Laser heating turning component outer circle surface residual stress prediction method based on ABAQUS and XRD experiments

Abstract

The invention discloses a method for predicting residual stress on the outer circle surface of a laser heating turning component based on ABAQUS and XRD experiments, and belongs to the field of machining. The method comprises ABAQUS finite element analysis, a laser heating assisted turning technology, an XRD experiment and a neural network prediction method. The method comprises the following steps: firstly, establishing a laser heating auxiliary turning model based on finite element analysis software ABAQUS to obtain a reasonable range of laser parameters and cutting parameters, carrying out laser heating turning on an outer circle component according to different process parameters obtained by the ABAQUS, and then carrying out XRD (X-Ray Diffraction) experiment to detect internal residual stress as a training sample of a neural network; the samples are used for training a neural network model to obtain a neural network with a prediction function, process parameters of verification samples are input into the neural network, and a predicted residual stress value is calculated and compared with an actual detection value.

Description

Technical Field

[0001] The invention belongs to the field of mechanical processing and relates to a method for predicting residual stress on the outer cylindrical surface of a laser-heated turning component based on ABAQUS and XRD experiments. Technical Background

[0002] 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, improving the quality of the processed surface.

[0003] Due to the interaction between cutting parameters such as cutting speed and cutting depth and laser parameters such as laser power, residual stress is almost unavoidable in laser-assisted processing, which has a significant impact on the precision, life and other performance of parts. Therefore, how to accurately predict residual stress and reversely derive more appropriate process parameters is of great research significance.

[0004] Neural networks can approximate highly nonlinear functions and find the relationship between output and input through adaptive learning of known samples. They have high fault tolerance and parallel processing capabilities for nonlinear information. They are suitable for modeling uncertain complex nonlinear system models and have been widely used. In terms of data prediction, radial basis function (RBF) neural networks have significant advantages. They do not need to establish mathematical models. Through a large amount of training, they can find special mappings between input and output. Applying them to residual stress prediction can improve prediction accuracy and speed. Taking advantage of the strong ability of RBF neural networks in predicting multivariable and complex linear information processing, a model suitable for predicting residual stress on the outer cylindrical surface of laser heated turning components is established. Summary of the invention

[0005] The purpose of the present invention is to provide a method for predicting residual stress on the outer cylindrical surface of a laser-heated turned component based on ABAQUS and XRD experiments. First, a laser heating-assisted turning model is established based on the finite element analysis software ABAQUS to obtain a reasonable range of laser parameters and cutting parameters. Laser heating and turning of outer cylindrical components are performed according to different process parameters obtained by ABAQUS. Then, an XRD experiment is performed to detect the internal residual stress. The results are divided into two parts, one as a training sample and the other as a verification sample. An RBF neural network prediction model is constructed, and the training sample is used to train the neural network model to obtain a neural network with a prediction function. The process parameters of the verification sample are input into the neural network, and the predicted residual stress value is calculated and compared with the actual detection value of the verification sample. The flow chart is shown as follows. Figure 1 .

[0006] The technical solution implemented in the present invention is as follows:

[0007] The residual stress prediction method of the outer surface of the laser-heated turning component based on ABAQUS and XRD experiments is as follows:

[0008] Step 1: Establish a finite element analysis model of laser heating assisted turning based on ABAQUS;

[0009] Step 2: Obtain reasonable parameters for laser heating turning of outer cylindrical components from step 1;

[0010] Step 3: Based on the parameters obtained in step 2, conduct laser heating turning experiments on external cylindrical components;

[0011] Step 4: Continue to perform XRD experiment to detect internal residual stress, and divide the results into two parts, one part as the training sample of the neural network and the other part as the verification sample to obtain the actual detection value;

[0012] Step 5: Establish an RBF neural network model for predicting residual stress;

[0013] Step 6: Use the training samples in step 4 to train the neural network to obtain a neural network with prediction function;

[0014] Step 7: Input the process parameters of the verification sample into the neural network to calculate the predicted residual stress value;

[0015] Step 8: Compare the predicted residual stress values ​​with the actual detected values.

[0016] The experimental equipment adopted by the present invention is: a laser, a temperature meter, a dynamometer and a cutting tool.

[0017] Beneficial effects of the present invention:

[0018] The method for predicting residual stress on the outer cylindrical surface of a laser-heated turned component based on ABAQUS and XRD experiments provided by the present invention can well predict the influence of turning parameters on the residual stress of the constructed surface and has great universality for workpiece materials. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 is a flow chart of the present invention;

[0020] Figure 2 It is a schematic diagram of the experimental equipment of the present invention;

[0021] Figure 3 This is a network structure design diagram of an embodiment of the present invention. DETAILED DESCRIPTION

[0022] The present invention will be described in detail below in conjunction with the embodiments of the present invention.

[0023] The embodiments described in this section are only used to explain the present invention and are not used to limit the present invention.

