Monitoring insulation surface conditions using neural networks

Surface tracking in solid insulators is an unwanted phenomenon, which cannot be accurately predicted. A wide range of relays can detect failure in a transmission line and prevent a total breakdown in the system, but in many cases it is too late to save the insulator from total damage. The method described here is mainly employed in detecting several conditions, such as discharges, leakage current, dry conditions, severe damage and tracking initiation. Initially a BPN (back propagation network) type neural network is trained with different signal types. Due to the nature of neural networks, which always require similar values for input nodes, the system uses the FFT of the input signal, which might have high amplitude frequency components other than the fundamental frequency due to the conditions on the surface. The system works on a real time basis and is able to make an estimate every 2-3 seconds, which can be reduced to milliseconds by using fast DSP boards. The program warns the user with the first indication of severe damage on the surface and can protect the insulator from excessive damage