Adaptive PID Algorithm Based on Online Robust LSSVM
XU Gui-yun
LIU Xiao-ping
LIU Yun-kai
ZHANG Xiao-guang
Abstract：An adaptive PID algorithm based on online robust least square support vector machines(LSSVM)was presented to solve the problem of large calculation resulted from the online least support vector machines updating the model in every sampling period.A two-step weights principle was used to improve robustness in LSSVM regression.The predictive error and prior process knowledge were combined to control model complexity.The precision and speed of the LSSVM were improved effectively.Based on the idea of predictive control,the online robust LSSVM modeling algorithm was applied to PID.The results show that the robust cost function combining robust Huber function with ε-insensitive loss function can be used to model for the partial nonlinear region of the system and adaptively identify the model of system with the changing of working point.This algorithm adapts to the control of time-varying parameters object and achieves higher control precision,stronger robustness and modeling speed.
Keywords：Robust least square support vector machinesonline learningmodel complexityPID
Publication Date：2010-01-01
Online Publishing Date：2025-08-15(First online date of this platform, not the publication date of the document)
Pages：7( 190-195,207 )