考虑主动避让行为的养老场所人员疏散研究

为明确主动避让行为对养老场所人员疏散的影响,建立考虑主动避让弱势群体行为的元胞自动机模型,模型分别考虑避让弱势群体前方行走区域、斜前方行走区域、前方和斜前方行走区域对人员疏散的影响;分析不同避让方式下弱势群体比例、行人密度、弱势群体位置分布对整体疏散结果的影响.结果 表明:主动避让弱势群体行为能够降低疏散时间,增加疏散...     

Experimental investigation on explosion flame propagation of wood dust in a semi-closed tube

In order to explore flame propagation characteristics during wood dust explosions in a semi-closed tube, a high-speed camera, a thermal infrared imaging device and a pressure sensor were used in the study. Poplar dusts with different particle size distributions (0–50, 50–96 and 96–180 μm) were respectively placed in a Hartmann tube to mimic dust cloud explosions, and flame propagation behaviors such as flame propagation velocity, flame temperature and explosion pressure were detected and analyzed. According to the changes of flame shapes, flame propagations in wood dust explosions were divided into three stages including ignition, vertical propagation and free diffusion. Flame propagations for the two smaller particles were dominated by homogeneous combustion, while flame propagation for the largest particles was controlled by heterogeneous combustion, which had been confirmed by individual Damköhler number. All flame propagation velocities for different groups of wood particles in dust explosions were increased at first and then decreased with the augmentation of mass concentration. Flame temperatures and explosion pressures were almost similarly changed. Dust explosions in 50–96 μm wood particles were more intense than in the other two particles, of which the most severe explosion appeared at a mass concentration of 750 g/m³. Meanwhile, flame propagation velocity, flame propagation temperature and explosion pressure reached to the maximum values of 10.45 m/s, 1373 °C and 0.41 MPa. In addition, sensitive concentrations corresponding to the three groups of particles from small to large were 500, 750 and 1000 g/m³, separately, indicating that sensitive concentration in dust explosions of wood particles was elevated with the increase of particle size. Taken together, the finding demonstrated that particle size and mass concentration of wood dusts affected the occurrence and severity of dust explosions, which could provide guidance and reference for the identification, assessment and industrial safety management of wood dust explosions.     

Integration of data analytics with cloud services for safer process systems,application examples and implementation challenges

Emerging sensors, computers, network technologies, and connected platforms result potentially in an immeasurable collection of data within plant operations. This creates the possibility of solving problems innovatively. Because most of the data appear to be unstructured or semi-structured, organizations shall design and adopt new strategies. Further, workflow architectures with data analytics are needed including machine learning tools and artificial intelligence techniques before proto-type solutions can be developed. We shall discuss several prospects of using (big) data analytics integrated with cloud services to produce solutions for improving plant operations. The paper outlines the vision and a systematic framework highlighting the data analytics lifecycle in the area of plant operation, process safety, and environmental protection. Four rather diverse example case studies are demonstrated including (1) deep learning-based predictive maintenance monitoring modeling, (2) Natural Language Processing (NLP) for mining text, (3) barrier assessment for dynamic risk mapping (DRA), and (4) correlation development for sustainability indicators. It further discusses the challenges in both research and implementation of proposed solutions in the industry. It is concluded that a well-balanced integrated approach including machine supporting decisions integrated with expert knowledge and available information from various key resources is required to enable more informed policy, strategic, and operational risk decision-making leading to safer, reliable and more efficient operations.     

基于极限学习机的炭化可燃物着火时间预测

机器学习技术近年来在许多传统科学领域取得了应用,针对火灾中炭化可燃物着火时间与物性参数及环境参数之间关系复杂的特点,提出了一种基于极限学习机的预测方法以实现不同物性及环境参数时着火时间的快速准确预测,为防治及扑救火灾提供参考。首先建立炭化可燃物热解数值模型,考虑了可燃物热解过程中的含水率以及热解反应、气体流动等复杂物理化学反应过程,然后搭建极限学习机,以数值模拟数据为基础进行训练及验证工作。结果表明基于极限学习机的预测方法能够有效实现炭化可燃物着火时间的快速准确预测,平均相对误差小于3%。     

