基于LS-SVM预报的动态矩阵预测控制

针对实际工业过程中的非线性及时变特性,传统预测控制算法就难于建立精确的数学模型,从而提出了一种基于最小二乘支持向量机预报的动态矩阵预测控制模型.在整个过程中,首先建立基于最小二乘支持向量机的非线性动态矩阵预测控制结构,通过利用最小二乘支持向量机辨识被控对象模型,同时预测对象的未来输出,然后用动态矩阵控制算法进行滚动优化和反馈校正.仿真实例表明该模型对预测结果有很好的控制作用,有效消除输入干扰的影响,从而提高了预测精度.

Dynamic Matrix Predictive Control Based on LS-SVM Prediction

It is difficult to build an accurate mathematic model using the traditional predictive control algorithm in practical industry process because of its non-linear and time variable characteristics. A DMPC based on least squares support vector machine (LS-SVM) prediction model is proposed. In the whole course, the structure of the non-linear dynamic matrix predictive control based on LS-SVM is built first, LS-SVM is employed to identify the controlled plant and to produce predictive output, then the DMC algorithm is used to recede horizon optimization and revise feedback. Simulation example demonstrates that the model has a good control effect on the predictive results and eliminates the disturbance of inputs effectively thus improving the precision of prediction.