重介质选煤过程模型与数据混合驱动的自适应运行反馈控制

重介质悬浮液密度是决定重介质选煤产品质量的重要影响因素,但由于重介质选煤运行过程是一个时变的强非线性过程,导致根据实时工况的变化在线调整重介质悬浮液密度异常困难.为此,本文针对重介质选煤过程特性,提出一种模型与数据混合驱动的自适应运行反馈控制方法,用于在线调整重介质悬浮液密度设定值.所提方法首先将重介质选煤过程分解为低阶线性模型和未建模动态非线性项两部分;进而针对线性部分,将PI控制与一步最优控制相结合,设计了模型驱动的自适应PI控制器;并利用随机向量函数链接网络设计了数据驱动的虚拟未建模动态补偿器;最后分析了闭环系统稳定性,并在基于MATLAB和Unity3D的虚拟现实仿真平台上进行了对比仿真实验,验证了所提方法的有效性.

Model-data hybrid driven adaptive operational feedback control of dense medium coal preparation process

Density of dense medium is a key factor for the quality of coal preparation products. Unfortunately, the dense medium coal preparation (DMCP) process has time varying and strongly nonlinear characteristics, which make it more difficult to adjust the density of dense medium online according to the current operation condition. To tackle this issue, this paper proposes a model-data hybrid driven adaptive operational feedback control approach for DMCP process. To adjust the set-point of density of dense medium online, the DMCP process has been first divided into two parts: the low-order linear model and nonlinear unmodeled dynamics term. For the linear model, a model-driven adaptive PI controller is developed by combining PI control method with one-step optimal control. A data-driven virtual unmodeled dynamics compensator is proposed based on a random vector function link network. The stability of closed-loop system is analyzed, and comparative simulations are conducted on MATLAB and Unity3D based virtual reality simulation platform to verify the effectiveness of proposed method.