Optimization design method for sheet metal liquid-filled forming loading path
A technology of liquid filling forming and loading path, which is applied in neural learning methods, design optimization/simulation, biological neural network models, etc. It can solve problems such as suboptimal solutions, poor scientificity, and large randomness, and reduce production costs. , Guaranteed accuracy and improved optimization efficiency
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Publication Date
- 2021-03-12
Smart Images

Figure 1 
Figure 2 
Figure 3
Abstract
Description
technical field
[0001] The invention belongs to the technical field of sheet material liquid-filled forming, and in particular relates to a method for optimizing the design of a loading path of sheet material liquid-filled forming. Background technique
[0002] Liquid-filled sheet metal forming refers to the use of liquid oil or water as the force transmission medium to replace the traditional rigid die or punch, so that the blank is formed by fitting the punch or die under the pressure of the force transmission medium. Flexible medium-assisted forming technology, the parts produced by this technology have good overall structure and small springback, which can simplify the production of parts with complex shapes, have a high degree of flexibility, and greatly reduce the number of mold sets and costs. It has broad application prospects in high-tech industries such as industry and nuclear energy application technology.
[0003] In the process of liquid-filled forming, the pre...
Examples
Embodiment 1
[0049] A method for optimizing the design of the loading path of sheet metal liquid-filled forming in this embodiment, such as figure 1 As shown, the genetic method is used to iteratively optimize the loading path of liquid-filled forming, and the neural network prediction model is called to predict the result of liquid-filled forming in the iterative optimization process to obtain the optimal liquid-filled loading path, including the following steps:
[0050] Step 1. Determine the input parameters and output parameters of the neural network, use the coordinate variables of several feature points on the liquid-filled loading path as the input parameters of the neural network, and perform several formings of the liquid-filled forming part with the current liquid-filled forming path The quality evaluation parameters are used as the output parameters of the neural network;
[0051] The coordinate variables of the feature points on the liquid-filled loading path can intuitively re...
Embodiment 2
[0056] This embodiment is further optimized on the basis of Embodiment 1. The establishment of the liquid-filled forming prediction model in the step 3 includes the following steps:
[0057] Step 3.1, use the L-M learning method to train the BP neural network with a three-layer structure, use the trial and error method to determine the optimal number of hidden nodes for training, use the empirical formula to determine the minimum number of hidden nodes for training, and use the least hidden nodes Train the BP neural network with three-layer structure. The BP neural network of three-layer structure comprises input layer, hidden layer, output layer, and the transfer function between described input layer and hidden layer is tanh function " tansig ", and the transfer function between described hidden layer and output layer The function is a linear function "purelin", and the error is selected as 1.0×10 -7 .
[0058] Step 3.2, based on the same training sample, gradually increas...
Embodiment 3
[0066] This embodiment is further optimized on the basis of above-mentioned embodiment 1 or 2, and described step 4 comprises the following steps:
[0067] Step 4.1. Predict the forming quality evaluation parameters under different liquid-filled loading paths through the liquid-filled forming prediction model;
[0068] Step 4.2: Use the heuristic crossover method and adaptive feasible mutation method in Matlab as the genetic method, and perform iterative calculations according to the forming quality evaluation parameters corresponding to different liquid-filled loading paths obtained in step 4.1, so as to obtain the adaptation of the liquid-filled loading path degree function;
[0069] Step 4.3: Obtain the optimal liquid-filled loading path according to the fitness function and output the corresponding forming quality evaluation parameters under the optimal liquid-filled loading path.
[0070] Using the neural network's approximation function to the nonlinear model to predict...