Power transformer noise active control based on genetic radial basis wavelet neural network

In this paper, an intellectual technology for transformer noise reduction using radial basis wavelet neural network is investigated. The noise reduction technique proposed in this paper enhances noise reduction capability and improves the adaptability of the noise reduction system. The data required for the proposed method is transformer noise obtained during steady state operation. The data is then processed off-line using radial basis wavelet neural network in conjunction with GA to optimize the parameters of the neural network. The proposed technique is demonstrated using experimental data obtained from a transformer of Jiangdong Door substation in Nanjing. The simulation results confirm the effectiveness of GA to adaptively optimize the parameters of radial basis wavelet neural network and the big advantages of the proposed technique, such as improvements of noise reduction capability, stability and adaptability.