摘要
为促进人工冻结技术在地下工程地基处理中的推广应用,在综合分析人工冻土融沉系数影响因素的基础上,采用BP人工神经网络方法建立人工冻土融沉系数的预测模型。用南京地区典型土质淤泥质黏土、粉质黏土和粉砂的试验数据作为网络模型的学习训练样本和测试样本,对网络模型的预测结果与实测进行对比。结果表明,用人工神经网络方法预测人工冻土融沉系数,结果准确可靠,更接近于实际,是一种很好的预测人工冻土融沉系数的方法。
In order to promote the application of the artificial freezing technology in the foundation treatment of underground engineering, on the basis of analysis of the main faclors influencing the thaw settlement coefficient of artificial freezing soil, models to predict the thaw settlement coefficient of artificial freezing soil were, established by applying the theory of BP artificial neural network （ANN）. A large amount of test data from three kinds of soil of Nanjing region （silt clay, mealy clay and mealy sand） was used as learning and training samples to train and test ANN models and the calculated results of the ANN models and the test values were compared and analyzed, which showed that it was comparatively precise to predict the thaw settlement coefficient of artificial freezing soil by ANN technology.
出处
《森林工程》
2008年第5期18-21,共4页
Forest Engineering
基金
江苏省"六大人才高峰"培养计划资助项目(2004)
关键词
人工冻土
融沉系数
神经网络
预测方法
artificial freezing soil
thaw settlement coefficient
＇artificial neural networks
prediction method
