A Modified Sine Cosine Algorithm with Teacher Supervision Learning for Global Optimization

Abstract

The sine cosine algorithm (SCA) is a recently developed meta-heuristic algorithm for solving global optimization problems. It has shown excellent performance in meta-heuristic algorithms. But this algorithm also has shortcomings such as low accuracy, easy to fall into a local solution, and slow convergence speed. Aiming at these deficiencies of the SCA, a modified sine cosine algorithm with teacher supervision learning (TSL-SCA) for global optimization is proposed. First, the teacher supervision strategy can guide the population convergence and accelerate the convergence speed. Second, individuals perform reflective learning after the standard SCA position is updated, which can effectively prevent individuals from stagnating in the evolutionary process and increase population diversity. In addition, a hybrid inverse learning method is proposed. It can not only enhance the ability of finding global optimal solution and increase population distributivity, but also balance the exploration and exploitation capabilities. Differential evolution algorithm (DE), particle swarm optimization (PSO), cuckoo search (CS) algorithm, moth-flame optimization (MFO), whale optimization algorithm (WOA), Teaching-Learning-Based Optimization (TLBO), SCA and TSL-SCA are selected for simulation experiment to solve 33 benchmark optimization problems. The experimental results show that the TSL-SCA can significantly enhance the optimization accuracy and convergence speed. Furthermore, the effectiveness of the proposed method is examined by solving analog circuit fault diagnosis of filter circuit examples.