A new human-based metaheuristic algorithm for solving optimization problems on the base of simulation of driving...

Area of Science:

• Computational Intelligence
• Optimization Algorithms
• Machine Learning

Background:

• Stochastic optimization algorithms are crucial for solving complex problems.
• Existing algorithms often struggle with balancing exploration and exploitation.
• Inspiration from real-world learning processes can lead to novel algorithmic designs.

Purpose of the Study:

• Introduce a novel stochastic optimization algorithm named Driving Training-Based Optimization (DTBO).
• To mathematically model the driving training process for optimization.
• Evaluate DTBO's performance against established algorithms on diverse benchmark functions.

Main Methods:

• DTBO is mathematically modeled based on three phases: instructor training, student skill patterning, and practice.
• The algorithm's performance was tested on 53 standard objective functions, including unimodal, multimodal, and IEEE CEC2017 test functions.
• Comparative analysis was conducted against 11 well-known optimization algorithms.

Main Results:

• DTBO demonstrated an effective balance between exploration and exploitation capabilities.
• The algorithm achieved superior performance compared to 11 competitor algorithms across various test functions.
• DTBO provided appropriate and efficient solutions for the tested optimization problems.

Conclusions:

• Driving Training-Based Optimization (DTBO) is a promising new algorithm for stochastic optimization.
• The approach of mimicking human training processes can yield effective optimization strategies.
• DTBO offers a competitive and efficient alternative for a wide range of optimization applications.