A novel multi-level quantization scheme for discrete particle swarm optimization

This paper presents a novel multi-level quantization scheme which best approximates the sigmoid function for multi-value discrete variable transformation in PSO. We define the set of multi-level quantization as integer multiples of powers-of-two terms to efficiently approximate the sigmoid function in transforming particle´s position into multilevel discrete values. In this paper, the feasibility of the proposed technique was tested in photovoltaic (PV) system allocation problem, and a comparison study with genetic algorithm (GA) is performed to show the quality of the solutions obtained.