%0 Journal Article %T 基于LS_SVM的伺服姿态调整平台反步控制<br>Backstepping Control of Attitude Adjustment Servo Platform Based on LS_SVM %A 黄晓蓉 %A 高宏力 %A 毛润 %A 李世超 %A 文娟< %A br> %A HUANG Xiaorong %A GAO Hongli %A MAO Run %A LI Shichao %A WEN Juan %J 西南交通大学学报 %D 2017 %R 10.3969/j.issn.0258-2724.2017.03.025 %X 为提高具有复合干扰问题的高精度伺服系统的控制性能，采用两个最小二乘支持向量机（least squares support vector machine，LS_SVM）在线逼近高精度伺服系统中的复合干扰，应用粒子群算法离线优化了LS_SVM的核函数参数和正则化参数；根据反步控制理论，依次递推选择Lyapunov函数，设计了基于LS_SVM的自适应反步控制器；通过Lyapunov稳定性证明了系统的稳定性.仿真结果表明：LS_SVM能对系统复合干扰部分进行有效的补偿；反步自适应控制器与经典三环PID控制器相比，在未考虑复合干扰时，系统响应时间减少了20%；考虑复合干扰时，系统稳态精度提高了34.09%，系统响应时间减少了25%；其能有效抑制系统参数变化对系统性能的影响，说明LS_SVM的自适应反步控制器具有较强的鲁棒性.<br>： To improve the performance of high-precision servo systems with compound disturbance problem, two least squares support vector machine (LS_SVM) systems were employed to approximate the compound disturbance of the high precision servo systems. The kernel functions and regularization parameters of LS_SVM were attained by particle swarm optimization (PSO) algorithm offline. An adaptive backstepping control system based on the LS_SVM method was designed by using the backstepping control theory and choosing Lyapunov function in turn. The stability of the developed system was proved through the stability of Lyapunov. Simulations show that the compound disturbance of the system could be effectively compensated by LS_SVM. Without considering the compound disturbance, the proposed controller could reduce its response time by 20% compared with the classic three-loop PID controller. However, when considering the compound disturbance, the proposed controller could reduce its response time by 25%, and increase its steady-state precision by 34.09%compared with the classic three-loop PID controller. At the same time, the proposed controller could effectively inhibit the influence on the system performance caused by the change of system parameters, so it has strong robustness %K 伺服系统 %K 反步 %K 粒子群优化 %K 最小二乘支持向量机 %K 齿隙 %K < %K br> %K servo systems %K backstepping %K particle swarm optimization %K least squares support vector machine %K backlash %U http://manu19.magtech.com.cn/Jweb_xnjd/CN/abstract/abstract12447.shtml