Research Into Adaptive Multi-Dimensional Taylor Network Control Of Cement Calciner Outlet Temperature

Cement is widely used in the transformation or newly-built of various types of building facilities,and its role cannot be replaced in terms of universality and applicability.Cement calciner is a key high-temperature thermal equipment that undertakes the decomposition of raw materials in NSP cement production line.Its structural mechanism and physical and chemical reactions are complex,with multivariable,pure time delay,strong coupling,uncertainty and nonlinearity.Traditional manual operation control or PID control makes it difficult to achieve strict control of calciner outlet temperature.Therefore,in order to improve cement quality,save energy,increase output,reduce environmental pollution,achieve green transformation,and promote the harmony between ecological civilization and industrial civilization in the cement industry,in-depth research and innovation attempts are needed for outlet temperature control strategy.In this thesis,a control method of cement calciner outlet temperature based on adaptive multi-dimensional Taylor network(adaptive MTN)control is proposed,and this method is compared with the PID optimal control and adaptive BP neural network PID control.Their differences in control performance are given based on the comprehensive analysis of simulation results.The main work of this thesis is listed below:1.The trend of Chinese cement production in recent years is described.And introduce the development and process flow of NSP cement production and cement calciner,giving prominence to the importance of research.2.Establish a mathematical model of the cement calciner outlet temperature.Based on the actual subject conditions,the modeling analysis is carried out to determine the main influencing factors of the cement calciner temperature and the mathematical model structure is designed as a multi-input single-output data-driven form with time-delay.The cross-correlation method is used to identify the time-delay parameters of the model,and then the residual model parameters are identified based on the recursive least squares method.The prediction accuracy of the proposed combination strategy is verified.3.Design Controllers.The control methods of this thesis adopt PID optimal control based on improved simplex method,adaptive BP neural network PID control and adaptive MTN controll.In order to improve the real-time performance of the PID parameters,it is combined with the BP neural network algorithm.The algorithm principle and structure form of adaptive MTN control are derived and discussed,and the controller designed can update the network weight coefficient online and optimize the control performance.4.Simulation system design.A simulation system for the cement calciner outlet temperature control is established,and the simulation experiments of given temperature values and under interference and model parameters changing are carried out.Based on MATLAB GUI,the graphical user interface of the simulation platform is designed to realize the integration of each simulation contrast experiment and user-friendly interaction.The performance of each controller is compared and analyzed.The simulation results show that in most cases,the adaptive MTN controller has the best adaptability,the strongest anti-interference ability and the most outstanding robustness,adaptive BP neural network PID controller is the second,and PID optimal control is the worst.