神经网络技术在智慧炼厂节能方面的研究

Research on Energy Saving of Intelligent Refinery Based on Neural Network Technology

汽油加氢加热炉采用经典PID控制,未经过优化整定,炉热效率较低,回路不稳定。对加热炉控制系统进行神经网络建模,采集加热炉实时数据,通过神经网络深度学习,建立加热炉工艺模型,根据模型采用梯度下降全局搜索方式对加热炉PID控制参数进行优化整定,优化排烟温度和烟气氧含量控制,提升了加热炉热效率。对于常规PID控制系统的优化有借鉴意义。

The classic PID control is adopted in the heating furnace in the gasoline hydrogenation installation.Without optimization and tuning,the heating furnace is with low thermal efficiency and unstable loop.Neural network model is established for the control system of the heating furnace to acquine real-time data.The heating furnace process model is constructed through deep studying of neural network.According to the model,the PID control parameters of heating furnace are optimized and tuned by gradient descent global search method,and the exhaust gas temperature and oxygen content control of flue gas are optimized,and the heating furnace thermal efficiency is improved.It has a certain guiding and reference significance for optimization for existing conventional PID control system.