核电厂两种实时故障诊断系统的对比

Contrastive Research on Two Kinds of Real-Time Fault Diagnosis Systems of Nuclear Power Plants

分别利用模糊神经网络技术和数据融合技术,采用VB6.0编程语言开发了核电厂实时故障诊断系统,并对诊断系统中所应用的智能诊断算法进行了详细的阐述。为比较直观地对比两个诊断系统,利用数据通讯程序接口使该诊断系统与仿真机进行实时数据交互,并在仿真机上设置了4种故障对两个诊断系统进行在线测试。测试结果表明,应用模糊神经网络技术和数据融合技术均能对故障进行识别,但都存在各自的优点和不足。离线分析表明,针对不同的故障类型,当特征参量较少时,采用模糊神经网络诊断技术较好;而特征参量较多时,最好采用数据融合诊断技术。

In order to guarantee the safe operation of nuclear power plants, the real-time fault diagnosis systems are developed using neural network technology and data fusion technology and adopting VB6.0 programming languages, and then the intelligence diagnosis arithmetic is expatiated in detail. The fault diagnosis systems interchange the data with the simulator timely utilizing communication procedure interface of the data, and four faults are inserted on the simulator to test the two diagnosis systems on line. The test result indicates that the fuzzy neural network technology and the data fusion technology could carry out the recognition of the faults, but each has its merit and the insufficiency respectively. The off-line analysis shows that, for different fault types, when there is few characteristic parameters, the fuzzy neural network diagnosis technology is better; when there are many characteristic parameter, the data fusion diagnosis technology is better.