A Hybrid Particle Swarm Optimization for Numerical Optimization

Particle swarm optimization (PSO) has shown its good performance on numerical function problems. However, on some multimodal functions the PSO easily suffers from premature convergence because of the rapid decline in velocity. This paper presents a hybrid PSO for numerical optimization, namely HPSO, which employs opposition-based learning (OBL) and a modified velocity model. The OBL provides more chances to find solutions more closely to the global optimum. And the modified velocity model guarantees a non-zero velocity to help trapped particles jump out local minima. Experimental results on 6 benchmark functions show that the HPSO outperforms the standard PSO and opposition-based PSO in all test cases.