摘要
针对经典差分进化(DE)算法存在后期种群多样性低、优化速度慢等缺陷,提出一种基于精英强化策略的DE(EEDE)算法。对DE算法进行改进,引入随机缩放向量和随机全域变异,同时借用模式搜索理念改进算法的精英强化算子,使得改进算法在全局性、优化速度和精度得到提高。再用该改进算法将高阶模型辨识为一阶时滞模型,用ITAE评价指标对辨识的模型PID参数进行优化整定,将所得整定参数用于原系统控制,检验模型辨识与控制问题。最后,通过MATLAB实验平台将该算法整定的PID与ZN法整定进行对比验证,实验结果证明:该算法用于系统模型辨识和控制参数优化方面可行且有效。
Aiming at the shortcomings of the classical differential evolution(DE)algorithm,such as low population diversity and slow optimization speed,an elite enhanced DE(EEDE)algorithm is proposed.The DE algorithm is improved by introducing random scaling vector and random global variation,and using the idea of pattern search to improve the elite enhancement operator of the algorithm,which improves the global,optimization speed and precision of the improved algorithm.Then,the improved algorithm is used to identify the higher-order model as a first-order time-delay model,and the proportion integration differentiation(PID)parameters of the identified model are optimized and adjusted by the ITAE evaluation index,and the tuning parameters are applied to the original system control to test the model identification and control problems.Finally,the PID tuning of this algorithm is compared with that of ZieglerNichols(ZN)method by MATLAB experiment platform,and the experimental results prove that this algorithm is feasible and effective in system model identification and control parameter optimization.
作者
黄军付
谭飞
谢涛
HUANG Junfu;TAN Fei;XIE Tao(School of Automation and Information Engineering,Sichuan University of Science&Engineering,Yibin 644000,China;Artificial Intelligence Key Laboratory of Sichuan Province,Yibin 644000,China)
出处
《传感器与微系统》
北大核心
2025年第10期134-137,141,共5页
Transducer and Microsystem Technologies
基金
国家自然科学基金资助项目(61902268)
四川省科技计划项目(21ZDYF4052,2020YFH0124,2021YFSY0060)。
作者简介
黄军付(1998-),男,硕士研究生,研究方向为智能算法;通讯作者:谭飞(1972-),男,硕士,副教授,研究领域为智能控制和计算机应用研究。
