|
| 摘要: |
| 针对当前专家系统知识获取瓶颈的难题,提出了基于神经网络与遗传算法的汽轮机组数据挖掘方法。将汽轮机组历史故障数据首先进行模糊化及离散化处理,接着构造一个多层的前向神经网络,然后通过教师示教的方式训练构造好的神经网络,最后进行基于遗传算法的神经网络优化。以神经网络为知识本体,提出了汽轮机组故障诊断分类规则的挖掘算法,其实现过程有4个步骤:计算效果度量矩阵;提取规则;计算规则权重;基于遗传算法的规则修剪。实现了基于神经网络与遗传算法的汽轮机组数据挖掘和故障诊断仿真系统,其诊断正确率达到了84%。 |
| 关键词: 神经网络,遗传算法,数据挖掘,故障诊断,规则修剪,汽轮机 |
| DOI： |
| 分类号:TM311;TP206.3;TP183 |
| 基金项目:国家863计划项目(2004AA1Z2450)，国家自然科学基金项目(70572090)，华北电力大学博士学位教师科研基金资助项目(H0585)~~ |
|
| Data mining based on neural network and genetic algorithm for steam turbine unit |
|
GUO Qinglin zu Xiangrong
|
| Abstract: |
| As the knowledge acquisition is quite difficult in current expert systems,an approach of data mining based on neural network and genetic algorithm is presented for steam turbine unit. The historical fault data is fuzzed up and discretized first,and a multi - level BP neural network is then constructed,which is trained in teaching mode and optimized by genetic algorithm. Based on the neural network,the data mining algorithm for classified diagnosis rules of steam turbine faults is brought forward,which has four steps:computing the measurement matrix of effect;extracting rules; computing the weights of rules;shearing the rules by genetic algorithm. A simulative system is implemented and its diagnosis precision is 84 %. |
| Key words: neural network,genetic algorithm,data mining,fault diagnosis,rule shearing,steam turbine |
