BP神经网络在柴油机涡轮增压系统故障诊断中的应用*

将BP神经网络应用于柴油机涡轮增压系统的故障诊断。简要介绍了BP神经网络的基本原理和算法，采用BP算法对网络进行训练、学习，可以识别柴油机滤清器堵塞、空冷器空气测流阻增大、气缸进排气堵塞及废气涡轮流阻增大四种故障，诊断结果和实际结果非常吻合，诊断结果可靠性强，并可以准确预计故障等级。

Application of BP Neural Networks to Fault Diagnosis of Turbocharging System of Diesel Engine

This paper applies BP neural networks to the fault diagnosis of turbine supercharging system of diesel engine. The keystone and algorithm of BP neural networks are simply introduced. Then, the BP algorithm is applied to train the networks. The trained networks can identify four kinds of fault diagnosises, including the jam in air filter of diesel engine, air resistance increase in air freezing refrigeration, the jam of intake and exhaust system, air resistance increase in turbo. Furthermore, the diagnosing results are consistent with the factual results. So the dignosing results are very reliable, and we can estimate the grade of the faults exactly.