结合模糊神经网络(Fuzzy Neural Network，FNN)与DempsterFhafer(DS)证据理论，研制开发了船舶动力装置数字化监测与故障诊断系统。该系统由信号采集系统、软件控制系统和分析系统组成，主要对主推进机组、柴油发电机组、辅机进行监测。采用压力示功图法和瞬时转速法对主机运行状态进行实时动态监测，且并行构建热力性能参数和瞬时转速2个子模糊神经网络，再运用DS证据理论进行主机状态信息融合判断，最后得到故障诊断结果。试验表明，该系统的实时性、测量精度满足要求，而且运用FNN与DS

A digital monitoring and fault diagnosing system for ship power plant based on fuzzy neural network and dempstershafer(DS) evidence theory is developed. The system includes data acquisition part, software control part and analysis part, which can monitor the main propulsion device, diesel generator and auxiliaries. The main propulsion device can be realtime dynamicly monitored by acquiring cylinder pressure and instantaneous angular speed. Using thermodynamic parameters and instantaneous angular speed, two subfuzzy neural networks are parallel constructed. The fault of main propulsion device is judged by data fusing algorithm of DS evidence theory, and finally the result of fault diagnosis can be obtained. Test shows that the system’s realtime performance and measurement accuracy can meet requirements, and the fusion method of fuzzy neural network according to DS evidence theory has relatively high reliability.

重庆长江航道局科技计划项目(2008015)