Sensor Fault Diagnosis based on Application of Neural Network

This paper aims at the characteristic of sensor system, synthesizes the output data of related sensor, expands related information according to the nonlinear application of neural network, and on the basis of different condition of the fault in single sensor of several sensors, in which that to single sensor fault, we can identify the fault sensor according to certain data fusion diagnosis strategy and to several sensor fault we can use a new neural network-rough neural network to locate fault sensor under more complex condition on the basis of single sensor. After the identification of fault sensor, we can recover fault signal at the same time and develop the accuracy of target description finally. Simulated experiment shows that the structure can identify and recover some common faults in sensor and has better effect.