A Learning Sparrow Search Algorithm.

Area of Science:

• Computational Intelligence
• Optimization Algorithms
• Swarm Intelligence

Background:

• Traditional intelligent optimization algorithms often suffer from premature convergence to local optima.
• The standard sparrow search algorithm (SSA) exhibits high randomness and a tendency to get trapped in local optima, limiting its effectiveness.

Purpose of the Study:

• To address the limitations of existing optimization algorithms, particularly the tendency to fall into local optima.
• To propose and validate a novel learning sparrow search algorithm (LSSA) with improved search capabilities and robustness.

Main Methods:

• Introduction of a lens reverse learning strategy in the discoverer stage to enhance population diversity and search flexibility.
• Integration of an improved sine and cosine guidance mechanism in the follower stage for more detailed exploration.
• Application of a differential-based local search strategy to refine optimal solutions and prevent the omission of high-quality candidates.

Main Results:

• LSSA demonstrated superior performance compared to various established algorithms (SSA, GWO, PSO, etc.) on 12 benchmark functions and the CEC 2017 test suite.
• The algorithm exhibited good universality and effectiveness in solving complex optimization problems.
• LSSA proved practical and effective in robot path planning, showcasing good stability and safety.

Conclusions:

• The proposed learning sparrow search algorithm (LSSA) effectively mitigates the local optimum problem in intelligent optimization.
• LSSA offers enhanced population diversity, flexible search mechanisms, and robust solution refinement.
• The algorithm's practical applicability and reliability are confirmed through benchmark testing and robot path planning simulations.