基于机器学习方法的安全帽佩戴行为检测

施工现场作业人员是否佩戴安全帽主要依靠人工检查,存在监管效率低、时效性差等问题,为了实时自动监管作业人员是否佩戴安全帽,提出1种基于机器学习的安全帽佩戴行为检测方法。首先利用深度学习YOLOv3算法检测出现场视频中的施工人员脸部位置,根据安全帽与人脸的关系估算出安全帽潜在区域;然后对安全帽潜在区域图像进行增强处理,使用HOG（方向梯度直方图）提取样本的特征向量;再利用SVM（机器学习的支持向量机）分类器对脸部上方是否有安全帽进行判断,进而实现对施工人员安全帽佩戴行为的实时检测与预警。以某高铁站施工现场为例进行验证,研究数据表明在施工通道和塔吊作业区域,该方法可实时有效检测出工人未佩戴安全帽的行为,识别率达90%。     

基于支持向量机的管制疲劳检测模型研究

管制疲劳是影响民航安全的重大隐患,准确检测疲劳状态是进行疲劳预警、降低疲劳风险的关键。提出应用支持向量机模型融合多生理参数和眼动指标进行管制疲劳检测,通过MP150多导生理记录仪与眼动仪,采集模拟管制试验中正常与剥夺睡眠状态下被试的试验数据,同时记录其Karolinska疲劳等级和操作绩效。结果表明,RR间期、LF/HF、快慢波比值、PERCLOS和扫视速度均与管制疲劳呈较强相关,利用支持向量机融合五项指标构建管制疲劳检测模型,对于五级疲劳度的识别准确率为78. 1%,判断正常组与剥夺睡眠组的准确率为94. 2%。     

基于智能手机采集数据的施工活动识别

施工人员的活动监控对施工安全管理及预防职业疾病至关重要。为提升施工安全的智能化管理水平,对楼板钢筋工程施工中的8个主要活动进行识别。为弥补单一传感技术采集维度不足的缺陷和集成传感技术对系统灵活性的限制,采用智能手机内置加速度传感器和陀螺仪采集试验人员模拟施工人员活动时的加速度和倾角数据,并提取平均值、标准差、协方差、四分位距(IQR)为活动的特征矢量。通过决策树中的CART算法建立分类训练模型,采用“交叉验证法”对模型进行评估和验证。测试结果表明:对于样本个体的平均分类准确率为95.28%,预测准确率为92.86%;样本总体的分类准确率为89.67%,预测准确率为94.82%。研究表明,基于智能手机采集数据的决策树模型可以用于施工人员的活动识别。     

SVM application in hazard assessment: Self-heating for sulfurized rust

In order to assess the oxidation self-heating hazard of sulfurized rust, for particular ambient conditions in crude oil tanks, the support vector machine (SVM) technique is applied to predict the maximum temperature (T_max) of oxidation self-heating process. Five governing parameters are selected, i.e. the water content, mass of sulfurized rust, operating temperature, air flow rate and oxygen concentration in the respiratory/safety valve. The efficiency and validity of the SVM predictions are investigated in the case of two sets of data: more than 85 experiments performed in academic lab (China) and almost 17 additional results collected from existing literature. Two main steps are also discussed: the training process (on selected subsets of data) and prediction process (for the remaining subsets of data). It can be concluded that for both datasets the maximum temperature (T_max) values calculated by SVM technique were in good accordance with the experimental results, with relative errors smaller than 15% except for a few cases.The SVM technique seems therefore to be relevant and very helpful for complex implicit processes such as chemical reactions, as it is the case of the oxidation of sulfurized rust in oil tanks. Furthermore, such predictive methods can be continuously be improved through additional experiments feedback (larger databases) and can then be of crucial help for monitoring and early warning of hazardous reactions.     

基于模糊FMEA法的乏燃料剪切机剪切装置风险评估

为了有效控制乏燃料后处理立式送料剪切机剪切装置在运行过程中的风险,采用基于模糊理论的故障模式及影响分析(FMEA)对剪切装置的故障模式进行风险评价。首先将风险优先指数(RPN)的3个评价指标视为模糊变量并模糊化,接着采用专家打分的方式对各故障模式进行模糊评价,采用重心法解模糊化,计算得出各故障模式的模糊风险优先指数(FRPN),从而确定机械设备中各故障模式的风险控制顺序。利用模糊FMEA法对乏燃料立式送料剪切机剪切装置的故障模式进行了风险评估,结果表明,吊篮焊接件裂纹、主刀体破损、刀具退刀固定档板变形为风险较大的故障模式,分析结果与工程实际较吻合。     

厂矿企业大宗原、燃料卸车机械安全改造技术

厂矿企业卸车机械长期运行,存在设备老化、检修频繁的现象.通过对卸车机械存在安全隐患的分析,论述了机械系统的行走机构、挖掘机构、钢结构框架、皮带输送机构、操作室和电气系统的供电电源、大车行走电机回路、卷扬电机等部件的安全改造技术.     

