论文部分内容阅读
以离合器盖总成中的传力片作为研究对象,借助Deform-3D仿真软件模拟了传力片冲裁过程中的凸模磨损情况,依据正交仿真试验的数据以及BP人工神经网络对传力片冲裁凸模的磨损量进行仿真预测。将冲裁间隙、凹模刃口圆角半径与冲裁速度作为BP神经网络的输入层,将冲裁凸模的最大磨损深度作为BP神经网络的输出层,建立3-12-1的3层BP神经网络。BP神经网络通过训练之后,仿真预测的最大误差为1.14%。基于正交试验的仿真数据对BP神经网络的性能进行检验,BP神经网络的仿真预测值与数值模拟值之间的误差为2.09%,并利用冲压级进模对BP神经网络的仿真预测值进行试验验证,两者之间的相对误差为8.25%,验证了BP人工神经网络应用于传力片冲裁凸模磨损仿真预测的准确性。
Taking the force transmission film in the clutch cover assembly as the research object, Deform-3D simulation software was used to simulate the punch wear in the process of transferring the film. According to the data of the orthogonal simulation and BP artificial neural network, The punching punch wear amount is simulated and predicted. The blanking gap, die radius and punching speed as the BP neural network input layer, the punch punch maximum wear depth as the output layer of BP neural network, the establishment of 3-12-1 3 layers BP neural network. After BP neural network training, the maximum error of simulation prediction is 1.14%. The performance of BP neural network was tested based on the simulation data of orthogonal test. The error between the predicted value of BP neural network and the numerical simulation value was 2.09%. The simulation prediction value of BP neural network The experimental verification shows that the relative error between the two is 8.25%, which verifies the accuracy of BP artificial neural network in simulating the simulation of punch wear punch.