基于水泥细度软测量的优化控制

Optimal control based on cement fineness soft measurement

为解决水泥联合粉磨系统细度检验延时、时间滞后性长造成优化控制难的问题,本文提出了一种基于径向基神经网络(RBF-NN)和模型预测控制(MPC)算法的细度软测量及在线优化控制方法。首先,提取水泥联合粉磨系统中的相关特征变量并完成数据清洗,引入K-Means聚类算法优化径向基神经网络的参数选取方法,构建了径向基神经网络的水泥细度软测量模型,通过对模型周期性的调度,实现了细度的在线预测;其次,基于水泥细度预测结果和MPC算法,构建在线质量控制回路对水泥细度进行实时在线优化控制;最终,以某水泥厂的联合粉磨系统为例,验证了该方法的有效性。

To solve the problem of difficult optimization control caused by the long-time lag of fineness test delay and long-time lag in the cement combined grinding system,this paper proposed fineness soft measurement and online fineness optimization control method based on radial basis neural network(RBF-NN)and model predictive control(MPC)algorithm.First,relevant characteristic variables in the cement combined grinding system were extracted and data cleaning was completed.The K-Means clustering algorithm was introduced to optimize the parameter selection method of the radial basis neural network,and a cement fineness soft measurement model of the radial basis neural network was constructed.Through periodic scheduling of the model,online prediction of fineness was achieved.Secondly,an online quality control loop was constructed to perform real-time online optimal control of cement fineness based on the cement fineness prediction results and MPC algorithm.Finally,the effectiveness of this method was verified based on a certain cement taking the factory's combined grinding system as an example.