江汉大学学报（自然科学版）

金属电致塑性效应为一个多变量非线性灰色系统,分析各种外界因素对金属电致塑性效应、金属中微观组织的影响以及建立相应的本构方程是相当困难的.人工神经网络ANN具有自学习、自组织、自适应和非线性动态处理等特征,更容易通过仿真逼近实际的数学模型.以ANN对铝合金电致塑性效应进行仿真,经过多次的训练后逼近实际的非线性模型,然后,给定一组输入数据进行自适应预测输出,并根据实验所测的实际数据与预测结果进行对照和评判,从而获得了比较满意的仿真效果.

The Metal electroplastic effect is a multiple variable gray nonlinear system.It is very difficult to analyse the influence of various external factors on the electroplastic effect and the microstructure of the metal,also difficult to set up the relative equations.Artificial neural network(ANN) is characterized with self-studing,self-organizing,self-adapting and nonlinear dynamic treat-ing,and it is easier to approach the actual mathematical model through the simulation.Explores how to simulate the metal electroplastic effect through ANN and how approach the nonlinear model after many trainings.Then,also studies how to to give a set of input data for adaptive predicted output,compares and evaluates the actual measured experimental data with the predicted results,obtains a satisfactory simulation effect.