A self-adaptive particle swarm optimisation and bacterial foraging hybrid algorithm

摘要

When used to deal with complex functions with high dimension, Bacterial Foraging Algorithm (BFA) converges slowly and Particle Swarm Optimisation (PSO) algorithm tends to premature convergence and low accuracy. Aiming at these shortcomings, an improved hybrid optimisation algorithm based on PSO and BFA is proposed in the paper (ABSO for short). The ABSO algorithm adds extremum disturbance to PSO. It also adaptively improves learning factors and inertial weight of PSO, chemotaxis step-length and disperse probability of BFA, respectively. BFA is used as the whole frame of the hybrid algorithm. After the chemotaxis operation of BFA, PSO is introduced to help BFA escape from local optima. This combines organically the optimisation update mechanism of PSO and the chemotaxis update mechanism of BFA, and can well balance the global search and local development capabilities. Simulation results on four benchmark functions show that the ABSO algorithm is superior to BFA, PSO, self-adaptive PSO and two other kinds of BFA hybrid algorithm in convergence speed, accuracy and robustness. This proves the validity of the ABSO algorithm in high-dimensional function optimisation problems.