论文部分内容阅读
本文提出并建立了基于多神经网络多参数综合的旋转机械故障诊断系统。在多层多输出前向神经网络的算法基础上,对多个征兆域分别建立相应的诊断网络,然后利用加权方法进行综合评判,并且该诊断系统具有自学习、自适应能力,以便能够适应大型旋转机械,特别是汽轮发电机组等实际产生故障的振动原因的复杂性及诱发的振动征兆的多元性等特点,从而提高了故障诊断的可靠性和诊断精度。本系统对工程应用具有较高的实用价值。
This paper presents and establishes a multi-parameter synthesis based on multi-neural network of rotating machinery fault diagnosis system. Based on the algorithm of multi-layer and multi-output forward neural network, a corresponding diagnosis network is established for each of multiple symptom domains, and then a comprehensive evaluation is made by using the weighted method. The diagnosis system has the ability of self-learning and self-adaptation so as to be able to adapt to large-scale The complexity of the vibration causes of the rotating machinery, especially the steam turbine generator, and the diversity of induced vibration signs, thereby improving the reliability and diagnostic accuracy of the fault diagnosis. The system has high practical value for engineering application.