摘要
针对传统的BP神经网络PID(BP-PID)控制因其初始权值随机,导致系统的收敛速度较慢、控制前期会有较大误差和BP神经网络初始权值优化等问题,建立了滚珠丝杠进给系统伺服三环模型,设计了BP-PID控制器,提出了一种二阶振荡混沌映射粒子群算法(SCMPSO)优化滚珠丝杠进给系统BP-PID控制器。首先,混沌映射初始化粒子位置,使粒子均匀分布于空间,增加粒子解的多样性;随后,提出一种非线性余弦自适应惯性权重,以平衡算法的全局搜索能力和局部搜索能力;其次,在算法中引入二阶振荡环节,在面对突变多峰干扰时,能及时跳出局部最优解。研究结果表明:当加入外界干扰时,控制策略SCMPSO-BP-PID在正向进给时段的位移平均误差为0.013 mm,相比SAWPSO-BP-PID、LDWPSO-BP-PID、PSO-BP-PID这3种控制策略分别提升约45.8%、55.2%、61.7%;当加入阶跃响应时,SCMPSO-BP-PID的最大超调量仅为0.029,系统调节时间和峰值时间相比3种控制策略均有较大提升,具有较高的控制精度和稳定性。
Traditional BP neural network PID(BP-PID)control has a slow convergence speed due to its random initial weights,resulting in significant errors in the early stages of control.This paper focuses on the initial weight optimization problem of BP neural network,establishes a servo three loop model of the ball screw feed system,designs a BP-PID controller,and proposes a second-order oscillation chaotic mapping particle swarm optimization algorithm(SCMPSO)to optimize the BP-PID controller of the ball screw feed system.Firstly,the chaotic mapping initializes the particle position,making the particles evenly distributed in space and increasing the diversity of particle solutions.Then,a nonlinear cosine adaptive inertia weight is proposed to balance the global search ability and local search ability of the algorithm.Finally,a second-order oscillation link is introduced in the algorithm,which can timely jump out of the local optimal solution in the face of sudden multi peak interference.The results show that when external interference is added,the average displacement error of SCMPSO-BP-PID during the forward feed period is 0.013 mm,which is about 45.8%,55.2%,and 61.7%higher than the three control strategies of SAWPSO-BP-PID,LDWPSO-BP-PID,and PSO-BP-PID,respectively.When step response is added,the maximum overshoot of SCMPSO-BP-PID is only 0.029,and the system adjustment time and peak time are significantly improved compared with the three control strategies,representing high control accuracy and stability.
作者
吴沁
周顺仟
王星联
WU Qin;ZHOU Shunqian;WANG Xinglian(School of Mechanical and Electrical Engineering,Lanzhou University of Technology,Lanzhou 730050,China;Lanzhou Petrochemical Company,Lanzhou 730060,China)
出处
《西安交通大学学报》
EI
CAS
CSCD
北大核心
2024年第6期24-33,共10页
Journal of Xi'an Jiaotong University
基金
国家自然科学基金资助项目(51965037,52365057)
关键词
滚珠丝杠
控制器
混沌映射
二阶振荡粒子群
ball screw
control
chaos mapping
second-order oscillatory particle swarm
