摘要
针对如何通过聚合釜运行时的各项历史数据,进行聚合釜的故障诊断分析和预测,本文提出了一种方法,即基于粗糙集RS-BP神经网络相结合建立数据模型,并利用遗传算法进行模型结构优化。一方面利用遗传算法对数据的粗糙集进行属性约简,保留了必要属性,约去不必要的冗余数据,减少诊断模型的输入维数,降低过拟合现象。同时利用遗传算法对BP神经网络的初始权值和阀值进行结构优化,提高其预测精度,并将其应用于聚合釜的故障预测和诊断中,仿真实验验证了该方法的有效性。
How to analysis fault diagnosis and predict of polymerization kettle through the historical data of polymerization kettle running. In this paper, a method is proposed, which is based on rough set RS-BP neural network to establish the data model, and genetic algorithm is used to optimize the model structure. On the one hand, it is using genetic algorithm for attribute reduction in rough set data to retain the necessary attributes and reduce unnecessary data. Thus the input dimension of diagnostic model is less than before. At the same time it is using the genetic algorithm to optimize the BP neural network's initial weights and threshold and thus the prediction of accuracy becomes improving. Finally using this method applies in fault diagnosis and prediction of polymerizer, simulation results verify the validity of the method.
出处
《计算机与应用化学》
CAS
2017年第8期621-624,共4页
Computers and Applied Chemistry
基金
河北省高等学校科学技术研究青年基金项目(2011139)
国家自然科学基金资助项目(F2012203111)
关键词
粗糙集
属性约简
遗传算法
BP神经网络
故障诊断
rough set
attribute reduction
genetic algorithm
BP neural network
fault diagnosis