Comparison of several improved versions of particle swarm optimizer algorithm for parameter estimation of squirrel-cage induction motors / Mohammad Yazdani-Asrami, Mehran Taghipour Gorjikolaie and S. Asghar Gholamian

In this paper, three versions of Particle Swarm Optimization (PSO) are proposed to estimate the equivalent
circuit parameters of squirrel cage induction motor. It is believed that how inertia weight changes during iterations can impact on final results. Constricted coefficients, linear model and exponential version are used as inertia weight, each of them presents different variations for inertia weight and consequently for particle movements and speed of such movements. In the linear version, particles start searching process with high speed and their speed will decrease by constant ramp, this kind of variation let to search all solution space in a short time and local search at the final iterations with low speed, also exponential version presents same treatment as linear version with non-linear variations in inertia weight and speed of movement. But, mathematical analysis shows that they trap into local minima and scientists presents
constricted version to solve this problem. In order to evaluate proposed versions additional to make changing in PSO’s version, sensitivity of proposed methods is analyzed using three sets of data. Results confirm the ability of proposed method which can estimate parameters with a possible least error.