Digital Technology In The Simulation Of Air-conditioning And Fault Diagnosis

In this paper,the mathematical models of every part of air handling unit(AHU) are established in detail,a simulation software is programmed by matlab7.0.Later, modulation characteristic of system is analyzed.At the same time,reliability of the system is demanded,the faults detection and diagnosis system has received greater attention.In this paper,the application of BP(Back-Propagation) algorithm in fault diagnosis of the air-conditioning system is researched.The system is made up of two major parts:simulation and fault detection and diagnosis(FDD).The program is developed utilizing matlab7.0 in windowsXP operating system.The artificial neural network(ANN) is using the 3-layer BP neural network for FDD.The Simulation part consists of eight parts according to the components of AHU. The FDD part simulates 5 common faults of AHU.The input X to the network is designed to satisfy this condition:X∈[0,1].The hidden layer has 7 neurons,and the output layer has 5 neurons:each one corresponds to one kind of common fault.The BP network can be trained either in a non-noisy way or in a noisy way.In both cases it was trained using random input.The results show the network can predict the 5 common system faults correctly.