An Improved Teaching-Learning-Based Optimization with the Social Character of PSO for Global Optimization

Area of Science:

• Computational intelligence
• Optimization algorithms
• Swarm intelligence

Background:

• Teaching-Learning-Based Optimization (TLBO) is a population-based metaheuristic algorithm.
• Particle Swarm Optimization (PSO) is another popular swarm intelligence algorithm.
• Existing optimization methods may face challenges like premature convergence or high computational cost.

Purpose of the Study:

• To propose an improved Teaching-Learning-Based Optimization algorithm integrated with Particle Swarm Optimization (TLBO-PSO).
• To enhance the global search capability for function optimization problems.
• To address the stagnation issue in the original TLBO algorithm.

Main Methods:

• Modified the teacher phase of TLBO to incorporate student-teacher interactions and class performance.
• Determined new individual positions using old position, mean position, and best position of the current generation.
• Removed duplicate individual removal process to reduce computational cost.
• Incorporated a mutation operator to decrease the probability of local convergence.

Main Results:

• The improved TLBO-PSO algorithm effectively finds global solutions for function optimization problems.
• The method overcomes the stagnation problem where the mean student position equals the teacher position in original TLBO.
• Reduced computational cost by omitting duplicate individual removal.
• Demonstrated competitive performance against other existing optimization methods on benchmark functions.

Conclusions:

• The proposed TLBO-PSO is an effective and efficient optimization algorithm.
• The enhancements address key limitations of the original TLBO, improving its convergence properties.
• The algorithm shows promise for solving complex function optimization tasks.