| With the improvement of living standards,the use of refrigeration units in daily work and life is more and more widely.It is directly related to the normal operation of the quality of life and the production activities of the refrigeration equipment.People have put forward higher request to the fault diagnosis method and effect of the refrigeration unit.It has important theoretical value and practical significance to study the fault diagnosis of the refrigeration unit.Neural network algorithm is the main method of fault diagnosis of refrigeration system.But the neural network has its own defects,such as the contradiction between the performance of the neural network and the structure of the neural network,the contradiction between the speed of the network learning and the convergence of the neural network algorithm.It also restricts the diagnostic performance of the play.This paper studies the improvement method of the neural network algorithm and its application in the fault diagnosis of the refrigeration unit.The main work is as follows:(1)the operating mechanism of refrigeration cycle of refrigeration unit is studied in depth,and the main fault types of the unit are analyzed in detail,and then the fault diagnosis process of the refrigeration unit is optimized in this paper.7.(2)the structure and basic principle of BP neural network are described firstly,and according to the shortcoming of BP neural network algorithm in the fault diagnosis of refrigeration unit,the adaptive neural network algorithm is proposed.(3)based on the extension of matter-element,correlation function,extension set theory and so on,the structure of extension neural network and the steps of fault diagnosis based on this algorithm are designed.Based on the extraction of training samples,the hidden layer structure,the learning algorithm and the expected error of the system are determined.Finally,the extension neural network model for the fault diagnosis of the refrigeration unit is built.(4)MATLAB software is used to carry out numerical simulation,and the effectiveness of BP,momentum adaptive and extension neural network model is verified.The results show that compared with the BP and the momentum adaptive neural network model,the extension neural network can more accurately judge thefault condition of the refrigeration unit. |
