基于RBF神经网络的地铁振动预测分析  

The Subway Vibration Forecast Analysis Based on RBF Neural Network

作　　者：王秀丽[1] 潘雷[1] 彭桂力[1] 梁茵[2] 王贝贝[1] 

机构地区：[1]天津城建大学控制与机械工程学院,天津300384 [2]天津城建大学能源与安全工程学院,天津300384

出　　处：《天津城建大学学报》2016年第6期465-470,共6页Journal of Tianjin Chengjian University

基　　金：天津市高等学校科技发展基金计划项目(20140527)

摘　　要：随着城市地下轨道交通的大力发展,列车运行引起的环境振动及噪声对周边环境及人们生活的影响越来越大.如何准确预测地铁列车引起的振动,有效分析振动的传播规律,为减振降噪提出有价值的参考数据就显得尤为重要.实测的振级数据是引起地铁列车振动的所有因素的综合效应,隐含了地铁振动的规律.以实测的振级数据为样本,结合滑动窗口技术构建径向基神经网络预测模型,得出了振级的拟合曲线,实现了对振动峰值的预测.通过实例说明,预测效果良好,建立的预测模型能充分反映地铁振动系统的非线性函数的映射关系.Along with the development of city rail transport,the influence of the vibration and noise caused by train movement are bigger and bigger on the environment and people’s life. How to forecast the vibration accurately and analyze itspropagation rule effectively are very important. It also can present some valuable reference data to depress vibration and noise. The measured data of the vibration level is combined effect of all factors which cause the subway vibration,implying the rule of the subway vibration. Using the measured data of the vibration level as sample,radial basis function （RBF） neural network forecasting model is structured with combinations of the sliding window technology. At last the fitted curves are obtained and the vibration’s peak value forecast is realized. Some examples illustrate that the forecasting result is goodand the forecast model fully reflects the mapping relation of nonlinear function about the subway vibration system.

关 键 词：地铁振动 加速度振级 径向基神经网络 预测 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程] U231[自动化与计算机技术—控制科学与工程]