A method for predicting bauschinger effect of metal sheet by using BP neural network
By constructing a finite element simulation model and a BP neural network, the problems of low efficiency and high cost in the traditional determination of the Bauschinger effect in thin metal plates are solved, achieving high-precision and low-cost prediction of the Bauschinger effect and simplifying the operation process.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- INNER MONGOLIA UNIVERSITY
- Filing Date
- 2022-05-17
- Publication Date
- 2026-07-14
AI Technical Summary
Traditional methods for determining the Bagsinger effect in thin metal plates are inefficient, costly, and difficult, and are prone to failure due to instability.
A finite element simulation model of three-point bending deformation under a specific pre-tension was constructed. The Bauschinger effect of the thin metal plate was predicted by a BP neural network. The load-displacement curve and Bauschinger stress parameters were obtained by finite element model simulation, and the BP neural network model was trained for prediction.
It achieves high-precision, low-cost prediction of the Bauschinger effect, simplifies the operation process, avoids equipment modifications, and improves measurement efficiency and accuracy.
Smart Images
Figure CN114970256B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to a performance testing technology for thin metal sheets, specifically a method for predicting the Bauschinger effect in thin metal sheets using a BP neural network. Background Technology
[0002] When a metallic material undergoes plastic deformation in one direction, its yield strength decreases if loading continues in the opposite direction; this phenomenon is known as the Bauschinger effect. The Bauschinger effect exists in most metallic materials, and its application in controlling the springback of sheet metal forming is of great significance. Especially with the widespread use of various alloys and high-strength steels in pipeline construction, automobile manufacturing, and other fields, accurate measurement of the Bauschinger effect has become crucial for improving product performance and optimizing manufacturing processes.
[0003] Traditional methods for determining the Bauschinger effect in thin metal sheets still rely on experimental methods. A key problem with this approach is that the thin sheet is prone to instability during reverse loading, leading to experimental failure. Therefore, this requires designing specialized fixtures and developing specialized devices to address this issue, resulting in low efficiency, high cost, and significant technical difficulty. Summary of the Invention
[0004] The purpose of this invention is to provide a method for predicting the Bauschinger effect in thin metal sheets using a BP neural network. This method is simple to operate, low in cost, has few influencing factors, and accurately predicts the Bauschinger effect.
[0005] To achieve the above-mentioned objectives, the present invention adopts the following technical solution:
[0006] This invention provides a method for predicting the Bagsinger effect in thin metal plates using a backpropagation neural network, comprising:
[0007] 1) Construct a finite element simulation model of the three-point bending deformation process under a specific pre-tension amount, and design a pre-tension three-point bending test with the same parameters to verify the reliability of the finite element model.
[0008] 2) The load-displacement curves of various hypothetical thin plate materials during three-point bending were obtained by finite element model simulation, and the Bauschinger stress parameters of each thin plate material were calculated by applying the Bauschinger effect theory under linear kinematic hardening constitutive model. A sample set containing the mapping relationship between load-displacement curves and Bauschinger stress parameters was constructed.
[0009] 3) Construct a prediction model of the Bauschinger effect based on a BP neural network and train it using a sample set.
[0010] 4) Obtain the load-displacement curve of the metal sheet to be tested by performing a pre-stretch three-point bending test with the same parameters as in step 1), input the trained BP network Bauschinger effect prediction model, and predict its Bauschinger effect.
[0011] According to one embodiment of the present invention, the pre-stretching amount mentioned in step 1) needs to be obtained by stretching the thin plate.
[0012] The tensile test is used to determine the tensile strength of the sheet metal. The stress applied to the sheet metal during the tensile test should be less than the ultimate tensile strength of the sheet metal. The purpose is to ensure that the Bauschinger effect is clearly characterized in the load-displacement curve obtained from bending, and to avoid fracture or back cracks in the specimen during pre-stretching and three-point bending continuous deformation.
[0013] According to one embodiment of the present invention, the evaluation index of the Bauschinger effect in step 2) is the Bauschinger stress parameter. Its value is calculated as follows:
[0014] (1)
[0015] Where: σ t —Stress at the positive unloading point, σ r —Reverse yield stress.
[0016] Where σ t σ r Applying the Bauschinger effect theory in linear follower-reinforced constitutive model (with appendix) Figure 1 ), calculated using equations (2) and (3):
[0017] (2)
[0018] In the formula: E, E P These are the elastic modulus and the plastic modulus, respectively. σ s For the positive yield stress, ε s For positive yield strain, ε t The strain is the strain at the positive unloading point.
