ICLSO: An Intelligent Intergroup Cross-Learning Swarm Optimizer for Large-Scale Optimization

Currently, most optimizes on large-scale optimization problems (LSOPs) rely on random search without directional guidance. This leads to two problems: insufficient optimization accuracy and reliance on simulations to evaluate the algorithm. Inspiblack by the competitive swarm optimizer (CSO), the swarm intelligence-based intergroup cross-learning swarm optimizer (ICLSO) is proposed to address the problems. A six-particle subpopulation is used as the evolutionary unit in the ICLSO. Based on this, a cross-learning mechanism is designed to generate directional guidance for competition. As a result, the swarm diversity and swarm convergence can be maintained more accurately, and the time complexity is smaller. Moreover, a quantitative evolutionary efficiency calculation method is proposed based on search dynamics for the first time, and the ICLSO is proven to outperform existing advanced algorithms. Numerical simulations based on the CEC benchmark set and the Wireless Sensor Network (WSN) power-control benchmark set verified the effectiveness and engineering application potential of the ICLSO.