Improvement of Sensitivity-Based Nonlinear Solver for Bidding Strategy Problem of Electricity Markets

Abstract

The optimal bidding strategy of players in an electricity market is a bi-level optimization problem. In the upper level of this problem, each player tries to maximize its profit by optimal setting of its strategic factor, while in the lower level the market operator wants to minimize the energy supplying cost in the electricity network. As the players' cost functions are nonlinear, a nonlinear solver is required to solve the problem. Solving methods based on derivative computing and sensitivity coefficients are faced with two major bugs in finding the optimal point of this nonconvex-nonlinear problem: 1. getting stuck in the local optimal points, and 2. the long calculation time. In this paper, the “passing MCP technique” is proposed to avoid getting caught in local optimal points and the “dynamic step size technique” is proposed to improve the search speed and reduce the computational burden. The results obtained from simulation show that these two techniques can significantly reduce the computational burden.