论文部分内容阅读
针对隧道工程塌方信息的缺失,基于粗糙集理论中的扩充差异矩阵和RSDIDA补齐算法,创立了隧道塌方案例影响因素丢失信息的修复技术。利用广义回归神经网络理论,构建了隧道塌方量预测的广义回归神经网络模型。将塌方信息修复技术与神经网络模型结合起来,构建起隧道塌方量预测的初始系统。收集到120多个公路隧道塌方案例,对案例有关缺失信息进行修复,构成完整的样本库。初始系统通过样本库训练,进一步发展为成熟的隧道塌方量预测系统。通过工程使用及对比,该系统预测结果体现出较好的准确性和效率,具有工程实用价值。
Aiming at the lack of information of landslide in tunnel engineering, a repair technique based on the extended difference matrix and RSDIDA in rough set theory is proposed to repair the information of tunnel collapse. Using generalized regression neural network theory, a generalized regression neural network model for predicting the amount of tunnel collapse is constructed. The collapse information restoration technology and neural network model are combined to build the initial system of tunnel collapse prediction. Collected more than 120 road tunnel collapse case, the case of missing information to be repaired to form a complete sample library. The initial system through the sample base training, and further developed into a mature tunnel collapse forecasting system. Through the use and comparison of the project, the forecast result of the system shows good accuracy and efficiency and has engineering practical value.