以实际的具有多变量、强耦合、强非线性特性的氧化铝强制循环蒸发系统为例,将一类多变量非线性系统分为线性模型和非线性未建模动态两部分,提出了由线性预测解耦控制器,神经网络非线性预测解耦控制器和切换机制组成的多模型预测解耦控制方法,并且证明了闭环系统的稳定性.最后,通过对强制循环蒸发系统的仿真实验,验证了所提方法的有效性. Taking an actual forced circulation evaporation system with multiple variables, strong coupling and strong nonlinearity as an example, a class of multivariable nonlinear systems is divided into linear model and non-linear unmodeled dynamics. Predictive decoupling controller, neural network nonlinear predictive decoupling controller and switching mechanism, and proves the stability of the closed-loop system.Finally, through the simulation experiment of forced circulation evaporating system, we verify The effectiveness of the proposed method.