基于神经网络模型的计量泵远程流量广义预测控制

Remote flow control of metering pump based on neural network and generalized predictive control

为了在流程工业中实现流体物料的远程投加,提出了一种隔膜计量泵流量远程控制方案。方案采用上层优化节点、双信道通信网络和本地控制器的分级控制策略,利用神经网络辨识计量泵流量和电机转速的动态模型,设计了基于自适应广义预测控制(GPC)的流量调节算法。通过对比仿真和实际平台验证了所提出方案的可行性和有效性。

A remote control scheme for diaphragm metering pump flow is proposed to realize the remote dosing of fluid materials in the process industry.In this scheme a hierarchical optimization control strategy involving the upper optimization node, dual channel communication network and local controller is utilized, the dynamic model of the metering pump flow and themotor speed are identified using neural networks, and a flow regulation algorithm based on adaptive generalized predictive control (GPC) is designed. The feasibility and effectiveness of the proposed scheme is illustrated by both numerical and practical examples.