Hybrid Particle Swarm Optimization based Day-Ahead Self-Scheduling for Thermal Generator in Competitive Electricity Market

This paper presents a hybrid particle swarm optimization algorithm (HPSO) to solve the day-ahead self-scheduling for thermal power producer in competitive electricity market. The objective functions considered to model the self-scheduling problem are 1) to maximize the profit from selling energy in day-ahead energy market subject to operational constraints and 2) at the same time, to minimize the risk due to uncertainty in price forecast. Therefore, it is a conflicting bi-objective optimization problem which has both binary and continuous optimization variables considered as constrained mixed integer nonlinear programming. To demonstrate the effectiveness of the proposed method for self-scheduling in a day-ahead energy market, the locational margin price (LMP) forecast uncertainty in PJM electricity market is considered. An adaptive wavelet neural network (AWNN) is used to forecast the day-ahead LMPs. The effect of risk is explicitly modeled by taking into account the estimated variance of the day-ahead LMPs.