摘要
将遗传算法与经典概率论相接合，利用知识库，针对核电厂故障诊断的特点，提出了一种故障诊断方法。本方法将核电厂部件状态与遗传算法中的群体相联系，利用专家知识和核电厂信号对群体进行约束，使该群体在诊断过程中不断发展变化，从而找出适合条件的个体，达到故障诊断的目的。在北京核电厂模拟培训中心950MW全尺寸模拟机上的实验表明，该方法对诊断过程中出现的虚假信号、专家知识不完备等问题有相当的适应性。
Via using the knowledge base, combining Genetic Algorithm and classical probability and contraposing the characteristic of the fault diagnosis of NPP, We put forward a method of fault diagnosis. In the process of fault diagnosis, this method contact the state of NPP with the colony in GA and transform the colony to get the individual that adapts to the condition. On the 950MW full size simulator in Beijing NPP simulation training center, experimentation shows it has comparative adaptability to the imperfection of expert knowledge, illusive signal and other instance.
出处
《核动力工程》
EI
CAS
CSCD
北大核心
2000年第4期362-367,共6页
Nuclear Power Engineering
关键词
遗传算法
故障诊断
非线性反演
知识
核电厂
Genetic algorithm
Fault diagnosis
Non-linear inversion
Knowledge base
