BP神经网络法在斜拉桥换索施工控制中的应用研究

运用神经网络理论,结合一座双塔三跨混凝土斜拉桥换索施工过程,对斜拉桥换索过程索力和主梁标高进行预测研究。运用平面杆系有限元理论计算出施工状态的理论索力和理论标高,再通过数值计算软件MATLAB7.1建立神经网络仿真模型得出各施工阶段的索力和标高预测值,为具体的换索工程提供参考。通过对比该斜拉桥换索过程中索力及主梁控制点标高的计算值和实测值可知,预测数据与目标数据相近,变化趋势一致,验证了BP神经网络在换索施工控制中进行参数预测的可行性。     

基于BP神经网络在森林防火专家系统不确定推理中的应用研究

提出将改进的BP神经网络应用于森林防火专家系统的不确定性推理中,其良好的自学习和泛化能力,可以解决基于可信度规则的知识表示在实际应用中导致的规则激增,推理速度缓慢的问题。该方法将不确定的知识用可信度区间表示,通过知识编码,设计并训练BP网络,最后用MATLAB进行仿真。实验结果表明:BP神经网络可以自动学习专家的典型经验,并且能将之准确的推广,隐含层神经元个数的确定和典型样本的选取决定了准确精度。在实际的专家系统不确定推理应用中具有应用价值。     

Pushing and pulling: an assessment tool for occupational health and safety practitioners

A tool has been developed for supporting practitioners when assessing manual pushing and pulling operations based on an initiative by two global companies in the manufacturing industry. The aim of the tool is to support occupational health and safety practitioners in risk assessment and risk management of pushing and pulling operations in the manufacturing and logistics industries. The tool is based on a nine-multiplier equation that includes a wide range of factors affecting an operator’s health risk and capacity in pushing and pulling. These multipliers are based on psychophysical, physiological and biomechanical studies in combination with judgments from an expert group consisting of senior researchers and ergonomists. In order to consider usability, more than 50 occupational health and safety practitioners (e.g., ergonomists, managers, safety representatives and production personnel) participated in the development of the tool. An evaluation by 22 ergonomists supports that the push/pull tool is user friendly in general.     

Reaction Moment at the L5/S1 Joint During Simulated Forward Slipping With a Handheld Load

The purpose of the study was to investigate the effects of load on the net moment response at the L5/S1 joint during simulated slip events. Six young individuals were instructed to take one step with a handheld load. Sudden floor movement was randomly introduced to simulate unexpected slips. Different loads conditions (0%, 10%, 20%, 30% of body weight) were introduced at random. Three-dimensional net moments at the L5/S1 joint were computed via downward inverse dynamic model. Peak joint moment generated at 30% load level was found to be significantly higher compared to no-load condition. No peak moment differences were found among no-load, 10% or 20% load levels. Additionally, the findings from this study indicated a flexiondominant net L5/S1 joint moment pattern during motion phase associated with slip-induced falls.     

A Multilevel Approach to Manual Lifting in Manufacturing Industries

Musculoskeletal injuries are often the consequences of wrong postural configurations used during Manual Materials Handling (MMH). This eventually leads to a large payout of worker’s compensation and loss of production time. A simulated study of back injury risks has been carried out on seven selected manufacturing industries to identify and evaluate harmful working postures. For each MMH task, Ovako Working Posture Analyzing System (OWAS) codes have been identified with the help of motion study pictures. Also, Chaffin's biomechanical model was used to calculate L5/S1 load compression values on the spine during MMH activities. The multilevel approach adopted was a combination of OWAS and Chaffin’s biomechanical model. The application of a digitizer enabled us to identify the coordinates and it made a subsequent evaluation of the angles of each body link possible.     

基于BP神经网络的建筑工人不安全行为预警模型

为减少建筑工人不安全行为,提高企业安全管理水平,采用事故统计分析、文献分析、质性访谈方法获取不安全行为影响因素,从组织、个人、外在环境、设备4个方面建立不安全行为预警指标体系,在此基础上,基于反向传播(BP)神经网络原理,将预警指标作为网络输入,不安全行为无警、轻警、中警、重警4种状态作为网络输出,进而设计编制预警问卷...     

