable to the above-mentioned specific case implementation method, but also applicable to the state monitoring of other mechanical equipment and their key parts. The present invention has good expansion capabilities and provides an effective reference for other technical personnel in this field.

[0082] It should be noted that the flowcharts and block diagrams in the accompanying drawings show the possible architectures, functions and operations of the systems, methods and computer program products according to multiple embodiments of the present invention. In this regard, each box in the flowchart or block diagram can represent a module, program segment or part of the code, and the module, program segment or part of the code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in an order different from that marked in the accompanying drawings. For example, two consecutive boxes can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flowchart, and the combination of boxes in the block diagram and / or flowchart, can be implemented using a dedicated hardware-based system that performs the specified function or action, or can be implemented using a combination of dedicated hardware and computer instructions.

[0083] Furthermore, the foregoing describes only some embodiments, which may be changed, modified, added and / or varied without departing from the scope and spirit of the disclosed embodiments, which are illustrative and not restrictive. Furthermore, the embodiments described relate to what are currently considered to be the most practical and preferred embodiments, and it should be understood that the embodiments should not be limited to the disclosed embodiments, but rather are intended to cover different modifications and equivalent arrangements that are included within the spirit and scope of the embodiments. Furthermore, the various embodiments described above may be used in conjunction with other embodiments, such as aspects of one embodiment may be combined with aspects of another embodiment to achieve yet another embodiment. Additionally, each independent feature or component of any given component may constitute another embodiment.

[0084] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or make equivalent replacements for some or all of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the scope of the technical solutions of the embodiments of the present invention, and they should all be included in the scope of the claims and description of the present invention.

Claims

1. A tool wear condition monitoring method based on graph Fourier transform and key order, characterized in that The following steps are involved: Step 1: Monitoring signal acquisition: Collect monitoring signals of the tool at different wear stages during the cutting process, and select signals of a given length as the original signal samples; Step 2: Graph signal conversion: According to graph spectrum theory, the one-dimensional signal in the original signal sample is converted into a graph signal, the graph Laplacian matrix of the graph signal is calculated, and the eigenvalues and eigenvectors of the graph Laplacian matrix are extracted; Step 3: Graph Fourier Transform (GFT): Based on the Graph Fourier Transform and the eigenvectors of the Laplace matrix obtained in step 2, the original signal samples are converted into a series of GFT values related to the order of the eigenvalues; Step 4: Extract key orders: Arrange the GFT values obtained from the monitoring signals at different wear stages in ascending order, take the eigenvalue orders corresponding to the first k GFT values as key orders, and combine the key orders of each wear stage to obtain a total of h (h ≥ k) key orders; Step 5: Classification of typical order combination patterns: Based on the h key orders and the dimension parameter m, all order combination patterns are obtained, and a typical pattern library with a good mapping relationship with different wear stages is obtained; Step 6: Identification of tool wear status: The order combination pattern obtained after the collected monitoring signal is processed through steps 1 to 5 is matched with the typical pattern library obtained after the training data is processed through steps 1 to 5 to realize tool wear status identification. When the wear status is in the rapid wear stage, it is instructed to replace the tool in time.

2. The tool wear condition monitoring method based on graph Fourier transform and key order according to claim 1, characterized in that: The different wear stages described in step 1 refer to the different stages that the tool goes through according to the wear characteristics when it is used continuously, including the initial wear stage, the steady wear stage and the rapid wear stage, a total of three stages.