[0019] (3)
[0020] According to one embodiment of the present invention, the input of the Bauschinger effect prediction model based on BP neural network in step 3) is a column matrix composed of the horizontal and vertical coordinates of 30 points randomly selected on the load-displacement curve, and the output is the Bauschinger stress parameter.
[0021] According to one embodiment of the present invention, the pre-tension three-point bending test parameters described in step 4) are completely consistent with the pre-tension three-point bending finite element simulation model in step 1).
[0022] Compared with the prior art, the beneficial effects of the present invention are as follows:
[0023] This invention constructs a pre-stretched three-point bending finite element analysis model to simulate the load-displacement curves of various hypothetical thin plate materials. Using the Bauschinger effect theory in linear kinematic hardening, the Bauschinger stress parameters of each hypothetical material are calculated. A sample set is constructed that includes the mapping relationship between the macroscopic response (load-displacement curve) and the Bauschinger stress parameter values of the thin plate material. A Bauschinger effect BP network prediction model is then trained, with the load-displacement curve as input and the Bauschinger stress parameters as output. The load-displacement curve of the metal thin plate under test during the three-point bending process is then input into the model to predict its Bauschinger stress parameter values. The prediction accuracy is high, effectively describing the Bauschinger effect of the metal thin plate.
[0024] This invention does not require the improvement or development of experimental equipment; all can be achieved using conventional equipment and methods. It is convenient and simple to operate, highly efficient, low in cost, has few influencing factors, and yields accurate parameters.
[0025] The Bauschinger stress parameter value predicted by this invention can play a key role in calculating the springback of thin metal sheets and optimizing the manufacturing process. Attached Figure Description
[0026] Figure 1 The theoretical diagram of the Bauschinger effect under the linearly responsive reinforced constitutive model.
[0027] Figure 2 This is a schematic diagram of the finite element simulation and test principle for pre-stretched three-point bending.
[0028] Figure 3 This is a flowchart illustrating the prediction of the Bagsinger effect in thin metal sheets according to the present invention. Detailed Implementation
[0029] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which this application belongs. The terminology used herein is for the purpose of describing specific embodiments only and is not intended to limit this application.
[0030] See Figure 3 This is a flowchart illustrating the prediction of the Bagsinger effect in thin metal sheets according to the present invention. The method of the present invention includes the following steps:
[0031] 1) Construct a finite element simulation model of the three-point bending deformation process under a specific pre-tension amount, and design a pre-tension three-point bending test with the same parameters to verify the reliability of the finite element model.
[0032] 2) The load-displacement curves of various hypothetical thin plate materials during three-point bending were obtained by finite element model simulation, and the Bauschinger stress parameters of each thin plate material were calculated by applying the Bauschinger effect theory under linear kinematic hardening constitutive model. A sample set containing the mapping relationship between load-displacement curves and Bauschinger stress parameters was constructed.
[0033] 3) Construct a prediction model of the Bauschinger effect based on a BP neural network and train it using a sample set.
[0034] 4) Obtain the load-displacement curve of the metal sheet to be tested by performing a pre-stretch three-point bending test with the same parameters as in step 1), input the trained BP network Bauschinger effect prediction model, and predict its Bauschinger effect.
[0035] The pre-tension amount mentioned in step 1) needs to be determined by performing a tensile test on the thin plate. In the tensile test...
[0036] The stress applied to the thin plate should be less than its ultimate tensile strength. This is to ensure that the Bauschinger effect is clearly characterized in the load-displacement curve obtained from bending, and to avoid fracture or back cracks in the specimen during pre-tensioning and three-point bending continuous deformation.
[0037] Step 2) The evaluation index for the Bauschinger effect is the Bauschinger stress parameter. Its value is calculated as follows:
[0038] (1)
[0039] Where: σ t —Stress at the positive unloading point, σ r —Reverse yield stress.
[0040] Where σ t σ r Applying the Bauschinger effect theory in linear follower-reinforced constitutive model (with appendix) Figure 1 ), calculated using equations (2) and (3):
[0041] (2)
[0042] In the formula: E, E P These are the elastic modulus and the plastic modulus, respectively. σ s For the positive yield stress, ε s For positive yield strain, ε t The strain is the strain at the positive unloading point.
[0043] (3)
[0044] The input to the Bauschinger effect prediction model based on the BP neural network described in step 3) is a column matrix composed of the horizontal and vertical coordinates of 30 randomly selected points on the load-displacement curve, and the output is the Bauschinger stress parameters.
[0045] The pre-tension three-point bending test parameters described in step 4 are completely consistent with the pre-tension three-point bending finite element simulation model in step 1).
