论文部分内容阅读
基坑工程由于受多种因素的影响,目前已成为岩土工程中的重点和难点。在基坑工程施工中,需要根据现场实际情况、周围环境、建筑安全等级等对变形进行严格控制。通过现场监测的深基坑围护结构变形信息资料,对实测数据进行整理和分析,利用神经网络对围护结构的变形做出预测的智能化施工成为基坑工程的发展趋势之一。研究了一种基于遗传算法的广义回归神经网络学习算法。该算法运用遗传算法寻找广义回归神经网络唯一参数光滑因子的最优解,将最优解赋予广义回归神经网络进行预测。在时间序列预测中,工程实例计算证明了遗传–广义回归神经网络预测的有效性和可行性,为时间序列预测提供了一种新途径。
Due to many factors, the excavation project has now become the key and difficult point in geotechnical engineering. In the construction of foundation pit, it is necessary to strictly control the deformation according to the actual conditions of the site, the surrounding environment and the level of building safety. Through the field monitoring of deep excavation deformation information of retaining structures, the measured data are collated and analyzed, the use of neural network to predict the deformation of the envelope structure of the intelligent construction of the pit as one of the development trend. A generalized regression neural network learning algorithm based on genetic algorithm is studied. The algorithm uses genetic algorithm to find the optimal solution of the smoothing factor of the unique parameter of the generalized regression neural network, and gives the optimal solution to the generalized regression neural network to predict. In the time series prediction, the engineering case calculation proves the validity and feasibility of the genetic - generalized regression neural network prediction, which provides a new way for the time series prediction.