Application of Fault Diagnosis Based on Neural Network and Least Square
Daoping Wang · Modern Electronics Technique · 2010
The application of BP neural network and nonlinear least squares in the fault diagnosis are introduced.The technical characteristics of both mentioned above are fused.The fault diagnosis technology based on parameter estimation for analyzing the time-sequential redundant signal are proposed.The generation of fault diagnosis neural network specimens and training of the neural network are introduced in detail.The final estimation of fault parameters is achieved with the least squares depending on system model by taking the output of neural network as the initial estimation which does not depend on the system model.The simulation verifies the feasibility of this fault diagnosis by the checkout with the output responses at different input excitation.