Scene prediction method for regional integrated energy system

The invention discloses a scene prediction method for a regional integrated energy system. The method comprises steps: the power of energy and load is predicted, counted and analyzed; energy and loadoutput power probability distribution is fit by adopting a non-parametric kernel density estimation method, and a probability distribution parameter is calculated; energy and load output power scenesare sampled according to the probability distribution parameter, random samples generated by sampling are substituted to energy and load output power cumulative probability distribution in each time period, and an actual energy and load power scene set is obtained; optimal clustering for the energy and load output power probability scene set is realized by adopting a particle swarm optimization algorithm and a K-mean algorithm; and through refined description on the system operation scene, a key typical scene can be quickly refined and selected, accurate and reliable decision-making basis is provided for design and ope