针对铝合金表面处理工艺复杂、加工设备工况动态变化、参数设置及控制完全依靠人工经验等问题进行了优化研究。研发工艺过程管控优化专家系统，以实际生产数据作为神经网络的训练样本，建立输入为工艺参数、输出为产品合格率预测值的铝合金表面处理工艺参数神经网络智能识别模型。并以该神经网络模型代替实际生产系统，在工艺参数取值范围之内，运用正交设计优化工艺参数。结果表明，经过训练的神经网络模型能较好地映射工艺参数与优化指标间的复杂非线性关系，结合正交设计，能准确进行任意条件下的参数优选和结果预测。

Aiming to the problems such as the complication of aluminum alloy surface treatment process, the dynamic changes of conditions of processing equipment and the complete dependance on the artificial experience of the parameter setting and control optimization was studied. Treatment process control optimization expert system was investigated and developed based on actual production data as the training samples of a neural network model. The aluminum alloy surface treatment process parameters neural network intelligent recognition model was build, in which the input was process parameters and the output was the predicted value of qualified product rate. And then the neural network model replaced the actual production system to optimize the process parameters by orthogonal design within the range of process parameters. The results indicated that the trained neural network model could well map the complex non-linear relationship between the process parameters and the optimization indicators. Combined with orthogonal design, the model can accurately optimize process parameters and forecast qualified product under arbitrary conditions.