针对灰狼优化算法存在的种群多样性不足、开发与探索不平衡以及过早收敛等问题,提出一种增强型改进灰狼优化算法(EIGWO)并应用于塑料激光焊接压力控制参数优化。改进主要包括三个方面:首先,提出一种基于维度学习的狩猎搜索策略(DLH),用...针对灰狼优化算法存在的种群多样性不足、开发与探索不平衡以及过早收敛等问题,提出一种增强型改进灰狼优化算法(EIGWO)并应用于塑料激光焊接压力控制参数优化。改进主要包括三个方面:首先,提出一种基于维度学习的狩猎搜索策略(DLH),用不同方法为每只狼构建一个可以共享信息的邻居,以此增加种群的多样性;其次,提出一种非线性的收敛策略来平衡勘探和开发,并通过修正参数A和C来模拟狼群狩猎时头狼和次头狼的交替行为;最后,对动态位置更新方程进行修正,保持种群多样性同时避免过早收敛。通过基准测试函数在多个维度下通过与原始的灰狼算法和两种改进的灰狼算法(IGWO, LGWO)以及粒子群算法(PSO)进行对比实验,实验结果表明,EIGWO具有更好的寻优性能和稳定性。在PID的参数优化应用中,通过改变参数和添加扰动,证明EIGWO具有更好的优化效果,且算法的稳定性和鲁棒性也更好。,br> Aiming at the problems of insufficient population diversity, unbalanced development and exploration and premature convergence in grey wolf optimization algorithm, an enhanced improved grey wolf optimization algorithm (EIGWO) is proposed and applied to the optimization of pressure control parameters in plastic laser welding. The improvement mainly includes three aspects: firstly, a new search strategy named dimension learning-based hunting (DLH) is proposed, and a neighbor who can share information for each wolf is constructed by different methods, so as to increase the diversity of the population. Secondly, a non-linear convergence strategy is proposed to balance exploration and development, and the alternating behavior of the first wolf and the second wolf is simulated by modifying parameters A and C. Finally, the dynamic position updating equation is modified to maintain population diversity and avoid premature convergence. In this paper, the benchmark function is compared with the original grey wolf algorithm, two improved grey wolf algorithms (IGWO, LGWO) and particle swarm optimization (PSO) in multiple dimensions. The experimental results show that EIGWO has better optimization performance and stability. In the application of PID parameter optimization, by changing parameters and adding disturbance, it is proved that EIGWO has better optimization effect, and the stability and robustness of the algorithm are also better.展开更多
文摘针对灰狼优化算法存在的种群多样性不足、开发与探索不平衡以及过早收敛等问题,提出一种增强型改进灰狼优化算法(EIGWO)并应用于塑料激光焊接压力控制参数优化。改进主要包括三个方面:首先,提出一种基于维度学习的狩猎搜索策略(DLH),用不同方法为每只狼构建一个可以共享信息的邻居,以此增加种群的多样性;其次,提出一种非线性的收敛策略来平衡勘探和开发,并通过修正参数A和C来模拟狼群狩猎时头狼和次头狼的交替行为;最后,对动态位置更新方程进行修正,保持种群多样性同时避免过早收敛。通过基准测试函数在多个维度下通过与原始的灰狼算法和两种改进的灰狼算法(IGWO, LGWO)以及粒子群算法(PSO)进行对比实验,实验结果表明,EIGWO具有更好的寻优性能和稳定性。在PID的参数优化应用中,通过改变参数和添加扰动,证明EIGWO具有更好的优化效果,且算法的稳定性和鲁棒性也更好。,br> Aiming at the problems of insufficient population diversity, unbalanced development and exploration and premature convergence in grey wolf optimization algorithm, an enhanced improved grey wolf optimization algorithm (EIGWO) is proposed and applied to the optimization of pressure control parameters in plastic laser welding. The improvement mainly includes three aspects: firstly, a new search strategy named dimension learning-based hunting (DLH) is proposed, and a neighbor who can share information for each wolf is constructed by different methods, so as to increase the diversity of the population. Secondly, a non-linear convergence strategy is proposed to balance exploration and development, and the alternating behavior of the first wolf and the second wolf is simulated by modifying parameters A and C. Finally, the dynamic position updating equation is modified to maintain population diversity and avoid premature convergence. In this paper, the benchmark function is compared with the original grey wolf algorithm, two improved grey wolf algorithms (IGWO, LGWO) and particle swarm optimization (PSO) in multiple dimensions. The experimental results show that EIGWO has better optimization performance and stability. In the application of PID parameter optimization, by changing parameters and adding disturbance, it is proved that EIGWO has better optimization effect, and the stability and robustness of the algorithm are also better.
