On Fault Prediction Method For A320 Aircraft Air Conditioning System

Air-conditioning system fault is relatively high frequency in the subsystems of A320 aircraft. Fault prediction is of great importance to maintain the aircraft air-conditioning system. Based on this, the fault prediction method of A320 aircraft air-conditioning system is studied.Firstly, A320 aircraft air-conditioning cooling system is studied in-depth. Taking aircraft maintenance simulator of laboratory for the experimental platform, three faults of air conditioning components(PACK) are simulated. The experimental data is statistically analyzed. The production rules between three air-conditioning PACK faults and two characteristic parameters are obtained.Secondly, grey theory is applied to air-conditioning PACK fault prediction. In order to improve the accuracy of traditional GM(1,1) prediction model, an improved GM(1,1)prediction model is established. The improved GM(1,1) prediction model introduces an operator of three-points to dispose of the original fluctuant sequences and corrects the background sequences using a new method. The predicted sequence of the two characteristic parameters of A320 air-conditioning PACK shows that the improved GM(1,1) prediction model satisfies the accuracy requirement and it is validated to judge whether there will be a fault.Finally, in this paper, grey prediction accuracy is not high on the long-term prediction. So the BP neural network is applied. Through repeated training, optimal learning rate and number of neurons are obtained. And the accuracy in long-running of air-conditioning forecast is high.