Power quality disturbance identification method based on multi-domain feature fusion

Power quality disturbance identification is vital for power quality study. However, noise, interference between disturbances and the effect of feature extraction method may lead to edge blurring of features extracted from different disturbances, thus affecting the accurate of disturbance recognition. Thence, the paper proposes a recognition method based on multi-domain feature fusion. Firstly, the neural network is trained preliminarily by mixed features of different domains, and then with the input characteristics of each domain, the action probability of hidden layer neurons in corresponding domain is determined according to the changes of cross-entropy before and after retention of the hidden layer neurons. Finally, work out the final results by the DS evidence theory, in which the independent evidence is transformed from identification results of unknown disturbances in different domains. The algorithm reduces the influence of errors in characteristics of a single domain on the identification accuracy, and is robust to noise and stable.