A New Approach in Applying Chaotic Cellular Automata in Particle Swarm Optimization

A New Approach in Applying Chaotic Cellular Automata in Particle Swarm Optimization

In PSO by using a random sequence with a random starting point as a basic parameters and by relying on this parameter it can update the positions and velocity of the particles. The main problem in using the PSO is in complex multi-peak search problems that usually leads to premature convergence. In this paper by using Chaotic Cellular Automata (CCA) a new and improved method for PSO is presented. By using Pseudo Random Number Generator (PRNG) in PSO we are able to produce chaotic numbers for inertial coefficient (ω), acceleration sufficient (C1 and C2) and random values (rand). These factors leads to the appropriate random behavior for the particles in the space’s problem which is capable for high exploitation ability. Furthermore this unpredictable behavior in changing the inertia coefficient ω with the combination of small steps by CCA help the proposed algorithm to avoid converging prematurely and falling in local minimums as well as covering the bigger problem space. In comparison with traditional algorithms, the proposed method illustrates faster and better performance for searching in space’s problems.

In PSO by using a random sequence with a random starting point as a basic parameters and by relying on this parameter it can update the positions and velocity of the particles. The main problem in using the PSO is in complex multi-peak search problems that usually leads to premature convergence. In this paper by using Chaotic Cellular Automata (CCA) a new and improved method for PSO is presented. By using Pseudo Random Number Generator (PRNG) in PSO we are able to produce chaotic numbers for inertial coefficient (ω), acceleration sufficient (C1 and C2) and random values (rand). These factors leads to the appropriate random behavior for the particles in the space’s problem which is capable for high exploitation ability. Furthermore this unpredictable behavior in changing the inertia coefficient ω with the combination of small steps by CCA help the proposed algorithm to avoid converging prematurely and falling in local minimums as well as covering the bigger problem space. In comparison with traditional algorithms, the proposed method illustrates faster and better performance for searching in space’s problems.