用BP神经网络预测了铝合金大气腐蚀,研究了网络的训练精度和预测精度的关系,建立7-5-1的模型结构,模型相关系数为0.8821,预测结果比较理想.利用单一因素敏感性分析,计算了合金元素和环境因素对于铝合金大气腐蚀速率的影响.
Atmospheric corrosion of aluminium alloy was predicted using BP neural network, and relation between training error and predicting error was researched. Artificial neural network(ANN) model is composed of seven input nodes, five hidden layer nodes and one output node. Correlation coefficient for model is 0. 8821, ANN prediction result is close to practical data. By sensitivity analysis method, metallurgical factor and environmental factor influence for atmospheric corrosion rate of aluminium alloy was researched.