An Improved Football Team Training Algorithm for Global Optimization.

Area of Science:

• Optimization algorithms
• Metaheuristic computing
• Artificial intelligence

Background:

• The Football Team Training Algorithm (FTTA) is a recent metaheuristic algorithm.
• FTTA exhibits good performance but struggles with convergence accuracy and local optima due to over-reliance on optimal individuals and insufficient perturbation.

Purpose of the Study:

• To introduce an Improved Football Team Training Algorithm (IFTA) addressing FTTA's limitations.
• To enhance exploration, exploitation, and convergence accuracy in metaheuristic optimization.

Main Methods:

• Developed a fitness distance-balanced collective training strategy to improve rational training and balance exploration/exploitation.
• Introduced a non-monopoly extra training strategy to enhance the ability to escape local optima.
• Implemented a population restart strategy to boost convergence accuracy and population diversity.

Main Results:

• IFTA demonstrated superior optimization performance compared to FTTA and six other algorithms on the CEC2017 test suite.
• Experimental results confirmed IFTA's strong optimization capabilities.
• IFTA showed potential for solving real-world engineering-constrained optimization problems.

Conclusions:

• The proposed IFTA effectively overcomes the limitations of the original FTTA.
• IFTA offers improved exploration, exploitation, and convergence accuracy.
• IFTA presents a promising approach for complex optimization tasks in real-world applications.