A comparative study of teaching-learning-self-study algorithms on benchmark function optimization

• Published in: The Korean journal of chemical engineering Vol. 34; no. 3; pp. 628 - 641
• Main Authors: Cho, Hyun-Jun, Ahmed, Faisal, Kim, Tae Young, Kim, Beom Seok, Yeo, Yeong-Koo
• Format: Journal Article
• Language: English
• Published: New York Springer US 01.03.2017; Springer Nature B.V; 한국화학공학회
• Subjects: Algorithms; Benchmarks; Biotechnology; Catalysis; Chemistry; Chemistry and Materials Science; Design optimization; Industrial Chemistry/Chemical Engineering; Machine learning; Materials Science; Mathematical analysis; Optimization algorithms; Process Safety; Process Systems Engineering; Self study; 화학공학; Teaching-learning-based Optimization; Teaching-learning-self-study; Comparative Study; Benchmark Function; Optimization
• ISSN: 0256-1115; 1975-7220
• DOI: 10.1007/s11814-016-0317-x

In typical optimization problems, the number of design variables may be large and their influence on the specific objective function can be complicated; the objective function may have some local optima while most chemical engineers are interested only in the global optimum. For any new optimization algorithms, it is essential to validate their performance, compare with other existing algorithms and check whether they provide the global optimum solutions, which can be done effectively by solving benchmark problems. In this work, seven typical optimization algorithms including the newly proposed TLBO (Teaching-learning-based optimization) based algorithms such as the TLSO (Teaching-learning-self-study optimization) algorithm have been reviewed and tested by using a set of 20 benchmark functions for unconstrained optimization problems to validate the performance and to assess these optimization algorithms. It was found that the TLSO algorithm shows the fastest convergence speed to the optimum and outperforms other algorithms for most test functions.