论文部分内容阅读
岩土体变形位移是受多种因素影响而发展演化的多维非线性动力系统,其过程既有确定性发展的趋势项,又有受不确定因素影响的随机项。根据灰色预报方法有较好的预报系统变化的总体趋势和神经网络具有逼近任意函数的能力,建立等维动态灰色-时序神经网络实时预报模型对趋势项和随机项进行预报。以基于实测资料的某铁路地基沉降预报为例,证实此模型临期预报可靠、精度高,同时中长期预报也具有参考价值;且模型原理简单,操作容易,可为岩土体变形位移控制、信息施工及修正设计方案提供参考。
Deformation and displacement of rock and soil are multidimensional nonlinear dynamical systems that evolve and evolve under the influence of many factors. The process includes deterministic trend and random items under the influence of uncertain factors. According to the gray forecasting method, the general trend of forecasting system change and neural network have the ability of approximating any function. The real-time forecasting model of isokinetic gray-time series neural network is established to forecast trend items and random items. Taking the subsidence forecast of a railway foundation based on the measured data as an example, it is proved that the forward prediction of the model is reliable and accurate, and the mid- and long-term prediction also has reference value. The model is simple in principle and easy to operate, Information construction and revision of design options for reference.