充填膏体流变参数优化预测模型

摘要

In order to predict the backfill paste rheological parameters more accurately, the prediction model was established based on the principal component analysis and the improved BP neural network. By doing the backfill paste mix proportion experimental results in a metal mine, factors as backfill paste mass fraction, sand-cement ratio in mass, slurry weight, collapsed slump, and etc influencing backfill paste rheological parameters were firstly processed by the method of principal component analysis. The main ingredients were obtained. Rheological parameters were then predicted accurately by the improved BP neural network model. The results show that model of backfill paste rheological parameter prediction relative errors of predicting outcomes are all controlled within 5％,and compared with the prediction errors by BP neural network without principal components analysis, the relative errors of yield stress are reduced by 0.48％-7.29％, and relative errors of viscosity are reduced by 1.67％-6.20％, which shows that the model of backfill paste rheological parameter prediction is reasonable and effective, the prediction precision of yield stress and viscosity are significantly improved. It provides a new method to the prediction of backfill paste rheological parameters.%为了更精确地对充填膏体流变参数进行优化预测,建立主成分分析法(PCA)和改进的BP神经网络(I-BPNN)相结合的优化预测模型.以某金属矿山充填膏体配比实验为基础,利用主成分分析法对充填膏体流变参数影响因素(膏体质量分数、砂灰质量比、料浆容重和坍落度等)进行预处理,得出主成分,再利用改进BP神经网络模型进行预测,最终得到更准确的充填膏体流变参数预测结果.研究结果表明:该模型对充填膏体屈服应力、黏度等流变参数优化预测的相对误差都控制在5％以内,较未经主成分分析的BP神经网络预测结果,经主成分分析后,屈服应力预测相对误差降低0.48％～7.29％和黏度相对误差降低1.67％～6.20％,表明该模型对充填膏体流变参数预测是合理、有效的,屈服应力与黏度的预测精度显著提高,为充填膏体流变参数优化预测提供了一种新思路.