基于改进的神经网络对航空发动机故障率预测研究

以某俄式发动机为研究对象，根据该发动机故障分布，使用优化后的BP神经网络对该故障率模拟。针对BP神经网络可能陷入局部极小值点的问题，在激励函数中加入模糊参数。与原网络相比，在不同学习速率的条件下，优化后的BP神经网络预测结果与实际故障率更为拟合，预测结果更为准确。该模型对于该型号发动机的故障预测具有一定参考意义。

The Study on Prediction of Aeroengine Failure Rate Based on Improved Neural Network

In this paper, a Russian engine is taken as the research object. According to the fault distribution of the engine, the optimized BP neural network is used to simulate the failure rate. Aiming at the problem that BP neural network may fall into local minimum point, fuzzy parameters are added into the excitation function. Compared with the original network, under the condition of different learning rates, the optimized BP neural network prediction results are more fitting with the actual failure rate, and the prediction results are more accurate. This model has some reference significance for the fault prediction of this type of engine.