GA-BP神经网络预测金属腐蚀速率  

Prediction of Metal Corrosion Rate Based on GA-BP Neural Network

作　　者：向乃瑞 闫海 王炜 刘闯[1] 陈思凡 XIANG Nai-rui,YAN Hai,WANG Wei, LIU Chuang, CHEN Si-fan(College of Electrical Engineering ＆ New Energy, China Three Gorges University, Yichang 443002, Chin)

机构地区：[1]三峡大学电气与新能源学院,湖北宜昌443002

出　　处：《电力学报》2018年第1期48-54,共7页Journal of Electric Power

摘　　要：以华南某地的土壤数据为基础对金属腐蚀进行研究。运用灰色关联度算法分析金属腐蚀速率与土壤中影响因素之间的关系,得到土壤的电阻率、氧化还原电位以及含水量是3个最主要的特征量。同时,针对目前常用的BP神经网络模型在计算时存在的一些局限性,应用遗传算法对其进行优化并利用改进后的模型对金属腐蚀速率进行预测。经计算,GA-BP神经网络相较于传统的BP网络,其模型的精度增加了3.13%,预测集的平均相对误差减少了3.22%。结果证明,建立的GA-BP神经网络预测模型是成功的。Using soil datas which came form a land in South China to study metal corrosion.The relationship between metal corrosion rate and influencing factors in soil was studied by using gray correlation degree.The soil resistivity,oxidation-reduction potential and water content were the three most important features.At the same time,according to the limitations of the commonly used BP neural network model in the calculation,Using the genetic algorithm to optimize and predict the corrosion rate of the metal.The GA-BP neural networks accuracy is improved 3.13%,and the average relative error of the prediction set is reduced by 3.22%.The results show that the GA-BP neural network prediction model is successful.

关 键 词：金属腐蚀 腐蚀预测 遗传算法 神经网络 

分 类 号：TG174.4[金属学及工艺—金属表面处理]