从信息融合的角度出发,以自组织特征映射网络SOM和误差反向传播网络BP为理论基础,提出了SOM-BP神经网络模型,引用某内燃机燃油系统故障及其相关参数作为训练样本建立诊断模型。通过仿真试验,验证了SOM-BP神经网络在机械故障诊断中的正确性和精确性。
From the view of information fusion, the SOM-BP network model based on the principal of self-organization network (SOM) and back pressure network (BP) was established. The faults and relative parameters of fuel injection system for internal combustion engine (ICE) were used for the training samples of the model. The correctness and accuracy of SOM-BP network for mechanical fault diagnosis were validated by the simulation test.