电熔镁砂产品单吨能耗混合预报模型  

A Hybrid Prediction Model of Energy Consumption Per Ton for Fused Magnesia

作　　者：吴志伟[1] 柴天佑[1,2] 吴永建[1] 

机构地区：[1]东北大学流程王业国家重点实验室沈阳110819 [2]东北大学自动化研究中心沈阳110819

出　　处：《自动化学报》2013年第12期2002-2011,共10页Acta Automatica Sinica

基　　金：国家自然科学基金(61020106003,61004009);国家重点基础研究发展计划(973项目)(2009CB320601)资助~~

摘　　要：产品的单吨能耗是反映电熔镁砂熔炼过程产品产量和能耗的综合生产指标.通过分析炉内电热转换关系,利用能量守恒原理建立了产品单吨能耗模型.针对模型的未知非线性和参数时变等综合复杂性提出了由基于机理分析的单吨能耗主模型和基于神经网络的补偿模型组成的产品单吨能耗混合预报模型.其中神经网络补偿模型用于补偿模型的未知非线性和参数不确定性对于预报模型准确性的影响.采用某电熔镁砂熔炼过程实测数据验证了所建立的混合预报模型是有效的.Energy consumption per ton is a main index of product quality and energy consumption in the smelting process of fused magnesia. By analyzing the electric heating conversion principle, an energy consumption per ton model of product is proposed by using the law of conservation of energy. For the complex characteristics such as unknown nonlinear and time-varying parameters of the energy consumption per ton model, a hybrid prediction model of energy consumption per ton which consists of a main model based on mechanism analysis and a compensation mode based on neural network is also proposed. The compensation model compensates the modeling errors caused by unknown nonlinearity and uncertainty of some parameters for the mathematical model. Industrial applications show the usefulness and effectiveness of the proposed hybrid prediction model for energy consumption per ton.

关 键 词：电熔镁砂 单吨能耗 预报模型 能量守恒 神经网络 

分 类 号：TF065.1[冶金工程—冶金物理化学]