【摘要】 针对非线性预测控制中的预测模型,设计了稀疏在线无偏置最小二乘支持向量机(SONB-LSSVM),并提出了基于SONB-LSSVM的有约束单步预测控制算法。在每个控制周期,该SONB-LSSVM递推地学习新样本,并删除贡献最小样本。该样本删除技巧能提高学习样本集的多样性和代表性;与ONB-LSSVM相比,SONB-LSSVM的泛化性能受输入信号频率影响较小。控制量由Brent优化方法计算。由于SONB-LSSVM能及时学习过程动态新特性,该预测控制方法具有良好的自适应能力.液位控制仿真表明,在多种波形的期望输出并有扰动情况下该预测控制方法都是有效的。更多还原

【Abstract】 Aiming at predicting model of nonlinear predictive control,a sparse online non-bias least square support vector machine(SONB-LSSVM) is designed,and a constrained single-step-ahead predictive control(PC) is proposed utiliz-ing SONB-LSSVM.During per controlling period,the SONB-LSSVM studies new sample and removes the least important one recursively.The skill for deleting sample can improve diversity and representative capacity of the training sample set;generalization of SONB-LSSVM is less affected by the input signal frequency compared with ONB-LSSVM.The control values are computed via Brent optimization method.Because SONB-LSSVM can study new dynamic properties of process in time,the predictive control strategy possesses excellent adaptation.Simulation results of liquid-level process control show the validity of the predictive control in various waveform expected output case with disturbance existing.更多还原