MATLAB Implementation of Two Intelligent Optimization Algorithms: Biogeography-Based Optimization and Particle Swarm Optimization

This article presents MATLAB implementations of two intelligent optimization algorithms: Biogeography-Based Optimization (BBO) and Particle Swarm Optimization (PSO). We begin by discussing the fundamental principles and applications of both algorithms, including their mathematical foundations and pseudocode representations. The BBO implementation typically involves habitat suitability index calculation, migration operations between habitats using cosine migration models, and mutation mechanisms for diversity preservation. The PSO code structure generally includes particle position and velocity initialization, fitness evaluation, personal best and global best updates, and velocity-position update equations with inertia weight adjustments. We then explore practical applications of these algorithms in solving real-world problems, including function optimization tasks (such as Rosenbrock and Rastrigin functions) and machine learning model parameter tuning. The MATLAB implementations demonstrate how to configure algorithm parameters like population size, migration rates for BBO, and cognitive/social parameters for PSO. We also discuss enhancement strategies for these algorithms, such as hybrid approaches combining BBO's migration mechanisms with PSO's velocity updates, or incorporating opposition-based learning for improved convergence. These modifications enable broader application across various domains and problem types. Finally, we summarize the comparative advantages and limitations of each algorithm: BBO's strength in maintaining population diversity versus PSO's rapid convergence characteristics, along with computational complexity considerations. We conclude by outlining future research directions, including multi-objective optimization extensions and GPU-accelerated implementations for large-scale problems.