Particle Swarm Optimization Based On The Principle Of Improvement Of Animal Feeding Studies

Particle swarm optimization (PSO) is a population-based swarm intelligent algorithm by simulating bird flocking and fish schooling. Due to the easy implementation and few parameters, it has been widely applied to many areas. In this article, we propose several modifications for PSO inspired by animal feeding behaviors.In standard PSO, the behavior of each particle is decided by the food resources. However, in nature, the animal feeding pattern is also influenced by living pressure especially for hungry risk. Therefore, a new variant--food-guided PSO is proposed in which each particle employs an inner living pressure index, and searching food according to the trade-off between food resources and inner index.Predation risk is another important factor for the pattern. To simulate this phenomenon, a new variant of PSO--risk-benefit PSO is introduced in which predation risk is incorporated into the food-guided PSO.In risk-benefit PSO, the hungry risk is decreased after finding new food resources. However, for some animals, they may maintain more motivations to seek the food in this case. Therefore, the incentive PSO is designed mimicking this behavior. To testify the performance, seven famous benchmarks are used to test. Simulation results show this variant is superior to other four variants of PSO including the food-guided PSO and risk-benefit PSO, especially for high-dimension multi-modal problems.