Thermal Unit Commitment Using Binary Differential Evolution

This paper presents a new approach for thermal unit commitment (UC) using a differential evolution (DE) algorithm. DE is an effective, robust, and simple global optimization algorithm which only has a few control parameters and has been successfully applied to a wide range of optimization problems. However, the standard DE cannot be applied to binary optimization problems such as UC problems since it is restricted to continuous-valued spaces. This paper proposes binary differential evolution (BDE), which enables the DE to operate in binary spaces and applies the proposed BDE to UC problems. Furthermore, this paper includes heuristic-based constraint treatment techniques to deal with the minimum up/down time and spinning reserve constraints in UC problems. Since excessive spinning reserves can incur high operation costs, the unit de-commitment strategy is also introduced to improve the solution quality. To demonstrate the performance of the proposed BDE, it is applied to largescale power systems of up to 100-units with a 24-hour demand horizon.