Day-ahead Scheduling of Multiple Thermal Power Plants with Output Constraints Based on PV Interval Prediction

This paper studies a day-ahead scheduling problem on power generation of thermal power plants and charge/discharge of battery energy storage systems, where the confidence intervals of the prediction of photovoltaic (PV) and demand power are available. This problem is reduced to an interval optimization problem, where parameters having any values in certain intervals are included. We have developed a method for efficiently solving this kind of problems using tools from the interval analysis. However, the class of the problem that we have studied there is limited in the sense that the output capacity constraints of thermal power plants are not taken into account. This paper proposes a new method for efficiently solving a more practical scheduling problem with the capacity constraints of thermal power plants added. To use our method, we need to analyze a matrix corresponding to a Jacobian of a solution to an optimization with respect to parameters on demand deviation, which is too complex to analyze in general. The key to overcome this difficulty is that we introduce a virtual thermal power plant whose output capacity constraint is not imposed. By using this idea, we can transform the above matrix into an appropriate matrix that is analyzed as easily as possible. The efficiency of the proposed method is shown by numerical simulations on the power system in the Tokyo area.