基于递归神经网络模型预测控制的模型平稳切换

不同生产条件下的控制系统可视多模型控制系统，但多模型控制在模型切换时会引起系统的瞬态响应。采用递归神经网络建立系统的多个模型，基于模型预测控制进行控制模型切换，克服了模型切换时引起的系统瞬态响应，实现系统的平稳切换。并通过仿真表明这种切换策略明显改善了模型切换过程的动态性能。

Smooth model switching scheme based on recurrent neural network model predictive control

The control system of differ product conditions could be regarded as a multi-model control system. But model switching will result in transient response of the plant controlled. A method to establish multiples models of system using recurrent neural network and complete switch process based on model predictive control was proposed. Which avoid transient response of switch process and achieve model switching smoothly. Simulation performances indicate that the model switching scheme improve dynamic quality of switch process greatly.