Particle Swarm Optimization-Differential Evolution Algorithm and Its Application in the Optimal Reservoir Operation

PSO Algorithm is an effective global optimizing algorithm which is based on swarm intelligence. However, as its single evolution strategy, the particle swarm shows a kind of convergence as a whole. It will be easy to converge too early, which results in sinking into local optimal. This paper puts forward the algorithm of chaotic Particle Swarm Optimization-Differential Evolution and makes improvements to the regular PSO algorithm from the following three perspectives. Firstly, the abilities of exploitation and development of the algorithm are improved by the addition of variable inertia weight and study factors. Secondly, the tough problem of sticking at local optimal caused by the randomness of initial position of the particles is solved by initializing the particle swarm, which uses chaotic sequences based on logical map instead of random sequence in standard PSO. Also, the possibility of obtaining the global optimal adaptive value is further reinforced. Lastly, the speed formula including aberrance gene is firmly established by introducing the ideology of crossover, mutation and selection in Differential Evolution into the standard PSO algorithm, which could solve the defect of single evolution strategy in regular standard PSO algorithm. This strategy can enable it to jump out of local circulation and seek out the orbit of particle in general scope when the particle swarm sinks into local optimal or converge too early, thus sinking into precocious. According to the actual application, the method of optimal reservoir operation with CPSO-DE algorithm is put forward after analyzing the mathematic model. The corresponding model is then established and solved by combining with specific project practice. Proven by example, the CPSO-DE algorithm has an advantage over traditional particle swarm algorithm with quick convergence in the study of the plan of optimal reservoir operation, which identified its practicality, feasibility and robustness.