连续搅拌釜式反应器（Continuous Stirred Tank Reactor,CSTR）是化工生产中进行物理变化和化学变化的重要反应装置。反应温度是CSTR生产过程最关键的工艺参数，反应温度的控制品质直接影响到产物质量和生产效率，在过程控制领域中CSTR反应温度优化控制研究一直受到广泛关注。随着单釜产能持续提升、强放热等特殊化工产业的快速发展，配有夹套与盘管双冷却装置的CSTR数量持续增加，双冷却结构为CSTR反应温度控制优化提供了条件。

本文主要研究如何利用夹套与盘管冷却装置实现CSTR聚合反应温度优化控制。首先，在双冷却CSTR聚合生产工艺流程、工作原理等分析的基础上，利用机理法建立了CSTR反应温度控制冷却过程的数学模型。为了充分发挥盘管与夹套冷却装置的互补优势，弱化二者单独冷却时的缺点，本文提出了双重控制策略，该控制策略一方面利用盘管冷却装置较快的响应速度，迅速消除动态偏差；另一方面，在反应平稳进行时，由冷却效率高的夹套冷却装置承担主要冷却负荷，减少冷却水消耗，使被控过程在动态响应和静态性能都获得较为理想的控制品质。最后，为了进一步改善双冷却CSTR反应温度双重控制系统的动态性能，利用非线性逼近能力很强的BP神经网络控制算法对主控制器进行优化，通过BP神经网络PID控制器在线对控制参数进行调整，加快控制系统收敛速度，提高CSTR反应温度控制过程自适应能力。

理论分析和仿真实验结果表明，采用基于BP神经网络控制的双冷却CSTR反应温度双重控制系统充分发挥了夹套与盘管冷却装置的冷却特性，实现了反应温度控制系统调节时间短、超调量小、抗干扰性强的控制效果，达到了节能降耗的目的，对提高CSTR放热反应过程稳定性和安全性具有重要理论意义和工业应用价值。

Continuous stirred tank reactor (CSTR) is an important reaction device for physical and chemical changes in chemical production. The most important parameter in the CSTR system is the reaction temperature, the control quality directly affects the quality of the product and the safety of the production process. With the rapid development of special chemical industries such as strong heat release and the continuous increase of single kettle capacity, the number of dual-cooling CSTRs equipped with jacket and coil has been increasing, the dual cooling structure provide conditions for the optimization of CSTR polymerization temperature control.

This article mainly studies how to use the jacket and coil cooling device to realize the optimal control of the reaction temperature of the CSTR system. First, the mathematical model of the CSTR reaction temperature control cooling process was established by the mechanism method, which are based on the analysis of the process flow and working principle of the dual-cooling CSTR. In order to give full play to the complementary advantages of the coil and jacket cooling device, weaken the disadvantages of them cooling alone. On the one hand, this control strategy uses the faster response speed of the coil cooling device to quickly eliminate the dynamic deviation. On the other hand, the jacket cooling device with high cooling efficiency bears the main cooling load when the reaction proceeds smoothly, which reduces the cooling water consumption. It can make the controlled process obtain better control quality in both dynamic response and static performance.Finally, to further improve the dynamic performance of the dual-cooling CSTR reaction temperature dual control system, The BP neural network control algorithm with strong nonlinear approximation ability is used to optimize the main controller, and the BP neural network PID controller is used to adjust the control parameters online to speed up the convergence of the control system and improve the adaptive ability of the reaction temperature control process.

Theoretical analysis and simulation experiment results show that the dual-cooling CSTR reaction temperature dual control system based on BP neural network control has fully utilized the cooling characteristics of the jacket and coil cooling device, which realizes the reaction temperature control system with short adjustment time. And the control effect of small volume adjustment and strong anti-interference achieves the purpose of energy saving and consumption reduction, which has an important theoretical significance and industrial application value for improving the stability and safety of the CSTR exothermic reaction process.