CAJ | 학술논문

在传统动态矩阵控制算法中，通过当前预测值和实际值的误差来对未来时刻的预测输出进行校正，这种方法在一定程度上可以克服扰动、噪声的影响，但是对模型失配问题仍有一些局限性。最小二乘相关向量回归（LS—RVR）是在相关向量回归的基础上发展起来的，相对于相关向量回归具有更好的回归性能。提出了一种基于最小二乘相关向量回归误差补偿的动态矩阵控制算法，并进行了仿真实验。仿真研究结果表明，该算法具有良好的控制性能。
재전통동태구진공제산법중，통과당전예측치화실제치적오차래대미래시각적예측수출진행교정，저충방법재일정정도상가이극복우동、조성적영향，단시대모형실배문제잉유일사국한성。최소이승상관향량회귀（LS—RVR）시재상관향량회귀적기출상발전기래적，상대우상관향량회귀구유경호적회귀성능。제출료일충기우최소이승상관향량회귀오차보상적동태구진공제산법，병진행료방진실험。방진연구결과표명，해산법구유량호적공제성능。
In traditional DMC, the predictive output for coming future is compensated by the error between current predictive value and actual value. This calculation method can overcome the influence from disturbance and noise. It still has some limitations on rnodel mismatch problem. The LS-RVR （Least Squares Relevance Vector Regression） is developed on the base of RVR （Relevance Vector Regression）, and has a better regression ability. The DMC based on LS- RVR error compensation is proposed and simulated. The simulation result indicates that this calculation method has better control performance.