裂隙介质VOCs赋存迁移特征与场地修复难点

梳理了裂隙介质污染的研究方法,分析裂隙介质中VOCs的赋存迁移特征与修复技术原理,结合我国实际情形分析该类场地修复治理的难点并提出相关建议.裂隙介质场地中,对流弥散、吸附解吸、跨界输移的协同作用影响VOCs的环境行为与归趋.在理论层面,基于分离、化学氧化与生物转化过程的修复原理能够实现裂隙中VOCs的去除,但由于氧化还原产物堵塞、吸附态污染物反向扩散等原因,裂隙介质VOCs污染修复过程可能存在拖尾、修复后可能存在反弹的问题.与孔隙介质场地相比,不确定性是造成裂隙介质场地修复困难的主要原因,主要体现在场地调查阶段样品代表性差、修复治理过程中裂隙对污染物的阻滞作用所致传质受限、污染物易发生跨介质迁移风险等方面.后续研究亟需在充分认识裂隙污染物迁移转化机制的基础上,进一步探索适合我国修复与再利用实际情形的修复和风险管控模式.     

FeCl3蚀刻液膜电解阴极液中Ni2+的萃取回收

采用自制的SSX萃取剂对FeCl₃蚀刻液膜电解阴极液（简称废液）中的Ni²⁺进行萃取回收。考察了萃取pH、SSX萃取剂含量、萃取相比（SSX萃取剂与废液的体积比）、萃取时间、萃取次数对Ni²⁺萃取率的影响，以及反萃剂HCl溶液浓度、反萃相比（反萃剂与萃取液的体积比）、反萃时间对Ni²⁺反萃率的影响。实验结果表明： 当SSX萃取剂质量分数20%、萃取pH 2.0、萃取相比1.0、萃取时间10 min、1次萃取时，Ni²⁺萃取率可达74.56%；当反萃剂HCl溶液浓度6.0 mol/L、反萃相比1.5、反萃时间10 min时，Ni²⁺反萃率达93.10%；再生后的SSX萃取剂重复使用4次后，Ni²⁺的累积萃取率达91.00%，萃取剂中Ni²⁺的质量浓度可达14.94 g/L；反萃液经浓缩、结晶处理可制备电镀用NiCl₂产品。     

南京江北2014-2016年PM2.5质量浓度分布特征及气象和传输影响因素分析

利用2014-2016年南京江北地区PM_2.5质量浓度和气象要素的小时数据,并结合HYSPLIT模式后向轨迹聚类分析和PSCF法分析了PM_2.5质量浓度的污染特征及其主要影响因素和主要来源特征.结果表明：2014-2016年PM_2.5质量浓度呈逐年下降趋势,下降幅度约为17.40%,由2014年的62.1 μg·m^-3下降至2016年的51.2 μg·m^-3,能见度由2014年的5.8 km上升至2016年6.6 km.PM_2.5质量浓度存在显著的月变化和季节变化特征,1月浓度最高,可达93.0 μg·m^-3;8月浓度最低,仅为38.8 μg·m^-3;冬季浓度最高,可达76.8 μg·m^-3,夏季浓度最低,仅为47.1 μg·m^-3.不同季节日变化均为单峰型分布.气象要素对PM_2.5质量浓度的影响较大,不同相对湿度下能见度和PM_2.5质量浓度具有较好的拟合关系.霾和非霾天PM_2.5质量浓度的阈值为15 μg·m^-3.不同季节的主导气团不同,春季主导气团为偏北气流和偏东气流,占比分别为43.50%和30.80%;夏季主导气团以东部气流为主,占比约为68.22%;秋季和冬季主导气团为来自北方的气流,总占比分别为83.52%和100%;偏北内陆气团PM_2.5质量浓度较大,偏东海洋性气团PM_2.5质量浓度较低.PM_2.5质量浓度潜在源区春冬季潜在源区范围较大,夏秋季潜在源区范围较小,季节变化显著.春季潜在来源主要分布在安徽、江西北部、江苏南部和浙江北部等地区,夏秋季分布在安徽东部、浙江北部和江苏南部等地区,冬季分布在安徽、河南东部,山东和江苏等地区.