Improved differential evolution algorithm for integrated energy system scheduling

Optimization models for an arbitrary integrated energy system (IES) are usually of high dimensions, and thus cannot be directly solved by classical optimization algorithms. To tackle this challenge, this paper analyzes advantages and disadvantages of common heuristic algorithms, and the differential evolution algorithm (DE) is selected for its fast convergence speed and robustness. Considering the fact that the results of traditional DE often contain some infeasible values, an improved DE algorithm by adding a filtration step is proposed to avoid infeasible values and improve the overall accuracy. To deal with the operational constraints of the IES that are difficult to handle by DE algorithms, the sequential unconstrained miniaturization technique is further adapted. Case studies are conducted for two typical objectives, i.e., operational cost minimization, carbon emission minimization, to illustrate the improved accuracy and applicability of the proposed algorithm.