An Application of Integrated Neural Networks to Fault Diagnosis
Xiao Gan · Mechanical Science and Technology · 2001
Based on signal type and different feature vectors, we put forward the idea of integrated neural networks from the view of data fusion. Integrated neural networks are composed of three level serial networks, and the second level network is composed of a single parallel network. Data distribution network carry out the task of data location. Diagnosis networks carry out the task of networks training and diagnosing. Decision making and fusion network receives diagnosing conclusions and fuses them. Application show that integrated neural networks are quite efficient in fault diagnosis.