| Based on fuzzy set theory and BP neural network technology, this paper puts its focuses on diagnosis theory and method based on their combined mechanism, and an online intelligent diagnose model of FCM-BP is carried out and applied to fault diagnose on turbine blower's vibration. The field application shows the model has fairly high practical value.To begin with, this paper deals with the importance and underground of the research subject, giving the research approach and implementing principles. Secondly, by analyzing detectors of turbine blower units as well as combing with feather of the field and requirement of the system, an implementation scheme of state monitoring and fault diagnose is established. Thirdly, an online intelligent diagnose model of FCM-BP is carried out. Finally, software design and application about the system is introduced in details.From the position of intelligent diagnosis technology, a thorough research made in the paper is described as follows: (1) The theory and method of Fuzzy C-Means clustering and BP neural network are thoroughly studied. According to the common faults of units, a diagnose model of FCM-BP based on their "simple layer" combination mechanism is put out and applied successfully to fault diagnose on vibrations of turbine blower. (2) On the basis of front end data acquisition, parallel dignosis network is replaced by generic dignosis network in the system, this not only lose no diagnosis accuracy, but also increase diagnosis speed, this is no doubt an improvement to traditional parallel dignosis network. (3) Fuzzy cluster analysis is used as preconditioning to select the sample set which includes characteristics of all members, which can solve the problem of study speed, and the method of analyzing the result of training is developed. This process can be a necessary supplement of the application of BP neural network. (4) The knowledge express based on figure is adopted and the man-machine interface is friendly. It is convenient modifying expert knowledge. The way can incrementally collect and save expert knowledge without any model. It is useful to the situation without mathematic model and it can help user understand the whole course of diagnosis.Application of the system solve the problem of fault diagnosis with long time, unclear locations and unknown causes, making fault diagnosis of turbine blower unit vibration scientific and standardized, improving diagnosis accuracy and speed. Since it was devoted to pre-running, the system had been becoming more and more useful through continuous amelioration and consummation, and the function of system was validated through application. It satisfied the customers and was prided by them.
|
PDF Full Text Request