Hybrid bare-bones PSO for dynamic economic dispatch with valve-point effects

• A new PSO-based hybrid algorithm is presented to solve dynamic economic dispatch problems with valve-point effects.
• An improved bare-bones PSO and a directionally chaotic search are well combined to achieve better performance.
• An adaptive disturbance factor and a new genetic operator are incorporated into the improved BBPSO.
• An effective constraint handing mechanism is introduced to solve complicated equality and inequality constraints.
• The proposed algorithm shows good performance to solve the DED problem.

This paper presents an efficient hybrid particle swarm optimization algorithm to solve dynamic economic dispatch problems with valve-point effects, by integrating an improved bare-bones particle swarm optimization (BBPSO) with a local searcher called directionally chaotic search (DCS). The improved BBPSO is designed as a basic level search, which can give a good direction to optimal regions, while DCS is used as a fine-tuning operator to locate optimal solution. And an adaptive disturbance factor and a new genetic operator are also incorporated into the improved BBPSO to enhance its search capability. Moreover, a heuristic handing mechanism for constraints is introduced to modify infeasible particles. Finally, the proposed algorithm is applied to the 5-, 10-, 30-unit-test power systems and several numerical functions, and a comparative study is carried out with other existing methods. Results clarify the significance of the proposed algorithm and verify its performance.

A new genetic operator by integrating the mutation and the crossover operators, called the M+C operator, where a genetic probability pe is used to control the recombination speed among particles. In the 8-th line, when G  (0, 1) takes a small value, like the crossover operator in an evolutionary algorithm, crossing with random Pbestlk can help a particle to construct good schemas rapidly; when G(0, 1) takes a big value, like the mutation operator, the value of G(0, 1) × rang can help the particle to escape from local optima.Figure optionsDownload as PowerPoint slide