金矿浮选回收率预测模型

Prediction model of floatation recovery ratio for a gold mine

摘要: 浮选回收率是金矿选矿过程重要的生产指标，目前主要是通过人工化验的方法检测获得，人工检测周期较长，造成金矿厂不能及时把握浮选工艺水平.在大量现场生产数据的基础上，分别采用多元线性回归和BP神经网络的方法，建立了金矿厂浮选回收率的预测模型.预测误差分析表明，BP神经网络预测模型能较好地预测金矿厂的浮选回收率，当预测相对误差在±3%范围内时，模型的预测精度达到91%，对于实际生产具有良好的参考作用.

Abstract: As an important production index in the present gold-mine beneficiation process, floatation recovery ratio is mainly ob-tained by laboratory test, which has long cycle time and is hard for the staff to control the flotation process standard. Based on massive actual production data, two prediction models of floatation recovery ratio for a gold mine were established respectively by using multiple linear regression and BP neural network method. By analyzing the predictive errors of the two models, it is approved that the prediction model based on BP neural networks can provide a better accuracy. When the relative prediction errors are within ±3%, the prediction accuracy reaches 91%, thus applying a good reference for practical production.