[0024] Step 1: Establish a finite element analysis model for laser heating assisted turning based on ABAQUS, define the workpiece material as GH4169, and the Johnson-Cook constitutive model and failure criterion are:

[0025] <![CDATA[d1]]> <![CDATA[d2]]> <![CDATA[d3]]> <![CDATA[d4]]> <![CDATA[d5]]> Melting point Transition temperature Reference strain rate 0.11 0.75 -1.45 0.04 0.89 1573 20 1

[0026] A B n m Melting temperature Transition temperature 1290 895 0.526 1.55 1573 20

[0027] Step 2: Obtain reasonable parameters of the outer surface of the component analyzed by ABAQUS from step 1;

[0028] Step 3: Based on the parameters obtained in step 2, the laser heating turning experiment of the outer cylindrical component is carried out. The schematic diagram of the experimental equipment is as follows: Figure 2 ;

[0029] Step 4: Continue to perform XRD experiments to detect internal residual stress, with one part used as training samples for the neural network and the other part used as verification samples to obtain actual test values;

[0030] Step 5: Establish an RBF neural network model for predicting residual stress based on ABAQUS process parameters. First, the radial basis function expansion coefficient SPREAD is selected as 0.9; secondly, the RBF neural network includes a three-layer structure of input layer, hidden layer and output layer, and the number of nodes is set to n, t, and m respectively. The influencing factors of laser heating turning residual stress include cutting depth a p , cutting speed V c and feed rate per revolution f, laser power P, so the number of input neurons of the network is n = 4; the output result is residual stress, then the number of output neurons is m = 1. The input vector is represented as P = [p1, p2, ..., p n ], where p i Because [a p , V c , f, P], and the corresponding expected output is expressed as where y i =σ′ s The output of the i-th hidden unit is the radial basis function φ(‖Xc i ‖), and c=[c i1 , c i2 , …, c in ] is the center of the function, ω ij(i=1, 2, ..., t; j=1) is the weight between the i-th hidden unit and the only output unit (residual stress), then the output of the RBF network is: where σ i is the expansion coefficient of the basis function of the i-th node in the hidden layer (σ i >0), which is usually determined through debugging or optimization. The specific network structure design is as follows: Figure 3 .

[0031] Step 6: When setting the input and output vectors of the network, unify the parameter dimensions through appropriate normalization, as follows: Where: p i is the i-th component of the input vector; and They are p i The output is the residual stress σ s , the corresponding output vector is Y = [σ s ], to ensure data uniformity, the output vector is processed in the same way as above. Other parameters are set as follows: the network fitting error target value is set to 0.001, the SPREAD value is set to 0.9, the number of hidden layer units can be adaptively programmed, and the training samples in step 4 are used to train the network to obtain the optimal network structure.

[0032] Step 7: Input the process parameters of the verification sample into the neural network to calculate the predicted residual stress value;

[0033] Step 8: Compare the predicted residual stress values ​​with the actual test values ​​to verify the model.

Claims

1. A method for predicting residual stress on the outer surface of a laser-heated turning component based on ABAQUS and XRD experiments, characterized in that: Here are the steps: Step 1: Establish a finite element analysis model of laser heating assisted turning based on ABAQUS; Step 2: Obtain the parameters of laser heating turning of the outer cylindrical component according to step 1; Step 3: Based on the parameters obtained in step 2, conduct laser heating turning experiments on external cylindrical components; Step 4: Continue to perform XRD experiment to detect internal residual stress, and divide the results into two parts, one part as the training sample of the neural network and the other part as the verification sample to obtain the actual detection value; Step 5: Establish an RBF neural network model for predicting residual stress; Step 6: Use the training samples in step 4 to train the neural network to obtain a neural network with prediction function; Step 7: Input the process parameters of the verification sample into the neural network to calculate the predicted residual stress value; Step 8: Compare the predicted residual stress values ​​with the actual detected values.

2. The method for predicting residual stress on the outer surface of a laser-heated turning component based on ABAQUS and XRD experiments according to claim 1, characterized in that: In step 3, the experimental equipment includes: a laser, a thermometer, a dynamometer, and a tool.

3. The method for predicting residual stress on the outer surface of a laser-heated turned component based on ABAQUS and XRD experiments according to claim 1 or 2, characterized in that: In step 5, the RBF neural network includes a three-layer structure of an input layer, a hidden layer and an output layer, and the number of nodes is set to n, t, and m respectively. The influencing factors of the residual stress of laser heating turning include the cutting depth a p , cutting speed V c and feed per revolution f, laser power P, so the number of input neurons of the network is n=4; if the output result is residual stress, the number of output neurons is m=1.

4. The method for predicting residual stress on the outer surface of a laser-heated turned component based on ABAQUS and XRD experiments according to claim 1 or 2, characterized in that: In step 6, the parameter dimensions are unified by normalization processing, and the method is as follows: Where: p i is the i-th component of the input vector; and They are p i The maximum and minimum values ​​of .

5. The method for predicting residual stress on the outer surface of a laser-heated turned component based on ABAQUS and XRD experiments as claimed in claim 3, characterized in that: In step 6, the parameter dimensions are unified by normalization processing, and the method is as follows: Where: p i is the i-th component of the input vector; and They are p i The maximum and minimum values ​​of .

6. The method for predicting residual stress on the outer surface of a laser-heated turned component based on ABAQUS and XRD experiments as claimed in claim 1, 2 or 5, characterized in that: In step 5, the output of the RBF network is: where σ i is the expansion coefficient of the basis function of the i-th node in the hidden layer (σ i >0).

7. The method for predicting residual stress on the outer surface of a laser-heated turned component based on ABAQUS and XRD experiments as claimed in claim 3, characterized in that: In step 5, the output of the RBF network is: where σ i is the expansion coefficient of the basis function of the i-th node in the hidden layer (σ i >0).

8. The method for predicting residual stress on the outer surface of a laser-heated turned component based on ABAQUS and XRD experiments according to claim 4, characterized in that: In step 5, the output of the RBF network is: where σ i is the expansion coefficient of the basis function of the i-th node in the hidden layer (σ i >0).

Images