3. The tool wear condition monitoring method based on graph Fourier transform and key order according to claim 2, characterized in that Step 2 further includes the following steps: For an undirected and connected weighted graph containing N vertices, it is represented as G = (V, E, A), where G is the graph signal of the weighted graph, V = (v1, v2, ..., v N ) represents the set of vertices in the graph, |V|=N means there are N vertices in total; E=(e1,e2,…,e M ) represents the set of edges in the graph, |E|=M means there are M edges in total. When two vertices v i With v j Adjacent, that is, vertex v i With v j When there is an edge connection between them, it is recorded as e ij , i=1,2,…,N and j=1,2,…,N; A is the weighted adjacency matrix, A=(a ij ) N×N , when two vertices v i With v j When adjacent, a ij Represents edge e ij The weighted ij =1; when two vertices v i With v j When not adjacent, a ij =0, the matrix A is a matrix in which all the sub-diagonal elements are 1 and all other elements are 0; Define the degree diagonal matrix D = diag(d1,d2,...,d N ) is a diagonal matrix, the diagonal elements d i Let it be vertex v i The degree of matrix D is a diagonal matrix with the first and last elements of the main diagonal being 1 and the remaining diagonal elements being 2. The graph Laplacian matrix is defined as L = DA. The graph Laplacian matrix L is based on the weighted adjacency matrix A and the degree diagonal matrix D. It contains both the degree information of the vertices and the association information between the vertices. The graph Laplacian matrix L is a real symmetric matrix. After performing a complete orthogonal decomposition, a set of non-negative real eigenvalues σ(L) = {λ1,λ2,…,λ N } and the corresponding eigenvector Y={y1,y2,…,y N }, that is, the graph Laplace matrix L satisfies: Ly i =λ i y i Where i = 1, 2, ..., N represents the sequence number of the vertex in the graph, λ i is the eigenvalue, and the eigenvalue σ(L) satisfies 0=λ1≤λ 2≤ …≤λ N =λ max ,y i represents the eigenvalue λ i The corresponding eigenvector.

4. The tool wear condition monitoring method based on graph Fourier transform and key order according to claim 3, characterized in that Step 2: Converting monitoring signals into road maps: The monitoring signal is a road graph, in which each sampling point in the monitoring signal corresponds to a vertex in the road graph, the value of the sampling point corresponds to the vertex value of the road graph, and the sequential connection relationship of the sampling points obtained according to the sampling frequency in the monitoring signal corresponds to the edge connection relationship in the road graph. According to this conversion relationship, the one-dimensional monitoring signal is converted into a graph signal G = (V, E, A) in the form of a road graph.

5. The tool wear condition monitoring method based on graph Fourier transform and key order according to claim 4, characterized in that: The graph Fourier transform in step 3 includes decomposing the graph signal G according to the eigenvector Y of the graph Laplace matrix L, specifically including: The GFT of signal X is calculated as follows: Where, GFT l represents the eigenvalue λ l The corresponding GFT value, l = 1, 2, ..., N represents the eigenvalue order of the GFT of signal X, x n represents the amplitude of the nth sampling point in the signal X, represents the eigenvalue λ l The corresponding eigenvector The nth element value in .

6. The tool wear condition monitoring method based on graph Fourier transform and key order according to claim 5, characterized in that: The key order described in step 4 refers to the eigenvalue order closely associated with each wear stage of the tool, reflecting the wear stage of the tool. For each type of wear stage, the N GFT values of the original signal are calculated, and the eigenvalue order corresponding to the first k GFT values is taken as the key order, resulting in a total of h (h ≥ k) key orders.

7. The tool wear condition monitoring method based on graph Fourier transform and key order according to claim 6, characterized in that: The order combination pattern described in step 5 is an order combination pattern determined according to h key orders and the dimension parameter m. The calculation method of the order combination pattern is as follows: Determine the dimension parameter m (m≤h), and randomly select m from the h key orders to obtain all possible combinations. There are C combinations in total, which are calculated as follows: Each order combination mode is recorded as c represents the order number of the combination mode, c = 1, 2, ..., C. The order of the order in the combination is arranged in descending order of the GFT amplitude corresponding to the order, indicating the importance of the order.

8. The tool wear condition monitoring method based on graph Fourier transform and key order according to claim 7, characterized in that: The typical pattern library described in step five is to select typical order combination patterns that have mapping relationships with the three tool wear stages, and to construct a correlation model between the arrangement pattern and the tool wear state.

9. The tool wear condition monitoring method based on graph Fourier transform and key order according to claim 8, characterized in that: The pattern matching with the typical pattern library described in step six refers to obtaining the order combination pattern of the signal sample to be tested according to steps one to five, and matching the order combination pattern of the signal sample to be tested with the typical order combination pattern library. If the order combination pattern of the signal sample to be tested exists in the typical pattern of a certain wear stage, then the signal sample to be tested is identified as the wear stage.

Images