An improved Coati Optimization Algorithm with multiple strategies for engineering design optimization problems.

Area of Science:

• Computational Intelligence
• Optimization Algorithms
• Metaheuristics

Background:

• The Coati Optimization Algorithm (COA) faces limitations in late-stage optimization, including population diversity loss and premature convergence to local optima.
• Existing COA struggles with insufficient exploration and exploitation capabilities, hindering its overall performance.
• Need for enhanced optimization algorithms that maintain diversity and escape local optima for complex problems.

Purpose of the Study:

• To propose an improved Coati Optimization Algorithm (COA), termed TNTWCOA, addressing the performance issues of the original COA.
• To enhance the convergence speed, accuracy, and robustness of the COA through novel strategies.
• To validate the effectiveness of TNTWCOA on benchmark functions and real-world engineering design problems.

Main Methods:

• Introduction of a chaotic sequence mechanism for initializing population positions, ensuring a more uniform distribution and higher quality initial solutions.
• Incorporation of a nonlinear inertia weight factor to balance local exploitation and global exploration capabilities.
• Integration of an adaptive T-distribution variation strategy to increase population diversity and escape local optima, coupled with an alert update mechanism for adaptive searching.

Main Results:

• The TNTWCOA algorithm demonstrated significantly improved convergence speed and optimization accuracy compared to COA, ICOA, GJO, OOA, SCSO, and SABO on 29 IEEE CEC2017 test functions.
• Robust performance was observed across various optimization tasks, indicating the algorithm's reliability.
• TNTWCOA showed a strong solution advantage in engineering design problems, including pressure vessel, welding beam, three-bar truss, gear train, and speed reducer designs.

Conclusions:

• The proposed TNTWCOA algorithm effectively overcomes the limitations of the standard COA, offering superior performance in terms of speed, accuracy, and robustness.
• The combination of chaotic initialization, nonlinear inertia weight, adaptive T-distribution variation, and alert update mechanisms contributes to enhanced exploration and exploitation.
• TNTWCOA exhibits significant potential for practical application in complex engineering optimization problems.