为解决武器装备全寿命周期费用高、经济可承受性差的难题,开展航空装备的故障预测技术研究。采用免疫算法改进隐含层激励函数得到免疫神经网络,用以进行装备特征参数的跟踪预测,结果表明免疫算法改进的神经网络可在故障前3小时实现预测,较BP网络性能有较大改善。 In order to solve the problem of high cost and low economic affordability of the whole life cycle of weaponry and equipment, the fault prediction technology research of aviation equipment is carried out. The immune algorithm is used to improve the hidden layer excitation function to get the immune neural network, which is used to track and predict the characteristic parameters of the equipment. The results show that the improved neural network with immune algorithm can predict three hours before the failure, which has a better performance than the BP network.