An Improved Particle Swarm Optimization Algorithm with Sentient Mode

This paper presents a modification of the sentient particle swarm optimization algorithm intended to introduce some psychology factor of emotion into the algorithm. At the same time, the algorithm compares the real velocity and position error with the threshold for finding conditionality condition of dissipative particle. It guides individuals to behave reasonably with the capability of self-adaptation in activities of self-cognition and sociality via quantifying the sensation intension as a result of environmental stimulation according to the sensation model. Considering the complexity of a swarm intelligent system at the level of sensation brings about optimization of the comprehensive capability of global, local searching and cooperating with each other. The resulting algorithm is shown to perform significantly better than the original algorithm on different benchmark optimization problems. Avoiding premature convergence allows the improved SPSO to continue search for global optima, reaching better solutions than SPSO with a much faster convergence speed.