基于BP神经网络的动车组客室空调故障识别与预警研究

摘要

动车组客室空调良好的制冷效果是旅客舒适度和动车组稳定有序运行的重要保证,但客室空调故障的发生具有突发性和隐蔽性特点,给空调系统的日常检修维护带来极大困难.采用BP神经网络算法建立客室室温预测模型,并运用Matlab编程计算实现客室室温理论预测.根据预测模型在CRH380B(L)型动车组客室空调系统中的实际验证情况,制定客室空调故障识别与预警的阈值、规则和等级,为后期客室空调系统故障自动识别与预警系统的开发奠定基础,完善动车组客室空调故障识别和预警机制,对动车组客室空调故障的在线实时识别与潜在故障的预警具有重要意义.%Good refrigeration effect of emu passenger compartment air conditioning is very important to ensure passenger comfort,emu stability and orderly operation,but the faults of the air conditioning system characterized by sudden occurrence and concealment challenge the daily maintenance.In this paper,BP neural network algorithm is adopted to establish temperature prediction model of passenger compartment,and Matlab programming and calculation are conducted to fulfill theoretical prediction of compartment temperature.Based on the analysis and verification of prediction model in CRH380B (L) EMU compartment air conditioning system,the thresholds,rules and grades for passenger compartment air conditioning fault identification and early warning are formulated,which lays the foundation for the development of fault automatic identification and early-warning system and for the improvement of fault identification and early warning mechanism,and is of vital importance to the potential failure on-line real-time identification and warning of emu passenger compartment air conditioning.