一种基于核ridge回归的解耦控制系统

提出了一种新的基于核ridge回归的解耦方法。该方法具有传统径向基(RBF)神经网络解耦方法对被控对象数学模型依赖性小的特点，同时又能有效地克服RBF神经网络解耦方法对训练样本要求高、噪声敏感和解耦速度慢的缺点，经核ridge解耦器补偿后的控制系统具有被调节量和调节量之间耦合作用小、动态特性好、稳定性强的优点。补偿后的控制系统具有很强的校正能力，对外界各种干扰也有较强的解耦效果和控制质量。仿真试验表明，采用核ridge解耦器的多变量控制系统能够有效地解除系统各变量之间的耦合作用，且结构简单、易于实现，大大增强了解耦控制系统的实用性能。

A Decoupling Control System Using Kernel Ridge Regression

Under the analysis of classical decoupling methods, a new decoupling method was proposed based on the kernel ridge regression. This method not only has the character of little dependence on the model of multivariable coupled systems which the conventional neural network decoupling methods have, but also overcomes their shortcomings of the need for highly quality training samples, sensitivity to noise and the long time for decoupling. The decoupled control system compensated by kernel ridge compensator has low coupling effects between controlling variables and controlled variables, good dynamical characteristics and excellent stability. The decoupled control system also has good emendation capability, decoupling effects and control capability to various disturbances. Simulation results were presented to show that the multivariable control system adopting kernel ridge compensator can decouple the coupling effects among those parameters. This method is relatively simple and easy to implement.