Parameter Identification of an Over-Determined and Over-Constrained System using PSO

This paper proposes a curve fitting method for an over-determined system by a function that is the sum of sinusoidal terms. A Particle Swarm Optimization (PSO) algorithm for this trigonometry curve fitting having a large number of unknown parameters is developed, which is needed in engineering applications. For the over-determined system, PSO based algorithm employs an additional hybrid term for velocity updation to determine the unknown parameters of the sine function. This additional hybrid term provides a better convergence and the iterations proceeds systematically. The algorithm is validated by fitting curve with a function having 18 parameters of mixed type. The code for the algorithm in MATLAB software is presented to establish the general approach.