Research on Accelerated Solving Method of Security-Constrained Unit Commitment Based on Heuristic Knowledge

Abstract

Aiming at the efficiency bottleneck problem of solving the security-constrained unit commitment (SCUC)with mixed-integer programming (MIP), a solution method based on heuristic knowledge and MIP(HKMIP) is proposed in this paper. Firstly, a mapping model of load and unit status is established through deep learning (DL). The historical data of load and solution are selected as the data set. Next, dual-threshold is introduced to judge the on/off state of the unit. In the new load scenario, the states of some units are determined by the mapping model. Simultaneously, as heuristic knowledge, the determined unit states are written into the original SCUC model. And the status and output of the remaining units are obtained through the solver. Finally, the IEEE-RTS96 test case is used as the experimental simulation platform. The cost and solution efficiency of MIP and HKMIP are compared. The results show that the model can significantly accelerate the solution efficiency while ensuring a high-quality solution, which verifies the feasibility and effectiveness of the discussed method.