Wear control method for high-temperature alloy laser heating cutting tool

Abstract

The invention discloses a wear control method for a high-temperature alloy laser heating cutting tool, and belongs to the field of machining. The method comprises the steps of laser heating auxiliary cutting technology, use of an infrared camera and a Kister three-way dynamometer, establishment of a BP neural network and determination of a membership function. Wherein the BP algorithm adopts a gradient descent method with momentum. Firstly, cutting force and cutting temperature in the laser-assisted cutting machining process are obtained and serve as input characteristic values, the tool abrasion loss is output through a BP neural network, then the tool abrasion level is obtained by determining a membership function of a tool state category through a tool cutting state mode, and therefore real-time monitoring of tool abrasion is achieved. And the tool wear level is jointly determined through the BP network and membership function calculation, so that the accuracy of the tool wear control method is improved, and the tool wear control method has a practical application value.

Description

Technical Field

[0001] The invention belongs to the field of mechanical processing, and in particular relates to a wear control method for a high-temperature alloy laser heating cutting tool. Technical Background

[0002] Metal cutting is a widely used processing method in modern mechanical manufacturing. Realizing intelligent metal cutting is one of the keys to intelligent mechanical manufacturing technology. Real-time monitoring technology of tool wear status is a key link in intelligent mechanical processing technology. Through the research of this technology, the intelligent metal cutting process can be promoted. In the material cutting process, the tool is in direct contact with the workpiece, and the gradual wear of the tool becomes an important factor affecting the workpiece processing dimensional accuracy, surface roughness and processing cost.

[0003] Laser heating assisted cutting technology was first proposed by Copley and Bass in 1978, and soon thereafter, it attracted widespread attention as a new method for processing difficult-to-process materials. It uses the high energy density of the laser to soften the surface of the material to be processed in front of the tool, and then the tool processes it, so that the material cutting process is transformed from the brittle domain to the plastic domain, thereby reducing the cutting force and extending the tool life.

[0004] From the principle of metal cutting, it is known that the tool wear state is closely related to the cutting force and cutting temperature. Therefore, by studying the cutting force and cutting temperature, the tool wear state can be monitored. Understanding the real-time state of tool wear is of great significance to improving cutting quality and efficiency and realizing intelligent mechanical manufacturing.

[0005] The existing single monitoring methods, whether direct or indirect, still have certain limitations in practical application. Based on summarizing the advantages and disadvantages of the existing monitoring methods, a multi-parameter fusion monitoring method is proposed. By selecting cutting force and cutting temperature as monitoring signals at the same time, the advantages of these two monitoring quantities can be used to complement each other, broaden the monitoring range, and improve the monitoring accuracy and success rate of discrimination.

[0006] During the machining process, Sun Yanjie et al. monitored the tool wear state based on cutting sound and cutting force, and used matlab to write a BP neural network. The output value of the neural network was basically consistent with the actual measurement value. Ding Ning et al. analyzed the tool wear state based on cutting temperature and established a mathematical model based on fuzzy membership function, which greatly improved the robustness of the traditional model and effectively solved the problem of real-time tool monitoring during machining. Xu Jinkai et al. conducted a laser heating-assisted turning experiment on nickel-based alloy GH4169 to study the effect of laser on tool wear during cutting. The experimental results showed that compared with ordinary turning, laser-assisted cutting can improve tool back face wear. Summary of the invention

[0007] The purpose of the present invention is to provide a method for controlling the wear of high-temperature alloy laser-heated cutting tools. The cutting force and cutting temperature in the laser-assisted cutting process are obtained and used as input characteristic values ​​to create a BP neural network, and the tool wear amount at the required time is output. Then, the membership function of the tool state category is determined by the tool cutting state mode to obtain the tool wear level, thereby achieving real-time monitoring of tool wear, and finally achieving the purpose of controlling tool wear. The tool wear level is determined by the BP network and the membership function calculation, which has practical application value.

