Advanced Optimization by PMSM of PSO and Neural Network

This paper suggests novel optimal approach by progressive mapping search method (PMSM) of neural network aided particle swarm optimization (PSO) that can obtain global optimal solution easily and speed searching time up by PMSM. The PMSM by NN and PSO has an important role as navigation when PSO is going to search all areas to have an optimal solution, it can help to increase searching capability of PSO. That is, the PMSM by NN and PSO is also trained to capture the PSO-searched data in system. To prove the method, we use four test function: De Jong’s function-1, Rosenbrock’s valley (De Jong’s function-2), Himmelblau function, Rastrigin’s function-6. The PMSM method suggested in this paper is faster than the traditional PSO method in four test function. We also apply this optimal approach into AVR (Automatic Voltage Regulator) system in thermal power plant. The response is quite faster and more stable. Key-Words: PSO, Neural Network, Hybrid system, Optimization, Learning system, Artificial intelligence.