Research On Control Scheme Of Fluidized Bed Polyethylene Process And Development Of Parameter Tuning Software

Polyethylene is an important material in daily production and life.In recent years,China ’s polyethylene industry has developed rapidly and the market competition is fierce.However,it faces the problems of serious product homogeneity,heavy dependence on imports of high-end products,technology dependence on introduction and weak innovation ability.Therefore,it is imperative to develop new high-end polyolefin technology and improve the operation level of the plant to achieve high quality,high efficiency and low-cost production.The gas-liquid fluidized bed polyethylene process adjusts the reaction zone with different temperature difference and concentration difference by introducing condensate at different positions on the side wall of the fluidized bed reactor,so as to produce high-end polyethylene products.The improvement of the process changed the product properties of the polymer and increased the complexity of the reaction system.In the industrial production of gasliquid fluidized bed polyethylene process,it is observed that the reactor temperature fluctuates greatly and the stability of the reactor deteriorates when some brands are switched and the reactor is started and stopped.This not only increases the production cost,but also brings security risks.From the perspective of process control,the cause of temperature fluctuation may be unreasonable control scheme and inappropriate PID parameters.Therefore,this paper focuses on the temperature control of polyethylene fluidized bed,carries out the research on the temperature control scheme design of fluidized bed reactor and the development of PID parameter tuning software,and explores the application of reinforcement learning algorithm in the temperature control of fluidized bed reactor.The main research contents and conclusions are as follows:(1)Four temperature control schemes of fluidized bed polyethylene process are designed and systematically evaluated and compared.Firstly,the steady-state model of fluidized bed polyethylene process is established.The steady-state model is verified by comparing the main equipment operating parameters,reactor temperature distribution and product quality index with industrial data,which can be used as a dynamic basis.Secondly,a complete control scheme is designed for the process by heuristic control design method.The dynamic model is verified by the dynamic response data of the heat exchange system and the step response data of the fluidized bed reactor temperature from the industrial device.The dynamic model can be used as a simulation platform for further temperature control scheme design and comparison.Finally,four control schemes are proposed,including direct control,reactor temperature-temperature regulating water temperature double loop cascade,reactor temperature-recycle gas temperature double loop cascade and reactor temperature-circulating gas temperaturetemperature regulating water temperature three loop cascade.The set point tracking and disturbance rejection capabilities of the four schemes are evaluated by seven simulation experimental systems.The results show that the reactor temperature-water temperature cascade control is slightly better than the other three schemes in terms of set point tracking performance.In terms of disturbance rejection,for the disturbances from the circulating gas subsystem,the reactor temperature-temperature regulating water temperature cascade control and direct control performance are better than the other two schemes;for the disturbances from the heat exchange system,the reactor temperature-temperature regulating water temperature cascade control and the reactor temperature-circulating gas temperature-temperature regulating water temperature three cascade control performance are better than the other two control schemes.In summary,the temperature control scheme of fluidized bed reactor should adopt reactor temperature-temperature regulating water temperature cascade control.The research results provide theoretical guidance for the design and transformation of temperature control scheme of fluidized bed polymerization process.(2)PID parameter tuning software with three functional modules of open-loop identification,closed-loop identification and parameter tuning is developed and applied to the parameter tuning of fluidized bed reactor temperature controller.Based on the first-order and second-order pure lag models commonly used in chemical processes,the PID parameter tuning algorithm with three modules of open-loop identification,closed-loop identification and parameter tuning is developed by continuous time integration method,frequency response method and internal model control principle respectively.The software user interface is designed.Multiple simulation cases show the effectiveness and accuracy of the software in tuning PID parameters under openloop and closed-loop identification conditions.The software is applied to the temperature closed-loop step experiment of industrial fluidized bed polyethylene reactor.The mean square errors of the two closed-loop identifications are 0.009 and 0.002,respectively.The process model can be accurately identified and the PID parameters with different response speeds can be calculated,which provides reference values for parameter setting and retuning of industrial PID controller.(3)A temperature control scheme of fluidized bed reactor based on deep reinforcement learning(DRL)is proposed.Based on the characteristics of nonlinearity and frequent disturbance of heat exchange system,DRL controller is designed,and DRL controller is obtained by step-by-step training strategy.The results show that compared with the traditional PID controller,the DRL controller has excellent set point tracking performance and disturbance rejection in the nonlinear process.The integral absolute error(IAE)value of the simulation experiment under various working conditions is much smaller than that of the PID controller.Compared with the industrial traditional PID-PID cascade control,the proposed PID-DRL cascade control has better set point tracking and disturbance rejection performance,and the IAE value of the simulation experiments is smaller.The main loop PID controller of the PID-DRL cascade control stabilizes the reactor temperature,and the response signal generated by various disturbances is transmitted to the secondary loop;the secondary loop DRL controller responds quickly to the main loop and suppresses various disturbances to the reactor temperature.The proposed compound cascade control structure provides a new idea for the temperature control design of polyethylene fluidized bed.