基于神经网络的整流罩空调系统传感器故障诊断

Sensor Fault Diagnosis for Air Conditioning System of Satellite Fairing Based on Neural Networks

针对卫星整流罩空调系统中传感器故障问题,分析了其典型故障模式及影响因素,构建了传感器故障诊断系统模型,提出了利用空调监控系统采集的历史数据建立径向基神经网络估计器的传感器故障诊断方法。将估计器应用于空调系统表冷器控制单元,对传感器的偏置故障和精度下降故障进行了模拟故障仿真应用实验,实验结果表明了该故障诊断方法的有效性和正确性。

Aiming at sensor faults in an air conditioning system of satellite fairing, this paper analyzes the typical fault modes and their influence factors, establishes diagnosis system model for the sensor faults, and introduces a data-driven diagnosis method based on radial basis function （RBF） neural network estimators to detect sensor faults. In accordance with the proposed method, the historical data acquired by monitoring system are selected to train the neural network to establish RBF neural network estimators. The simulation results for bias fault and precision degradation fault in surface cooler sensors verify the reasonableness and valida- tion of the proposed method.