神经网络在高浓度充填料浆流变特性中的应用

为了探究高浓度充填料浆中各组分对充填料浆流变参数的影响,以实验室测得15组配比实验数据为样本,建立了塑性粘度、初始切应力与其主要影响因素质量浓度、胶凝剂用量、粉煤灰用量、煤矸石用量之间的BP神经网络模型,通过各影响因素的权重值将料浆配比进行了优化,并以唐安矿的充填料浆配比为例加以验证。结果表明：建立模型的最大预测相对误差为-3.209%；矸石含量对流变参数的影响最大,胶凝剂含量对流变参数的影响最小；料浆浓度对塑性粘度的影响较大,而粉煤灰对初始切应力的影响较大；当浓度一定时,粉煤灰和矸石的比例应该尽量大一些；按照料浆配比的优化方案,唐安矿料浆输送达到了预期的目标。

Application of Neural Network to Rheological Character of High Concentration Filling Slurry

In order to explore the effect of material components on rheological parameters in high concentration filling slurry, the BP network model based on the 15 groups results of ratio test was established. The model established the relationship between plastic viscosity, initial shear stress and mainly effect factors including mass concentration, dosage of gelling agent, dosage of fly ash, coal gangue dosage. The optimal ratio of filling slurry was obtained by comparing the weighted values of effect factors. The ratio of filling slurry in TangAn Mine was taken to verify. The result was that, the maximum prediction relative errors is -3.209%; The coal gangue dosage is the most impact factor and the dosage of gelling agent has little influence on rheological parameters; The effect of mass concentration on plastic viscosity is greater than on plastic viscosity and the fly ash