Enhancing economics of power systems through fast unit commitment with high time resolution

With the rapid increase of renewables, the unit commitment (UC) problem with high time resolution should be implemented to guarantee the economy and security of the power system. However, as a mixed-integer programming problem, its computational burden exponentially increases with the time resolution. Because of the unacceptable computational burden, many markets still adopt the UC with an hourly scheduling horizon. In this paper, a new approach is proposed to perform a fast calculation of the UC problem with high time resolution. The basic idea is to identify the representative scheduling points (RSPs) and reduce the size of optimization while causing a limited impact on the commitment solution. A selection strategy is proposed to identify RSPs. Based on RSPs, the construction of commitment solution is presented to obtain the result with original high time resolution. To guarantee feasibility, a correction strategy is proposed. Numerical results based on the IEEE 30-bus and 118-bus systems and a 661-bus utility system show that 1) the proposed method can substantially accelerate the computational speed, 2) the proposed approach can provide a feasible commitment solution within the desired accuracy loss, and 3) the market-clearing payments solved by the proposed method reflect a more accurate price signal.