Distribution network state awareness and risk prediction based on ELM algorithm

At present, with the continuous access of photovoltaic, energy storage and charging piles, the distribution network field is facing a comprehensive transformation of digitalization, informatization and networking. However, the existing distribution network lacks the ability of comprehensive analysis and risk prediction of the overall operation of the distribution network. In this paper, the security situation of the distribution network information energy network is studied from two aspects: multi-source spatial-temporal data fusion and situation risk prediction. First, the artificial fish swarm algorithm (AFSA) is used to optimize the limit learning machine (ELM) to spatial-temporal fusion the equipment information and access energy parameters of the flexible resource access point in the distribution network line, and the risk pattern recognition method of AFSA-ELM multi-classification curve analysis is constructed; Secondly, the parameters obtained from AFSA-ELM algorithm network are converted into information utility values., and the risk values of distribution network risk points obtained from AFSA-ELM are quantified to obtain the coordinates of actual risk points; Then, the Autoregressive Integrated Moving Average model(ARIMA) time series prediction algorithm is used to predict the risk change process of the risk points and reconfirm the situation of the risk points; Finally, the model is verified to have good evaluation and prediction effect by using the data collected from the power grid.