【摘要】 采用在遗传规划中使用概率模型的新方法来解决一系列故障诊断问题。故障诊断可被看为是一个多级分类问题。遗传规划在解决复杂问题上有很大的优势,而这种优势在故障诊断中仍然显著。而且,使用概率模型作为适应函数能提高诊断的精确性,最后用这种方法解决机电设备的故障诊断。结果显示,使用基于概率模型的遗传规划解决机电设备的故障诊断比人工神经网络优越。更多还原

【Abstract】 A new method is presented to solve a seriesof faultdetection problems using probabilisticmodel(PM) in genetic programming(GP).Fault detection is seen as a problem of multi-class classification.GP methods used to solve problems have a great advantage in their powertorepresent solutions to complex problems,and this advantage remainsin the domain of fault detection.Moreover,diagnosis accuracy is improved by using PM.In the end,the fault detection of electro-mechanical device is solved.The results show that the method using GP with PM to solve fault detection of electro-mechanical device is superior to the artificial neural network(ANN).更多还原