Application of Artificial Neural Network and Information Fusion Technology in Power Transformer Condition Assessment

To meet the needs of assets management and risk assessment for power transformers in power systems, we proposed a condition assessment method of power transformer based on artificial neural network and information fusion technology. Taking preventative test parameters and on-line monitoring parameters as the example, we chose some representative part of them as static condition parameters, and chose the variation trends of parts of the static condition parameters as trend condition parameters. We normalized these condition parameters using a nonlinear index evaluation function, and established a model of multi-information fusion transformer condition assessment based on the artificial neuron network(ANN) and Dempster-Shafer(D-S) evidence theory. Moreover, we analyzed data of an example from a 500 kV power transformer, and the results verified the effectiveness of the proposed model. It is concluded that combining on-line monitoring parameters and their variation trends, the proposed method is helpful to improving the accuracy and timeliness of transformer condition assessment.