An Enhanced Grasshopper Optimization Algorithm with Outpost and Multi-Population Mechanisms for Dolomite Lithology...

Area of Science:

• Computational Intelligence
• Optimization Algorithms
• Machine Learning

Background:

• The Grasshopper Optimization Algorithm (GOA) is recognized for its simplicity and effectiveness.
• However, GOA struggles with high-dimensional and complex optimization problems.
• There is a need for improved optimization techniques to handle complex scenarios.

Purpose of the Study:

• To propose an improved variant of the Grasshopper Optimization Algorithm (GOA) named Outpost Multi-population GOA (OMGOA).
• To enhance both local exploitation and global exploration capabilities of the GOA.
• To validate the performance and applicability of OMGOA on complex optimization tasks and a real-world engineering problem.

Main Methods:

• Development of OMGOA integrating an Outpost mechanism for enhanced local exploitation and a multi-population mechanism for global exploration and diversity.
• Conducting ablation studies to evaluate the contribution of each novel mechanism.
• Performing comparative experiments against other algorithms on multi-dimensional test functions and a lithology prediction task.

Main Results:

• OMGOA demonstrated superior optimization performance compared to existing algorithms in comparative experiments.
• Ablation studies confirmed the effectiveness of both the Outpost and multi-population mechanisms.
• The algorithm achieved competitive classification performance when applied to lithology prediction from petrophysical logs.

Conclusions:

• OMGOA effectively addresses the limitations of the standard GOA in high-dimensional and complex optimization tasks.
• The proposed mechanisms significantly improve exploration and exploitation balance, leading to better optimization outcomes.
• OMGOA shows practical utility and competitive performance in real-world engineering applications like lithology prediction.