Imitation Learning Based Fast Power System Production Cost Minimization Simulation
Qinran Hu, Zishan Guo, Fangxing Li
Abstract
Production cost minimization (PCM) simulation is an important tool for long-term power system simulation and assessment. However, solving a PCM problem is always time-consuming for its numerous binary variables. Besides, as modern energy systems have various planning options, the slow solution speed of PCM problems cannot satisfy the requirement of quick assessment of various plans. Most previous works on accelerating PCM problems ignore the importance of accurate solutions on proper assessment but only provide approximate solutions. Therefore, this work provides a fast PCM simulation method with optimality guarantee based on imitation learning. Compared with the popular open-source solver SCIP under default rules, the proposed method can find the optimal solution faster or provide smaller gap when the preset solving time limit hits. Simulation results show the effectiveness of the proposed method.