[0008] The technical solution adopted by the present invention is:

[0009] A wear control method for high-temperature alloy laser heating cutting tools. First, a laser-assisted cutting processing platform is built, and the cutting temperature and cutting force are obtained by using an infrared temperature camera and a Kistler three-axis dynamometer. Then, a BP neural network is established, and the temperature and force are used as input neurons to output the tool wear. The specific steps are as follows:

[0010] Step 1: Clamp the workpiece on the selected machine tool;

[0011] Step 2: In order to obtain cutting force data, a Kistler three-axis dynamometer is used to obtain the three-axis cutting force;

[0012] Step 3: In order to obtain cutting temperature data, an infrared camera is used to record the temperature during the processing;

[0013] Step 4: In order to realize the laser heating assisted process, the laser spot and the tool need to remain relatively still, and the laser spot is located in front of the tool, such as Figure 2 ;

[0014] Step 5: Experiment with the new tool, set the machining parameters, and complete the cutting process;

[0015] Step 6: Get the cutting temperature T and three-way cutting force F during the cutting process x , F y , F z The change curve of

[0016] Step 7: Establish a BP neural network to detect the tool wear status to control tool wear.

[0017] Where T, F x , F y , F z is the input characteristic parameter, and O is the output of tool wear level.

[0018] The BP algorithm uses the gradient descent method with momentum and introduces the momentum factor η (0<η<1).

[0019]

[0020] x(k+1)=x(k)+Δx(k+1) (2)

[0021] Where α is the momentum term coefficient.

[0022] Step 8: Assume that the characteristic sample centers corresponding to the five tool states of new tool, initial wear, mid-term wear, late wear, and rapid wear are A1, A2, A3, A4, and A5, respectively, and A i ={μ i1 ,μ i2 ,…,μ im}, where μ im is the mth characteristic component of the characteristic sample center corresponding to the i-th tool state category; the characteristic sample of any tool cutting state mode is X={X1,X2,…,X n}Then the process of determining the membership function of the tool cutting state mode to the tool state category can be summarized as follows:

[0023] According to the Euclidean distance formula, the Euclidean distances between the cutting state mode of the tool to be identified and the known tool state category are:

[0024]

[0025] make

[0026] Then the closeness of the tool state pattern to be identified to the known tool state category can be obtained and used as the equivalent form of the membership function, that is:

[0027]

[0028] If μ Ai (X)>μ Aj (X), j = 1, 2, 3, 4, 5, j ≠ i then X∈A j ,, the tool cutting state model X to be identified can be considered to belong to the i-th type of tool state.

[0029] Beneficial effects of the present invention:

[0030] The high-temperature alloy laser heating cutting tool wear control method based on cutting force and cutting temperature provided by the present invention solves the problem of controlling tool wear in the laser-assisted cutting process and has great universality for workpiece materials. The method allows the use of a gradient descent method with momentum, which can reduce the network's sensitivity to local details of the error surface, thereby improving the network's learning speed and increasing the reliability of the algorithm. The tool wear level is determined jointly by the BP network and the membership function calculation, which increases the accuracy of the present invention and has practical application value. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 Schematic diagram of the process of the present invention;

[0032] Figure 2 This is a schematic diagram of the laser spot being located in front of the tool;

[0033] Figure 3 It is a schematic diagram of BP neural network;

[0034] Figure 4 This is a scatter plot diagram of Matlab. DETAILED DESCRIPTION

[0035] The present invention will be described in detail below in conjunction with the embodiments of the present invention.

[0036] The embodiments described in this section are only used to explain the present invention and are not used to limit the present invention.

[0037] Step 1: Clamp GH4169 on the selected machine tool;

[0038] Step 2: In order to obtain cutting force data, the three-dimensional cutting force is obtained using a Kister5017A three-dimensional dynamometer;

[0039] Step 3: To obtain cutting temperature data, use a FLIR ResearchIR Max4 thermal imager to record the temperature during processing;

[0040] Step 4: In order to realize the laser heating assisted process, the laser spot and the tool need to remain relatively still, and the laser spot is located in front of the tool, such as Figure 2 ;

[0041] Step 5: Experiment with the new tool, set the machining parameters, and complete the cutting process;

[0042] Step 6: Get the cutting temperature T and three-way cutting force F during the cutting process x , F y , F z The change curve of

[0043] Step 7: Establish BP neural network to detect tool wear status to control tool wear, such as Figure 3 .

[0044] Where T, F x , F y , F z is the input characteristic parameter, and O is the output of tool wear level.

[0045] Eight groups of characteristic parameter values ​​at different times are selected as input samples.

[0046] The BP algorithm uses the gradient descent method with momentum and introduces the momentum factor η (0<η<1).

[0047]

[0048] x(k+1)=x(k)+Δx(k+1) (2)

[0049] Where α is the momentum term coefficient.

[0050] According to the reference formula for selecting the best hidden unit:

[0051]

[0052] Among them, m is the number of output neurons, n is the number of input neurons, and a is a constant between [1, 10], so n1 is 8. That is, the optimal number of hidden layer neurons is 8. At the same time, the learning rate of the system is taken as 0.05.

