Yihui Zhou: an Improved Differential Evolution Algorithm for Power Optimization ..

Optimal reactive power dispatch of power systems is a complicated nonlinear combinatorial optimization problem. The Differential Evolution Algorithm (DEA) is an effective method to solve such problems. But DE still has some drawbacks like slow convergence speed and the parameters selection having a great effect on the algorithm performance. So a modified differential evolution is proposed in the paper. Analyzing the mutation operator for vector operations, a new mutation strategy with direction is put forward which improves the convergence rate of the algorithm. According to the individual evolution process, a self-learning mechanism of the parameters is introduced. The mechanism adjusts the mutation and crossover rate of the individuals adaptively to improve the robustness of the algorithm. In the optimal reactive power dispatch problem, it is important to deal with the inequality constraint violation: the conventional method uses a penalty function, but an appropriate penalty coefficient is difficult to choose. The paper adopts a non-parameter method to handle the constraint conditions, which can effectively guarantee the state variables within the limits. The effectiveness of the proposed modified differential evolution and the adopted non-parameter method to handle constraint conditions are demonstrated on IEEE 30-bus, 57-bus and 118-bus test power systems.