论文部分内容阅读
应用BP神经网络方法,研究高粘聚阴离子纤维素、羧甲基纤维素、磺化酚醛树脂以及磺化褐煤树脂四种常用处理剂对蒙脱土悬浮液表观粘度与塑性粘度的影响。采用弹性算法和最优停止法对神经网络进行优化,避免了过拟合现象,提高了神经网络的训练速度和泛化能力。以实际数据作为验证样本对神经网络模型进行检验,模型计算结果与实际结果比较,表观粘度的最大绝对误差率为8.75%,平均绝对误差率为1.12%;塑性粘度的最大绝对误差率为8.82%,平均绝对误差率为1.42%。最后运用该模型分析了单一处理剂对蒙脱土悬浮液流变参数的影响。
The effects of high viscosity polyanionic cellulose, carboxymethyl cellulose, sulfonated phenolic resin and sulfonated lignite resin on the apparent viscosity and plastic viscosity of montmorillonite suspension were studied by BP neural network. The neural network is optimized by the elastic algorithm and the optimal stopping method to avoid the over-fitting phenomenon and improve the training speed and generalization ability of the neural network. The actual data as the verification sample to test the neural network model, the model calculation results compared with the actual results, the apparent maximum apparent error of the absolute error rate of 8.75%, the average absolute error rate of 1.12%; plastic viscosity of the maximum absolute error rate of 8.82 %, The average absolute error rate of 1.42%. Finally, the model was used to analyze the influence of single treatment agent on the rheological parameters of montmorillonite suspension.