全尾砂絮凝沉降参数预测模型研究

为了得到最佳的絮凝沉降参数,运用BP神经网络和遗传学算法建立了全尾砂絮凝沉降参数预测模型.以絮凝剂单耗和尾砂浓度作为输入因子,以沉降速度作为输出因子;通过正交试验,确定网络学习、训练样本,建立神经网路预测模型;采用遗传算法对全尾砂沉降参数预测模型进行全局寻优,得到最佳絮凝沉降参数.将预测模型运用到和睦山铁矿,在絮凝剂单耗12 g/t,尾砂浓度17%条件下,沉降速度达到1.31 m/h,满足生产需要,比原生产所需絮凝剂单耗减少20%.应用结果表明,该预测模型有较高的实用性,为沉降参数优选提供了一种崭新的思路.

Study on the Parameters Prediction Model of Flocculating Sedimentation of Crude Tailings

In order to get the optimum parameters of flocculating sedimentation， back propagation neural network and genetic algorithm were applied to establish the flocculation sedimentation parameters prediction model of the crude tailings. The flocculating agent and tailings concentration consumption were used as the input data and the sedimentation speed was confirmed to be the output data. The learning and training samples were received by the orthogonal experiments to build neural network prediction model. Then， the optimum parameters of flocculating sedimentation were received after using genetic algorithm finding optimal in parameters prediction model of the crude tailings. The selected parameters prediction model was used in Hemushan iron mine. The results showed that the flocculating agent consumption is 12g/t and tailings concentration is 17%， the sedimentation speed is 1.31m/h， these parameters meet the production requirements. The flocculating agent consumption required is 20% less than the original production when using these flocculating sedimentation parameters. The application of the model indicates that it provides a new method to optimize the flocculating sedimentation parameters with a good effect.