基于LSSVM对盾构施工诱发地面塌陷变形预测

成都地铁1,2号线盾构在砂卵石地层施工诱发多次地面塌陷事故,其地面塌陷变形曲线受多种因素影响,且各因素对地面塌陷变形曲线的影响表现出非线性特性,因此地面塌陷变形曲线很难用显示的数学表达式进行求解。最小二乘支持向量机是基于统计学习理论的机器学习方法,该方法能避免传统神经网络诸多缺陷,能够分析复杂因素对结果影响的潜在规律,据此引入最小二乘支持向量机,以地层物理力学参数、盾构埋深和地层损失数量为输入参数,建立地面塌陷变形曲线预测模型。经过样本检验,预测模型具有较强的泛化能力,预测结果精度和可靠性较高。     

基于GIS考虑准动态湿度指数的滑坡危险性预测水文-力学耦合模型研究

降雨在区域浅层滑坡灾害危险性预测研究中起着关键作用。为综合斜坡地形因素和由降雨引起的地表水下渗、地下水径流对斜坡稳定性的影响,结合基于物理过程的无限斜坡模型和基于数字高程模型 DEM 的简化运动波模型,提出考虑准动态湿度指数的滑坡危险性预测水文-力学耦合模型。该模型基于地理信息系统GIS平台,采用矢量-栅格复合单元对斜坡进行危险性分析(即以斜坡单元为基本研究对象,用栅格数据进行单个斜坡单元稳定性分析)。首先,基于极限平衡理论推导无限斜坡模型;然后,在考虑降雨强度以及持续时间的情况下应用简化的运动波模型计算降雨入渗和地下水径流作用过程中的地形准动态湿度指数,得到滑坡土体饱和因子在时间和空间上的分布情况,并在此基础上实现水文模型与无限斜坡模型的耦合;最后,以三峡库区巴东新城区滑坡灾害危险性预测为例,验证模型在区域浅层滑坡灾害危险性预测中具有较高的精度。     

粗粒土BP神经网络本构模型研究

为了描述具有应力应变关系非线性和剪胀性的粗粒土本构特性,利用改进的BP神经网络算法,通过优选网络结构和对粗粒土的大型三轴固结排水剪试验数据样本学习,建立了一个以平均主应力p和广义剪应力q作为网络输入向量、以体应变εv和剪应变εs作为网络输出向量的粗粒土BP神经网络本构模型。利用此模型对粗粒土的应力应变关系进行了预测,整体预测结果的最大误差均在10%内。预测表明本神经网络模型具有良好的预测精度和适用性。     

稳定型橡胶改性沥青混合料动态模量预估

测试了8种混合料类别、12种级配组成的稳定型橡胶改性沥青混合料的动态模量,分析了稳定型橡胶改性沥青混合料的动态力学性能特点.针对Witczak模型对稳定型橡胶改性沥青混合料动态模量预测效果不理想的问题,采用Levenberg-Marquardt非线性回归方法修正了Witczak模型.结果表明:稳定型橡胶改性沥青混合料具有良好的动态力学性能;修正后的Witczak模型可以较好地预测稳定型橡胶改性沥青混合料的动态模量.     

Effects of urbanisation on the quality of the urban runoff for Delhi watershed

The current paper is the first study to report quality of urban runoff, focusing on the diffuse sources contributing to pollutant loads in the Delhi watershed. The quality of stormwater runoff from the urban watershed of Delhi was evaluated to assess the effects of different land-use practices on pollutant concentrations. Six sites within the Delhi watershed with different land-use characteristics were selected. The microbial quality of the urban runoff from sub-catchments of Delhi watershed is very poor with Faecal Coliform (FC) levels varying between 6 and 7 log orders. There was a good correlation between the chemical and biological quality of urban runoff with land-use characteristics such as impervious land and population density. Except for chemical oxygen demand all other parameters in urban runoff were found to exceed the effluent discharge standards set by the National River Conservation Directorate (NRCD), India. The public health risks associated with exposure to urban runoff via different routes have also been qualitatively estimated.     

Pesticides in rain and roof runoff waters from an urban region

The levels of organonitrogen, organochlorine and organophosphorus pesticides were monitored in rain and roof runoff waters in Gdańsk (Poland) over a period of 6 months (October 2000 – March 2001). Buildings included in the study were of different ages, and covered with a variety of roofing materials. Samples were collected during the precipitation events. Pesticides were determined by gas chromatography combined with NPD and ECD detection. Bromofos, heptachlor epoxide and o,p′-DDE were detected the most often in the samples collected. In general, pesticide concentrations were higher in roof runoff than in rainwater. The type of the roofing material had a significant effect on the pesticide levels in runoff waters. The highest analyte levels were recorded in October and November, when the temperature was the highest.     

Laboratory analysis of the outflow and detention processes from modular tray-based blue roofs

ABSTRACT

Novel laboratory experiments were carried out to investigate the outflow process and the detention effects of a tray module of a blue roof (BR) system designed for runoff control from building roofs. The experiments revealed the outflow from the tray to decline during time, due to a number of complex dynamics occurring in proximity of the outlet orifice. The tests also showed the system to provide potential for relevant attenuation of the flow peak during inflow events (i.e. precipitation) of different magnitude. Based on regression analysis, an empirical relationship was developed for the evaluation of the tray outflow as a function of time, thus providing a tool for improved design of BRs. Finally, combination of the identified relationship with a routine for the water balance simulation of the tray was shown to describe reliably the detention process in the BR tray module.     

A critical review of methods used to obtain flow patterns and volumes of individual domestic water using appliances

Determining the daily flow rates and water use patterns of specific household fixtures and appliances is important for water use monitoring, deriving effective strategies for water demand management, designing onsite water reuse systems, and efficient planning and investment in the urban water sector. Currently, there are a limited number of publications which address the need to measure flow patterns and volumes of individual household water-using appliances, or which analyze the alternatives to do so. This paper critically reviews and compares different methods of obtaining the specific flow patterns and volumes of water used by individual household water-using microcomponents (i.e. toilet, bath/shower, washbasin, washing machine, kitchen sink, dishwasher, etc.) in order to make recommendations about the appropriateness of these methodologies for various applications.     

Long-term monitoring and assessment of a precast continuous viaduct

Precast girders have recently been widely employed in the construction of bridges and viaducts. The new bridge over the Tagus River in Portugal, the Lezíria Bridge, comprehends a 9160 m long south approach viaduct, which was built with precast girders made continuous in situ. Given the relevance of this construction, a long-term monitoring system was implemented and measurements were taken since the start of the construction. The observed parameters were concrete strains and temperatures, deck rotations, joint displacements, accelerations and also environment temperature and relative humidity. The work presents the precast structure, the monitoring system and the appraisal of a statistical procedure for the long-term assessment of the structural behaviour. This procedure is based on prediction models, which establish the normal correlation patterns between environmental and material parameters (such as concrete temperature and shrinkage strains), and the observed structural response in terms of strains, rotations and movements of expansion joints. The calculation of the normal correlation pattern comprehends the minimisation of a square error. By applying the prediction model to the structural response measured in the south approach viaduct of the Lezíria Bridge, it was found that this methodology is a feasible tool for real-time damage detection of bridges.