The Fault Diagnosis Research Of Refrigeration System Of High-low Temperature Shock Chamber Based On The PCA And PNN

With the continuous development of science and technology and the improvement of industrial automation,The application range of the refrigeration equipment is more and more widely,especially as an important part of high and low temperature shock test chamber,In the detection of industrial products the reliability of the temperature and humidity environment is of great significance.Due to the complexity of refrigeration system structure and high failure frequency,failure is difficult to be detection and recognition,that can lead to increase of energy consumption and to negative effects such as equipment reliability,security and economy.Analyze the failure of cooling system,therefore,find a way to able to quickly identify and eliminate failure is a hot research topic in this field.This article uses the method of principal component analysis and probabilistic neural network to diagnosis common failure of refrigeration equipment of high and low temperature shock test chamber.Firstly,this paper introduces the composition and structure of the impact of high and low temperature test chamber and the working principle of refrigeration system,analysis system of common failures and causes,determine the relationship between the fault and symptoms,select could reflect the measurement of system running status parameters,to solve the current problems encountered in the process of refrigeration system fault diagnosis,design of the overall solution.Secondly,introduce the PCA,learning and Analysis the realization process of PCA,optimize the algorithm To determine the optimal principal component and to simulate the optimization algorithm.Thirdly,By studying the principle of BP neural network to establish a fault diagnosis system base on optimal parameters of the BP neural network to diagnosis three kinds of fault of refrigeration system,though experiments proved that the system can diagnose fault but exist problems of such as complex structure,network convergence is slow etc.So introduce the probabilistic neural network,andestablishing a fault diagnosis system base on optimal parameters of the PNN to diagnosis three kinds of fault of refrigeration system by using Matlab2012.Determining the PNN network not only has advantages of simple structure and fast training speed and good stability compared with the BP network diagnosis.Finally,through the transformation model for CERC-CJ-70 B high and low temperature shock test chamber to establish the experimental platform of the simulation of the refrigeration system to simulate 7 kinds of typical fault.And through the platform for getting training samples and testing samples.The original sample input fault diagnosis system respectively based on BP network?PNN network?PCA-PNN network.Comparing three kinds of network training process and diagnosis results to prove that the performance of PNN network is better than that of BP network,the performance of PCA-PNN is better than that of PNN network.