[0053] In order to ensure faster convergence when the program is running, normalization processing must be performed before training to normalize the data to [-1, 1]. A matlab program is written and the following is obtained: Figure 4 The scatter plot shown.

[0054] Step 8: Assume that the characteristic sample centers corresponding to the five tool states of new tool, initial wear, mid-term wear, late wear, and rapid wear are A1, A2, A3, A4, and A5, respectively, and A i ={μ i1 ,μ i2 ,…,μ im}, where μ im is the mth characteristic component of the characteristic sample center corresponding to the i-th tool state category; the characteristic sample of any tool cutting state mode is X={X1,X2,…,X n}Then the process of determining the membership function of the tool cutting state mode to the tool state category can be summarized as follows:

[0055] According to the Euclidean distance formula, the Euclidean distances between the cutting state mode of the tool to be identified and the known tool state category are:

[0056]

[0057] make

[0058] Then the closeness of the tool state pattern to be identified to the known tool state category can be obtained and used as the equivalent form of the membership function, that is:

[0059]

[0060] If μ Ai (X)>μAj (X), j = 1, 2, 3, 4, 5, j ≠ i then X∈A j ,, the tool cutting state model X to be identified can be considered to belong to the i-th type of tool state.

Claims

1. A method for controlling wear of a high-temperature alloy laser-heated cutting tool, characterized in that: Here are the steps: Step 1: Clamp the workpiece on the selected machine tool; Step 2: Obtain three-dimensional cutting force data; Step 3: Obtain cutting temperature data and use an infrared camera to record the temperature during processing; Step 4: The laser spot and the tool must remain relatively still, and the laser spot is located in front of the tool; Step 5: Experiment with the new tool, set the machining parameters, and complete the cutting process; Step 6: Get the cutting temperature T and three-way cutting force F during the cutting process x , F y , F z The change curve of Step 7: Establish a BP neural network to detect the tool wear state to control tool wear; Where T, F x , F y , F z is the input characteristic parameter, O is the output of tool wear level; Step 8: Assume that the characteristic sample centers corresponding to the five tool states of new tool, initial wear, mid-term wear, late wear, and rapid wear are A1, A2, A3, A4, and A5, respectively, and A i ={μ i1 ,μ i2 ,…,μ im }, where μ im is the mth characteristic component of the characteristic sample center corresponding to the i-th tool state category; the Euclidean distances between the cutting state mode of the tool to be identified and the known tool state category are: make The closeness of the tool state pattern to be identified to the known tool state category is obtained and used as the equivalent form of the membership function, that is: If μ Ai (X)>μ Aj (X), j = 1, 2, 3, 4, 5, j ≠ i then X∈A j , that is, the cutting state model X of the tool to be identified can be considered to belong to the i-th type of tool state.

2. A high temperature alloy laser heating cutting tool wear control method as claimed in claim 1, characterized in that: In the step 2, a Kistler three-axis dynamometer is used to obtain the three-axis cutting force.

3. A high temperature alloy laser heating cutting tool wear control method as claimed in claim 1 or 2, characterized in that: In step 3, an infrared camera is used to record the temperature during the processing.

4. A high temperature alloy laser heating cutting tool wear control method as claimed in claim 1 or 2, characterized in that: In step 7, the BP algorithm adopts a gradient descent method with momentum, introducing a momentum factor η (0<η<1); x(k+1)=x(k)+Δx(k+1) (2) Where α is the momentum term coefficient.

5. A high temperature alloy laser heating cutting tool wear control method as claimed in claim 3, characterized in that: In step 7, the BP algorithm adopts a gradient descent method with momentum, introducing a momentum factor η (0<η<1); x(k+1)=x(k)+Δx(k+1) (2) Where α is the momentum term coefficient.

6. A method for controlling wear of a high-temperature alloy laser-heated cutting tool as claimed in claim 1, 2 or 5, characterized in that: In step 8, the characteristic sample of any tool cutting state mode is X = {X1, X2, ..., X n }.

7. A high temperature alloy laser heating cutting tool wear control method as claimed in claim 3, characterized in that: In step 8, the characteristic sample of any tool cutting state mode is X = {X1, X2, ..., X n }.

8. A method for controlling wear of a high-temperature alloy laser-heated cutting tool as claimed in claim 4, characterized in that: In step 8, the characteristic sample of any tool cutting state mode is X = {X1, X2, ..., X n }.

Images