Hybrid of Particle Swarm Optimization and Simulated Annealing for Multidimensional Function Optimization

The classical Particle Swarm Optimization (PSO) Algorithm is very efficient and effective in solving optimization problems (both minimization and maximization). But PSO algorithm has a shortcoming of converging prematurely after getting trapped into some local optima (local optimum solution point) and considers it to be the global optima (global optimum solution point). Moreover, when we apply it to a multi-dimensional complex problem scenario, then due to some constraints it becomes nearly impossible to get out from that local optima (apparent global optima) and reach out for the global optima. Instead, all the particles starts getting converged to that apparent optimum solution. On the contrary, Simulated Annealing (SA) Algorithm can hinder the premature convergence to the local optima and diverges the particles using its strong ability of local search. Here, we propose a new hybrid algorithm of Particle Swarm Optimization (PSO) and Simulated Annealing (SA) in optimization (We applied and concentrated on minimization problems) of complex, multidimensional functions. The proposed algorithm is fundamentally based on the PSO algorithm, whereas, SA method is used to slow down the convergence of the swarm and to increase the swarm’s probability of reaching the global optima by increasing the diversity of Moitree Basu, Pradipta Deb, Gautam Garai