Knowledge Production and Learning for Sustainable Landscapes: Seven Steps Using Social–Ecological Systems as Laboratories

There are multiple challenges regarding use and governance of landscapes’ goods, functions and intangible values for ecosystem health and human well-being. One group of challenges is to measure and assess principal sustainability dimensions through performance targets, so stakeholders have transparent information about states and trends. Another group is to develop adaptive governance at multiple levels, and management of larger geographical areas across scales. Addressing these challenges, we present a framework for transdisciplinary research using multiple landscapes as place-based case studies that integrates multiple research disciplines and non-academic actors: (1) identify a suite of landscapes, and for each (2) review landscape history, (3) map stakeholders, use and non-use values, products and land use, (4) analyze institutions, policies and the system of governance, (5) measure ecological, economic, social and cultural sustainability, (6) assess sustainability dimensions and governance, and finally (7) make comparisons and synthesize. Collaboration, communication and dissemination are additional core features. We discuss barriers bridges and bridges for applying this approach.     

Eigenbehaviors: identifying structure in routine

Longitudinal behavioral data generally contains a significant amount of structure. In this work, we identify the structure inherent in daily behavior with models that can accurately analyze, predict, and cluster multimodal data from individuals and communities within the social network of a population. We represent this behavioral structure by the principal components of the complete behavioral dataset, a set of characteristic vectors we have termed eigenbehaviors. In our model, an individual’s behavior over a specific day can be approximated by a weighted sum of his or her primary eigenbehaviors. When these weights are calculated halfway through a day, they can be used to predict the day’s remaining behaviors with 79% accuracy for our test subjects. Additionally, we demonstrate the potential for this dimensionality reduction technique to infer community affiliations within the subjects’ social network by clustering individuals into a “behavior space” spanned by a set of their aggregate eigenbehaviors. These behavior spaces make it possible to determine the behavioral similarity between both individuals and groups, enabling 96% classification accuracy of community affiliations within the population-level social network. Additionally, the distance between individuals in the behavior space can be used as an estimate for relational ties such as friendship, suggesting strong behavioral homophily amongst the subjects. This approach capitalizes on the large amount of rich data previously captured during the Reality Mining study from mobile phones continuously logging location, proximate phones, and communication of 100 subjects at MIT over the course of 9 months. As wearable sensors continue to generate these types of rich, longitudinal datasets, dimensionality reduction techniques such as eigenbehaviors will play an increasingly important role in behavioral research. This contribution is part of the special issue “Social Networks: new perspectives” (Guest Editors: J. Krause, D. Lusseau and R. James). An erratum to this article can be found at     

Developing a theory of change for a community‐based response to illegal wildlife trade

The escalating illegal wildlife trade (IWT) is one of the most high‐profile conservation challenges today. The crisis has attracted over US$350 million in donor and government funding in recent years, primarily directed at increased enforcement. There is growing recognition among practitioners and policy makers of the need to engage rural communities that neighbor or live with wildlife as key partners in tackling IWT. However, a framework to guide such community engagement is lacking. We developed a theory of change (ToC) to guide policy makers, donors, and practitioners in partnering with communities to combat IWT. We identified 4 pathways for community‐level actions: strengthen disincentives for illegal behavior, increase incentives for wildlife stewardship, decrease costs of living with wildlife, and support livelihoods that are not related to wildlife. To succeed the pathways, all require strengthening of enabling conditions, including capacity building, and of governance. Our ToC serves to guide actions to tackle IWT and to inform the evaluation of policies. Moreover, it can be used to foster dialogue among IWT stakeholders, from local communities to governments and international donors, to develop a more effective, holistic, and sustainable community‐based response to the IWT crisis.     

Evolution of disaster relief law under multiple transformations: Progressive learning or walking in a circle?

The aim of this paper is to examine the association between double-loop disaster learning and transformations of political systems. The particular question is how disasters increase the rapidity and complexity, with which the transforming regimes face the unsuitability and inadequacy of the past forms and processes of disaster governance. The centennial evolution of the disaster relief law at a country scale (Czech Republic, resp. Czechoslovakia, 1918-present) is used as an example. The thorough search in digital repository of Czech and Slovak Parliamentary Library was performed to obtain legislative norms and protocols from parliamentary debates, whereas major disasters that induced new laws (1927 flash flood, 1947 drought, 1997 and 2002 floods) were identified from existing databases and from detailed reconstructions of historical disastrous events. The described cases enabled to add new empirical evidence contesting (i) the process scheme of disaster politics during critical junctures and institutional transformations, and (ii) the progressive nature of the double-loop learning approach related to disasters. It is shown that the double-loop learning from disasters and its benefits should be considered as time-dependent and related to permissive and productive conditions that facilitated the adoption of the new disaster relief law (i.e. the learning process). Finally, the implications for the research agenda (historical disaster research) and for disaster policy (evaluation of disaster relief effectiveness) are summarized.