Automatic electrical fault detection method for power supply and distribution based on fuzzy C-mean

Effective fault risk early warning and timely risk prevention and control measures are of great significance to improve the reliability of power supply. In this paper, a distribution network fault early warning model based on LSTM and FCM clustering is proposed, and an operating parameter prediction method based on the fused fault characteristics is proposed. After that, this paper selects FCM clustering algorithm to identify the boundary conditions of fault state and non-fault state of distribution network, and uses LSTM to predict the characteristic quantities of distribution network in the short term, and further combines the prediction results of characteristic quantities and the obtained state distinguishing boundary conditions to predict the short-term fault occurrence moment in the future, so as to realize fault early warning. This method has a good prediction effect in the actual example data, and can effectively predict the fluctuation of distribution network characteristic quantity and its specific occurrence moment, which can be used as the technical support for subsequent fault early warning. Through simulation analysis, it can be seen that the method proposed in this paper can effectively detect the traveling wave signal when the line fault is weak, and the detection results are not easily affected by the fault type, load fluctuation and white noise.