Study on weather-related natural contaminant deposit prediction of insulators based on neural network

According to the statistics, the faults of power system caused by pollution flashover are the most familiar kind of transmission net faults. As a result, it´s very important to avoid pollution flashover. If the amount of the dirt can be forecasted right, the just in time maintenance can be taken. As a result, the cost can be cut down greatly and the effect can be better than to clean up regularly. During the research, the testing insulators are hanged on the pylons in typical areas. The equivalent salty deposit density (ESDD) of these insulators surfaces are measured regularly. At the same time, the weather data of these areas are gathered. To forecast the condition of pollution, a neural network which uses ESDD as outputs is established. By training the network with gathered data, the regular pattern of pollution is simulated. The result of this research can be used in field, and can initiate just in time maintenance.