GA-BP神经网络的钛合金微弧氧化膜层厚度预测模型的建立

Establishment of GA-BP neural network model for prediction of the thickness of micro-arc oxidation coating on titanium alloy

利用正交试验法获得的TC4钛合金微弧氧化实验数据建立了基于4-11-1(即4个输入神经元,11个隐含层节点,1个输出神经元)结构的BP神经网络预测膜层厚度的模型,并引入遗传算法(GA)对其权值和阈值进行优化。以微弧氧化工艺参数中的电流密度、脉冲频率、占空比和氧化时间作为网络的输入向量,氧化膜层厚度作为网络的输出向量,对比和分析了BP与GA-BP模型的预测结果。与BP网络模型相比,GA-BP网络模型稳定性能较好,并能高精度预测膜层的厚度,GA-BP网络模型预测值的平均误差为0.015,最大误差仅为0.036,而BP模型预测结果的平均误差为0.064,最大误差为0.099。

A model based on BP（back propagation） neural network with 4-11-1（i.e. 4 input neurons, 11 hidden layer nodes, and 1 output neuron） structure for prediction of the thickness of micro-arc oxidation coating on TC4 titanium alloy was established using the data obtained by orthogonal test. The weights and thresholds were optimized by genetic algorithm（GA）. The prediction results of the BP and GA-BP models were analyzed and compared using current density, frequency, duty cycle, and oxidation time as input vectors and the thickness of oxidation coating as output vector. Compared to BP model, the GA-BP model possesses better stability and predicts the coating thickness with higher precision. The average and maximum prediction errors are 0.015 and 0.036 respectively for GA-BP model, and 0.064 and 0.099 respectively for BP model.