改进BP算法用于液体发动机涡轮泵故障诊断

Fault Diagnosis for Turbo Pump of LRE by an Improved BP Algorithm

涡轮泵是液体火箭发动机系统的核心设备,工作环境十分恶劣,极易出现故障,所以对涡轮泵进行准确、快速的故障诊断显得尤为重要。因此提出了利用BP神经网络对其进行故障诊断。而常规BP算法存在收敛速度慢、易陷入局部最小等缺点,运用一种改进算法—自适应率BP算法,通过对液体火箭发动机涡轮泵的常见故障进行诊断。诊断结果表明,该改进算法具有收敛速度快,学习记忆稳定,诊断准确等特点。

Turbo pump is an important equipment of liquid rocket engine with a very badly working environment,so it is easy to be caused fault.So BP algorithm is used to diagnose fault.In order to improve the training velocity and solve the problem of falling in local least point for traditional BP algorithm,an improved algorithm-self adaptve BP algorithm is presented to diagnose the fault of the turbo pump of liquid rocket engine.The result shows that the method has faster convergence velocity,more stable study and memory,and more accurate diagnosis.