Fault diagnosis of shield machine based on SOM-BP neural network fusion

In view of the potential operation harms caused by manual checking construction faults of shield, a method to diagnose construction faults of shield is proposed based on neural network(NN) information fusion. The parameters of shield excavation are as the input of self-organizing feature maps(SOM) to perform partial fusion. Then the output of SOM is as the input of back propagation(BP) network to complete final fusion. The construction faults of shield can be diagnosed according the results of final fusion. The analysis of an illustrating example indicates that the proposed method is more effective and feasible. The diagnosis results can be a beneficial guidance for online-diagnosing the construction faults of shield.