电熔镁砂产品单吨能耗混合预报模型

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.

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

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.