A Multi-Verse Optimizer with Levy Flights for Numerical Optimization and Its Application in Test Scheduling for...

Area of Science:

• Computational Intelligence
• Optimization Algorithms
• Meta-heuristic Computing

Background:

• The Multi-Verse Optimizer (MVO) algorithm, while effective, can suffer from stagnation.
• Stagnation occurs when the MVO algorithm's exploration phase repeatedly focuses on solutions near the current best, limiting its ability to find global optima.
• Efficiently solving numerical and engineering optimization problems requires robust algorithms that avoid premature convergence.

Purpose of the Study:

• To introduce a novel meta-heuristic algorithm, the Levy flights Multi-Verse Optimizer (LFMVO).
• To enhance the exploration capabilities of the Multi-Verse Optimizer (MVO) by incorporating Levy flights.
• To address the stagnation problem inherent in the original MVO algorithm for numerical and engineering optimization tasks.

Main Methods:

• Integration of Levy flights, known for superior large-scale search capabilities, into the MVO framework.
• Modification of the MVO's exploration mechanism by applying Levy flights to the best-found universe to escape local optima.
• Validation of the LFMVO algorithm on 23 benchmark test functions and a Network-on-Chip (NoC) test scheduling problem (an NP-complete problem).

Main Results:

• The proposed LFMVO algorithm demonstrated a significant improvement over the original MVO and other peer algorithms.
• Experimental results confirmed that LFMVO achieves higher quality solutions compared to existing methods.
• The LFMVO algorithm exhibited faster convergence speeds in solving the tested optimization problems.

Conclusions:

• The Levy flights Multi-Verse Optimizer (LFMVO) effectively overcomes the stagnation issue of the MVO algorithm.
• LFMVO provides a more competitive approach for numerical and engineering optimization, offering enhanced solution quality and convergence.
• The integration of Levy flights presents a promising strategy for improving the performance of meta-heuristic optimization algorithms.