旋流-静态微泡浮选柱分选指标的BP神经网络预测

Separation index prediction of cyclonic-static micro-bubble flotation column based on BP neural network

采用BP神经网络模型,对不同操作条件下的分选指标进行了预测。选取捕收剂用量、起泡剂用量、循环压力和进气量作为输入因子,精煤灰分和可燃体回收率作为输出因子,建立了分选指标与操作参数的BP神经网络预测模型。结果表明:BP神经网络模型能准确预测分选指标,预测值与试验值之间误差小,精煤灰分和可燃体回收率的预测值与试验值的相对误差一般小于5%。

BP neural network model was used to predict the cleaned coal ash and combustible recovery in different operational conditions.Taking the circulation pressure,air flow,dosage of collector and frother as input set,the BP neural networkas prediction model was proposed to estimate the cleaned coal ash and combustible recovery as outputs.It is shown that BP neural network model can estimate the separation index quite satisfactorily in which the relative errors between the predicted values and experimental values are both less than 5% for the cleaned coal ash and combustible recovery.