Optimizing Laying Hen Diet using Multi-Swarm Particle Swarm Optimization

Abstract

Formulating animal diet by accounting fluctuating cost, nutrient requirement, balanced amino acids, and maximum composition simultaneously is a difficult and complex task. Manual formulation and Linear Programming encounter difficulty to solve this problem. Furthermore, the complexity of laying hen diet problem is change through ingredient choices. Thus, an advanced technique to enhance formula quality is a vital necessity. This paper proposes the Multi-Swarm Particle Swarm Optimization (MSPSO) to enhance the diversity of particles and prevent premature convergence in PSO. MSPSO work cooperatively and competitively to optimize laying hen diet and produce improved and stable formula than Genetic Algorithm, Hybridization of Adaptive Genetic Algorithm and Simulated Annealing, and Standard Particle Swarm Optimization with less time complexity. In addition, swarm size, iteration, and inertia weight parameters are investigated and show that swarm size of 50 for each sub-swarm, total iteration of 16,000, and inertia weight of 6.0 should be used as a good parameter for MSPSO to optimize laying hen diet.