改进PID在反应釜温度控制系统中的应用研究

Application of Improved PID in Reactor Temperature Control System

反应釜炉温控制是化工生产过程中主要的控制系统之一,其温度控制具有大滞后、时变、非线性等特点。针对常规PID控制效果不佳的缺点,提出一种改进的模糊RBF神经网络智能控制方法。将系统的输入误差及误差变化率进行模糊化,并利用RBF神经网络算法对PID控制参数进行在线学习、运算和整定。在RBF神经网络控制算法中,设定初始权值在一定范围内服从高斯分布和均匀分布,对权值不断优化,使得反应釜温度达到良好的控制效果。经Matlab仿真验证,结果表明和常规PID相比,该方法提高了系统的控制精度并具有较强的鲁棒性。

The reactor temperature control is one of the main control systems in the chemical production process. Its temperature control has the characteristics of large hysteresis, time variation and nonlinearity. To overcome the shortcomings of the conventional PID control, an improved fuzzy RBF neural network intelligent control method was proposed. The input error and error rate change of the system were fuzzified, and the RBF neural network algorithm was used to on-line learn, calculate and adjust the PID control parameters. In the RBF neural network control algorithm, the initial weight was set in a certain range by Gauss distribution and uniform distribution, and the weight was optimized continuously, so that the temperature of the reactored reached a good control effect. The Matlab simulation results show that the proposed method improves the control precision of the system and has strong robustness.