金精矿提金三相循环流化床系统的神经网络建模

Neural Network Modeling of System of Extracting Gold from Gold Concentrate with Three-phase Circulating Fluidized Bed

三相循环流化床床层压力是影响提金效率的一个重要因素,但床层压力大小受进气速度、固含率、金精矿粒径大小、循环流化床自身设计结构等多种因素影响,关系复杂,难以建立精确的数学模型;针对以上问题,首先通过实验确定了影响床层压力大小的主次因素,为简化问题进行建模做了准备;然后对三相循环流化床提金过程的压力特性进行多工况实验,采用基于LM(Levenberg-Mar-quardt)算法的BP神经网络建立了金精矿提金三相循环流化床床层压力的神经网络模型,并进行了校验;研究结果表明该算法收敛速度快,所建模型精度高且泛化能力强,该模型为三相循环流化床的监控提供了基础。

The stratified pressure of Three-phase Circulating Fluidized Bed（TCFB） is one important factor which affects the efficiency of extracting gold from gold concentrate.But the size of the stratified pressure is affected by many factors,such as the rate of the flowing gas,solid holdup,the design structure of TCFB,etc.And their relationship is complicated,it＇s difficult to establish accurate mathematical model.Aimed at this problem,firstly,the main factors and the secondary factors which affect the size of the stratified pressure have been determined by experiments,based on this,the problem was simplified to prepare for the modeling.Then the stratified pressure of TCFB in extracting gold from gold concentrate process was experimentally investigated,and taking advantage of BP neural network based on Levenberg-Marquardt（LM） algorithm,the neural network model is established and verified.The result showed that the LM algorithm has a rapid convergent speed,the model has high precision and good generalization ability,and the model provides a foundation to monitor TCFB.