船用核动力装置神经网络故障诊断技术研究

摘要

针对船用核动力装置故障模式多、故障诊断难度大的问题,开展了基于遗传神经网络的核动力装置故障诊断技术研究.将遗传算法引入到BP神经网络训练中,克服了神经网络训练难度大、精度低的不足;构建了由单一故障模式诊断与复合故障模式诊断相协调的综合推理系统,克服了单一神经网络推理精度低、难以实现对大量故障模式精准识别的问题.通过仿真分析可以看出,所设计的综合故障诊断系统,可以实现对故障的高效率诊断,在提高诊断可靠性的同时,大大降低了误诊率的发生.%In order to conquer the problem of large amount fault models and fault diagnosis difficulties on ship nuclear power plant,nuclear power plant fault diagnosis technology based on genetic neural network is studied in this article.The genetic algorithm was used in BP neural network training to conquer the training difficulty and low precision.The coordinating comprehensive reasoning system based on single fault model and multiple fault models was built to conquer the low reasoning precision of single fault model neural network and the difficulties of large amount fault models diagnosis.It can be seen from the result of simulation analysis that comprehensive reasoning system built in this article can diagnose fault models efficiently with higher diagnosis reliability and lower diagnosis faults.