| This project is designed to mimic automation of analog filter analysis to examine some efficient algorithm useful in filter synthesis. The process involves formation of MNA matrix to create symbolic transfer functions in s domain, continuous and discrete sizing of LC components using evolutionary algorithms; and finally, the performance of each algorithm is studied based on fixed error criterion and adaptability to discrete problem.;Efficiency of the clever algorithms in optimizing piecewise filter response is ultimately dependent on the quality of the fitness function. A unique measure of error called Sum of Maximum Deviation (SMD) is implemented which evaluates the performance of global optimizer by weighing important details per unit sampled frequency.;From global optimization point of view, it is certain that discrete evolutionary algorithms lacks the absoluteness of brute force analysis; however, the general continuous optimization is stretched to accommodate a new proximity estimator alongside its elementary constraint. |
