针对装甲车辆柴油机喷油器故障诊断不能满足实时在线监测的问题，提出一种基于多层双向长短时记忆 网络(bidirectional long short term memory，Bi-LSTM)的装甲车辆柴油机喷油器故障诊断方法。对柴油机喷油器故障 进行模拟实验，利用多层双向长短时记忆网络具备较长距离的时序分析能力的优势，分别将压力波特征值和压力波 时序信号作为输入进行故障模式识别验证。结果表明：该方法具有较高的识别精度和较快的分类速度，能够满足实 时在线监测的要求。
Aiming at the problem that the fault diagnosis of armored vehicle diesel engine injector can not meet the real-time on-line monitoring, a fault diagnosis method of armored vehicle diesel injector based on multi-layers bidirectional long short term memory (Bi-LSTM) network is proposed. The simulation experiment of diesel injector fault is carried out. Taking advantage of the long-distance time sequence analysis ability of multi-layer bidirectional long-short term memory network, the fault pattern recognition verification is carried out by taking the pressure wave feature value and pressure wave time sequence signal as input respectively. The results show that the method has higher recognition accuracy and faster classification speed, and can meet the requirements of real-time online monitoring.