Teaching Learning based Optimization Applied to Mechanical Constrained Design Problems

Abstract

Amidst all the evolutionary optimization algorithms Teaching–Learning-Based Optimization (TLBO) seems to be a promising technique with relatively competitive performances. It outperforms some of the well-known metaheuristics regarding constrained benchmark functions, constrained mechanical design, and continuous non-linear numerical optimization problems. This dissertation presents the application of TLBO to various problems of mechanical engineering. Both constrained and unconstrained optimization has been performed on some manufacturing processes and design problems. Parametric optimization of three non-conventional machining processes namely electro-discharge machining, electrochemical machining and electro-chemical discharge machining, have been carried out and the results are compared with other evolutionary algorithms. Improvement in the existing TLBO algorithm has been incorporated in this dissertation using two schemes namely bit string mutation and replacement of worst solutions with fresh ones. Performance evaluation of these modifications have been presented in this dissertation by solving six optimization problems using original TLBO and proposed modifications. It has been found that better results are achieved in reaching the global optimal values by the use of these modifications. However, the results prefer the use of bit string mutation over scheme of replacing the worst solutions with fresh solutions in addition to the original logic of TLBO. The bit wise mutation and replacement of the worst solutions with fresh ones, proved an added advantage to the existing algorithm. Both these modifications resulted in a steeper convergence rate and finally provided global optimal solutions, and in some cases even better solutions than previously published results.