Within-day rolling optimal scheduling problem for active distribution networks by multi-objective evolutionary algorithm based on decomposition integrating with thought of simulated annealing

Highlights: • Employing rolling dispatch mode to handle uncertainty of renewable sources in ADN. • Building rolling multi-objective optimal dispatch model considering four objectives. • Integrating simulated annealing algorithm to balance global and local searching ability of MOEA/D. • Using IEEE 33-bus distribution system to test and verify the proposed model and algorithm. The prediction accuracy plays a very important role in the optimal dispatch problem of active distribution networks (ADN). Basing on the fact that the prediction accuracy will be greatly improved when approaching prediction domains and decreasing prediction time periods, rolling optimization provides an alternative approach to handle uncertainty of renewable sources in ADN. Considering the characteristics of power supplies and loads in ADN, a comprehensive model of rolling multi-objective optimal dispatch is established. Four objective functions of minimizing the ADN operation cost, minimizing adjustment of the active power outputs to the day-ahead plan, minimizing the total active power loss, and minimizing the total voltage deviation of the system are considered simultaneously. Then, the thought of simulated annealing is integrated into the multi-objective evolutionary algorithm based on decomposition (MOEA/D) to change the process of generating new solutions and the neighborhood updating process. Therefore, the proposed MOEA/D with the thought of simulated annealing (MOEA/D-TSA) is adopted to solve this rolling multi-objective optimal dispatch problem. In the proposed approach, a dynamic probability value is employed to select which neighbor subproblem should be updated using the new generated solution in order to balance global and local searching ability of the algorithm. Finally, the performance of the proposed approach is tested and verified on an improved IEEE 33 node test system. The comparisons of the proposed method with the original MOEA/D and NSGA II on non-dominated solutions, extreme solutions, optimal compromise solution (OCS) and statistical indicators are well illustrated. The results show that MOEA/D-TSA algorithm has better convergence and comprehensive performance than other considered algorithms and its application in rolling dispatch problem is also presented through this test system.