An improved ant colony optimization with an automatic updating mechanism for constraint satisfaction problems

• Published in: Expert systems with applications Vol. 164; p. 114021
• Main Authors: Guan, Boxin, Zhao, Yuhai, Li, Yuan
• Format: Journal Article
• Language: English
• Published: New York Elsevier Ltd 01.02.2021; Elsevier BV
• Subjects: Algorithms; Ant colony optimization; Automatic updating mechanism; Constraint satisfaction problem; Convergence; Evolutionary algorithms; Optimization; Constraint satisfaction problem; Ant colony optimization; Evolutionary algorithms; Automatic updating mechanism
• ISSN: 0957-4174; 1873-6793
• DOI: 10.1016/j.eswa.2020.114021

•The paper proposes an improved ant colony optimization.•An automatic updating mechanism is incorporated into the proposed algorithm.•The proposed algorithm can solve constraint satisfaction problems efficiently. Constraint satisfaction problem (CSP) is defined as a set of variables whose values need to satisfies a set of constraints. Ant colony optimization (ACO) has been proved to be a promising algorithm for solving the CSP, but the solution quality and convergence speed of existing ACO-based algorithms are not satisfactory. To overcome these drawbacks, this paper proposes an improved ant colony optimization with an automatic updating mechanism (AU-ACO). The idea of the automatic updating mechanism is to optimize an assignment without giving up the excellent variable-value pairs of the assignment. Under the impact of this mechanism, AU-ACO can only optimize the non-excellent variable-value pairs of a selected assignment, which results in the algorithm having a greater chance of finding better solutions. Furthermore, by optimizing only some variable-value pairs rather than all variable-value pairs, the convergence speed of the proposed algorithm is improved. AU-ACO is compared with eight other state-of-the-art algorithms on a wide range of binary problems, and experimental results demonstrate that AU-ACO a more effective and efficient for solving the CSP.