融合社会学习和莱维飞行的QPSO自平衡控制参数优化  

PARAMETER OPTIMIZATION OF SELF-BALANCING CONTROLLER BASED ON QPSO INTEGRATING SOCIAL LEARNING AND LEVY FLIGHT

作　　者：董慧芬[1] 沈鹏飞 Dong Huifen;Shen Pengfei(Control Science and Engineering,Institute of Robotics,Civil Aviation University of China,Tianjin 300300,China)

机构地区：[1]中国民航大学机器人研究所控制科学与工程,天津300300

出　　处：《计算机应用与软件》2023年第3期88-93,209,共7页Computer Applications and Software

基　　金：天津市自然科学基金项目(17JCYBJC18200)。

摘　　要：非同轴两轮自平衡车LQR控制器Q、R矩阵整定过程,QPSO存在信息共享机制单一等问题,提出融合社会学习、莱维飞行的改进QPSO。建立平衡车动力学模型,根据适应度函数采用动态线性递减LSL-QPSO进行Q、R优化,对LSL-QPSO控制进行仿真。结果表明:改进LSL-QPSO控制更加有效,相对于QPSO,起摆倾角和进动角峰值分别降低18.2%和21%,调节时间分别缩短16.6%和14.3%;抗干扰倾角和进动角的调节时间分别缩短20%和12.5%,进动角峰值降低15.2%,有效提升系统动态性能。In the process of tuning Q and R matrix of the LQR controller of a self-balancing car,the QPSO has a problem of simple information sharing mechanism.To solve this problem,an improved QPSO integrating social learning and Levy flight is proposed.The dynamic model of the balance car was established.A linear decreasing LSL-QPSO was used to design Q and R optimization based on the fitness function.The LSL-QPSO was simulated.The results show that the improved LSL-QPSO is more effective.Compared with QPSO,the peak tilt angle is reduced by 18.2%and the adjustment time is shortened by 16.6%.The precession angle peak is reduced by 21%,and the adjustment time is shortened by 14.3%.The adjustment time of anti-interference tilt angle is shortened by 20%.And the peak of precession angle is reduced by 15.2%,and the adjustment time is shortened by 12.5%.

关 键 词：陀螺平衡车 量子粒子群 线性二次型调节器 社会学习 莱维飞行 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]