运用Bayes网络研究不同因素对磷酸镁水泥凝结时间的影响

Factors Influencing Setting Time of Magnesium Phosphate Cement by Bayesian Network

具有快硬早强特点的磷酸镁水泥凝结时间受多个因素的影响。采用概率方法统计分析硼砂缓凝剂掺量、死烧MgO粉比表面积、温度及水固比这4个因素对磷酸镁水泥凝结时间的影响。首先通过回归法建立相关模型关系,然后从试验数据和不同因素的条件概率出发,构造Bayes网络,将影响因素的模型进行误差分析。结果表明:在所研究的因素中,死烧MgO粉比表面积对磷酸镁水泥凝结时间的影响最明显,其他几种因素的影响程度按缓凝剂掺量、环境温度和水固比依次递减,各模型与试验结果偏差最大的是水灰比,其次为温度、硼砂掺量和死烧MgO粉比表面积。

There are some factors affecting the setting time of the rapid hardening magnesmm pnospnat~ ~ al. Based on some previous experimental results, this paper investigated the effects of factors （i. e. , borax content, specific sur- face area of MgO, temperature and water-to-solid ratio） on the setting time of M_PC using a probability approach. The re- gressed relationship between each factor and setting time was established. The Bayesian network was used to anatyze the condi tional probability of the factors. The quantitative effect degrees and error analysis of all the factors were obtained through the Bayesian network. The results show that among these factors, the specific surface area of magnesium oxide has the most signif- icant effect on the setting time, then the borax content, temperature and the water-to-solid ratio. The water-to-solid ratio has the greatest error between the model and the experimental data, then the environmental temperature, the borax content and the specific surface area of magnesium oxide.