[0046] This invention specifically uses a thin sheet of X80 pipeline steel to be tested, with dimensions of 156mm × 35mm × 6mm, to predict the Bauschinger effect of this sheet, including the following steps:
[0047] Tensile tests were conducted on the X80 pipeline steel sheet used, and the elongation after fracture was found to be 20.35%. Therefore, a finite element model of the pre-tensioned three-point bending of the sheet material with a pre-tension of 18% and dimensions of 156mm×35mm×6mm was established. A pre-tensioned three-point bending test with the same specimen, parameters, and material was designed to verify the accuracy of the load-displacement curve obtained by the model.
[0048] The load-displacement curves of 125 hypothetical thin plate materials were obtained by simulating using a verified finite element model. Based on the Bauschinger effect theory under linear kinematic hardening constitutive model, the elastic modulus, yield strength, and plastic modulus of these 125 hypothetical materials were used to calculate the Bauschinger stress parameter value for each material. A sample set containing the mapping relationship between the "load-displacement curve" and the "Bauschinger stress parameter value" for each hypothetical material was constructed.
[0049] A 60×12×1 BP network model was constructed, and 30 coordinate points that could describe the characteristics of the curve were randomly selected on the load-displacement curve of each hypothetical material in the sample set. The column matrix composed of its horizontal and vertical coordinates was used as the input of the BP network model, and the Bauschinger stress parameter value of the hypothetical material was used as the output of the BP network model. The BP network model was then trained.
[0050] The load-displacement curve of X80 pipeline steel sheet material obtained from the pre-tension three-point bending test is input into the trained BP network model to obtain the predicted value of the Bagsinger stress parameter of the material.
[0051] By using uniaxial tensile and compressive tests, the experimental values of Bauschinger stress parameters of the X80 pipeline steel sheet were obtained. By comparison, the prediction method mentioned in this invention has a prediction error of 3%, which is highly accurate and efficient.
[0052] In summary, the Bauschinger effect prediction method for thin metal sheets proposed in this invention requires no modification or redesign of the experimental setup, and is simple to operate, highly efficient, and highly accurate. Thin sheet processing plants generally lack dedicated equipment for measuring the Bauschinger effect in thin sheets. However, using the method proposed in this invention, the Bauschinger effect of thin sheets can be measured quickly and accurately, the springback amount during processing can be accurately calculated, process optimization can be performed, and high-quality thin sheet parts can be obtained.
[0053] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.
Claims
1. A method for predicting the Bagsinger effect in thin metal sheets using a BP neural network, characterized in that, include: 1) In the tensile test, a stress lower than the material strength limit is applied to the thin plate to determine the pre-tension amount. A finite element simulation model of the three-point bending deformation process under the pre-tension amount is constructed, and a three-point bending test with the same pre-tension parameters is designed to verify the reliability of the finite element model. 2) The load-displacement curves of various hypothetical thin plate materials during three-point bending were obtained by finite element model simulation, and the Bauschinger stress parameters of each thin plate material were calculated by applying the Bauschinger effect theory under linear kinematic hardening constitutive model. A sample set containing the mapping relationship between load-displacement curves and Bauschinger stress parameters was constructed. 3) Construct a Bauschinger effect prediction model based on a BP neural network and train it using the sample set. The input of the constructed Bauschinger effect prediction model based on a BP neural network is a column matrix composed of the horizontal and vertical coordinates of multiple points randomly selected on the load-displacement curve, and the output is the Bauschinger stress parameter. 4) Obtain the load-displacement curve of the metal sheet to be tested through a pre-stretched three-point bending test, input it into the trained BP network Bauschinger effect prediction model, and predict its Bauschinger effect.
2. The method for predicting the Bagsinger effect in thin metal plates using a BP neural network as described in claim 1, characterized in that: In step 2), the evaluation index of the Bauschinger effect is the Bauschinger stress parameter, the value of which is calculated by the following method: Where: σ t —Stress at the positive unloading point, σ r —Reverse yield stress, where σ t σ r Using the Bauschinger effect theory in linear kinematic reinforced constitutive modeling, the following method is used to calculate: In the formula: E, E P These are the elastic modulus and the plastic modulus, respectively, σ s For the positive yield stress, ε s For positive yield strain, ε t The strain is the strain at the positive unloading point.
3. The method for predicting the Bagsinger effect in thin metal plates using a BP neural network as described in claim 1, characterized in that: In step 3), the input of the constructed Bauschinger effect prediction model based on BP neural network is a column matrix composed of the horizontal and vertical coordinates of 30 points randomly selected on the load-displacement curve, and the output is the Bauschinger stress parameter.
4. The method for predicting the Bagsinger effect in thin metal plates using a BP neural network as described in claim 1, characterized in that: In step 4), the pre-tension three-point bending test parameters are completely consistent with the pre-tension three-point bending finite element simulation model in step 1).