Coevolutionary particle swarm optimization using AIS and its application in multiparameter estimation of PMSM

Area of Science:

• Computational Intelligence
• Optimization Algorithms
• Machine Learning

Background:

• Particle Swarm Optimization (PSO) is a widely used metaheuristic for optimization.
• Existing PSO variants often struggle with premature convergence and limited global search capabilities.
• Artificial immune systems offer principles for robust and adaptive optimization.

Purpose of the Study:

• To propose a novel coevolutionary particle swarm optimization (PSO) algorithm integrated with artificial immune principles.
• To enhance the global search ability and convergence speed of PSO.
• To apply the proposed algorithm for multiparameter estimation in permanent magnet synchronous machines.

Main Methods:

• A coevolutionary framework dividing the population into elite and normal subpopulations.
• Hybrid method for creating new individuals using diverse operators to ensure subpopulation diversity.
• Adaptive wavelet learning operator to accelerate convergence of personal best (pbest) particles.
• Immune-clonal-selection operator for elite subpopulation optimization and migration for inter-subpopulation information exchange.

Main Results:

• The proposed algorithm exhibits faster convergence and improved global search ability on standard benchmark functions.
• Significant performance improvement over existing PSO methods in multiparameter estimation of permanent magnet synchronous machines.
• Successful simultaneous estimation of machine dq-axis inductances, stator winding resistance, and rotor flux linkage.

Conclusions:

• The proposed coevolutionary PSO algorithm effectively enhances optimization performance.
• The integration of artificial immune principles offers a robust approach to complex optimization problems.
• The algorithm shows practical applicability in accurate parameter estimation for electrical machines.