乙烯装置蒸汽管网用能配置与实时优化

摘要

The polyethylene unit steam piping network system has a complex structure. It has various kinds of steam equipment, the parameters for the model are complicated and its energy consumption and configuration depend on experience. This paper puts forward the support vector machine as the soft sensor model to establish the accurate non-linear foundation model for the optimization of the steam piping network. A steam piping network model from the viewpoint of energy consumption was constructed based on material balance, energy conservation, and product quality. This model was a problem of mixed integer nonlinear programming ( MINLP ) calculated with adaptive mixed coding differential evolution (AMCDE). The polyethylene unit steam network of a petrochemical plant was simulated with the model and optimized.%乙烯装置蒸汽管网系统结构复杂,用汽设备多,模型参数复杂多变,其能量配置多依赖经验给出.针对运行过程中管网模型参数的变化,利用支持向量机建立了关键设备的透平蒸汽流量软测量模型,从而为蒸汽管网优化问题提供准确的非线性模型基础.根据物料平衡、能量守恒和产品质量要求建立了能耗视角下的蒸汽管网系统模型,该模型本质上属于混合整数非线性规划问题(MINLP).采用自适应混合编码差分进化算法(AMCDE)对模型进行求解.最后以某石化厂蒸汽管网系统为研究实例进行仿真,得到了令人满意的优化结果.