基于人工神经网络预测Fe-Si合金层性能及工艺优化研究

Artificial Neural Network-based Prediction of Properties and Process Optimization of Fe-Si Alloy Coating

采用Visual C# 2005计算机高级编程语言开发人工神经网络智能软件,应用此软件,建立了Fe-Si合金化工艺参数与合金层性能指标之间映射关系的模型。经过学习,可以验证或预测得出的饱和磁化强度和腐蚀电流密度,且与实际试验结果接近。建立合金化多指标综合性能评价模型,并建立合金化工艺参数对合金层各性能指标的影响,通过对合金层各性能指标权重的调整,根据需求确定出合金层的综合性能值,从而给出最优的工艺参数。从预测结果可以得出,合金化工艺为Si摩尔分数为0.23~0.35,温度在1000 ℃时,可得到最优性能,合金化样品的两项性能指标达到最优。

An artificial neural network intellective software was developed by Visual C#2005 computer high-level programming language.Application of the software can establish mapping relation model between Fe-Si coating technology parameters and its overall performance.The saturation flux density and corrosion electricity density closed to actual experimental results.A comprehensive evaluation model was established to evaluate two indicators which were saturation flux density and the cathode current efficiency.To get the optimal craft parameters,the weight value of two indicators is adjusted respectively and the Fe-Si coating comprehensive performance value is calculated by the model.From the predicted results,it can be concluded that the optimal properties can be obtained when the alloying process is Si mole fraction 0.23-0.35 and the temperature is 1 000℃,and the two performance indexes of the alloyed sample are optimal.