分段线性化多模型动态矩阵控制算法的设计

The DMC strategy of piecewise linear multi-model system

线性控制算法具有计算简单、实时性好等优点,但是工业过程通常具有复杂的非线性特性,而且非线性程度越强,实际过程和模型之间的误差越来越大,常规的线性控制器性能会随着非线性程度的增强而显著降低。对于一般的非线性控制策略而言,它们具有建模成本较高、难以选取合适的权重因子等难点,本文采用分段线性化的方法,将非线性系统过程分为多个分段的线性化模型,将动态矩阵控制策略推广到各个分段线性模型,设计了分段线性化多模型DMC算法,通过组合以实现其预测控制的目的。为了验证该算法的有效性,本文在Matlab/Simulink软件平台上进行了1套强非线性pH值中和过程实验装置的较全面的控制仿真,包括阶跃及方波给定值跟踪、抗强干扰,以及存在较大模型失配时的给定值跟踪性能仿真。仿真结果表明,针对强非线性的过程,分段线性化的DMC比不分段单模型的DMC过程具有更好的控制性能。

Linear control algorithms have some advantages,such as simple calculated,good real-time,and so on.But industrial processes usually have complicated non-linear characteristics and the stronger the degree of nonlinearity,the bigger the deviation between the model and actual process. The performance of conventional linear controller will be significantly stepped down with enhanced nonlinearity of the controlled process.For the general nonlinear control strategy,there are some troubles also,such as high modeling cost,difficult to select appropriate weighting factors for the control algorithms and etc.Based on the idea of piecewise linearization,the nonlinear system process is divided into several of piecewise linear models.Then the Dynamic Matrix Control（DMC） strategy is extended to every piecewise linear model to design the DMC strategy of piecewise linear multi-model system.The predictive control of nonlinear process is implemented through the control of combination of multiple piecewise linear models.In order to verify the effectiveness of the multi-model DMC algorithm,a comprehensive simulation of a strong nonlinear pH experimental process is carried out based on Matlab/Simulink software platform,including the tracking of step and square-wave set points,dealing with strong disturbance,and the simulation of set points tracking performance with large model mismatch.The simulation control results show that this multi-model DMC strategy gives better control performance than the DMC with single linear model.