An Expert System for Diagnosis of Dynamometer Cards based on the Integration of Rough Sets and Neural Network
In order to increase the efficiency of diagnosis for the dynamometer cards of the pumping unit, an expert system is introduced. The structure of artificial neural network (ANN) in the expert system can be simplified by using the rough sets. Firstly, the training data of the dynamometer cards are inputted into rough sets and they are classified. Next, the classified data are fed into the feedback ANN and ANN is trained according to the data. Finally, a real dynamometer card is inputted into the expert system to get the diagnostic results. Examples show that the expert system has a higher training speed than the conventional ANN. When the training target is 0.001, it converges after training 3165 times. The correct percentage of diagnosis is 91％.
Expert system Rough sets Artificial neural network Dynamometer card Diagnosis
Hangxin Wei Wei Wu Liguang Zhi
School of Mechanical Engineering, Xian ShiYou University, Xian City, China
国际会议
武汉
英文
835-839
2010-06-06（万方平台首次上网日期，不代表论文的发表时间）