基于深度学习的污染场地作业人员着装规范性检测

为解决污染场地修复作业中缺乏统一的人员安全保障管理措施以及人工监管困难的问题,结合HSE理念提出1种基于优化Faster R-CNN的作业人员着装规范性检测算法。该方法在回归损失函数中引入L2正则项,既保证模型的泛化能力,又提高深层网络模型收敛速度。基于自建着装规范数据集（Dress Code Dataset）进行实验,评价算法检测时间和mAP等指标。结果表明:所提出的着装规范性检测算法检测时间为44 ms,mAP为88.17%,解决了传统检测算法中实时性和准确率低的问题,且模型具有更好的泛化性和鲁棒性。     

Using European naturalistic driving data to assess secondary task engagement when stopped at a red light

Problem: Some evidence exists that drivers choose to engage in secondary tasks when the driving demand is low (e.g., when the car is stopped). While such a behavior might generally be considered as rather safe, it could be argued that the associated diversion of attention away from the road still leads to a reduction of situational awareness, which might increase collision risk once the car regains motion. This is especially relevant for texting, which is associated with considerable eyes-off-the-road-time. Nonetheless, it seems that previous research has barely addressed the actual engagement in secondary tasks while waiting at a red light (as compared to just addressing the tasks’ mere prevalence). Objective: The present study investigated secondary task engagement while stopped at a red light using European naturalistic driving data collected through the UDRIVE project. Attention was given to the whole engagement process, including simple prevalence and the tasks’ relation (in terms of start/end) to the red light period. Moreover, given that texting is one of the most problematic forms of distraction, it was characterized in more detail regarding glance behavior. Method: Videos of 804 red light episodes from 159 drivers were annotated. Glance behavior was also coded for a sub-set of 75 texting events and their matched baselines. Results, conclusions and practical applications: Drivers engaged in at least one secondary task across almost half of the annotated red light episodes. Drivers who texted while stopped spent most of the time looking at their cell phone. Consequently, drivers might not have been prepared for potentially unexpected events once the light turned green. Further, drivers concluded texting a considerable number of times well after the red light period, which has potential implications for traffic safety.     

Digital screen use for a road safety campaign message was not associated with road safety awareness of passers-by: A quasi-experimental study

Introduction: Recent evidence suggests fatality risks for cyclists may be increasing in Britain. Understanding how to increase levels of cycling while keeping risk low is paramount. Educating drivers about cyclists may help with road safety, and mass-media messaging is a possible avenue, potentially utilizing digital displays screens in public areas. However, no studies have examined the use of these screens for road safety campaigns. Methods: A quasi-experiment was conducted to examine if digital screens may be effective to raise awareness of a campaign message and encourage recall of car drivers. A digital campaign image was selected that encouraged car drivers and cyclists to ‘look out for each other,’ and stated than 80% of cyclists owned a driving license. Views and knowledge on driver priorities around cyclists were examined before (control) and after campaign exposure (intervention), and tested using regression modelling. Results: 364 people were interviewed over five days. Those interviewed on intervention days were more likely to rank ‘Look out for cyclists’ as being more important compared to those interviewed on control days (OR 1.20), but this was not statistically significant (p = 0.355). Those who said they had seen the image did not rank ‘Look out for cyclists’ higher than those who said they had not seen it (p = 0.778). The disparity between reported and displayed percentage of cyclists with a driving license did not differ between intervention and control days, but was 8% higher amongst those who claimed to have seen the image (p = 0.026). Conclusions: We did not find strong evidence that use of an image on digital screens increased public awareness or recall of a casualty reduction campaign message. Work is needed to investigate the effects of longer-term exposure to road safety images. Practical Applications: Short-term use of digital signage is not recommended for raising awareness of road safety campaigns.     

基于深度学习算法的叉车危险操作行为检测

随着物流仓储行业快速发展及叉车数量的不断增多,针对叉车作业过程中存在的人员碰撞、挤压、坠落等潜在风险,迫切需要对危险行为进行及时检测和预警。为解决人员值守易漏报误报及传统方法检测精度低的问题,建立基于图像特征识别的叉车检测深度学习模型和算法。通过采集、处理现场视频图像素材,完成模型的训练及性能评价,建立相应的报警规则和报警阈值,搭建测试环境并进行仿真测试,开发相应的软件系统。结果表明:模型检测速率为130 ms/帧,人员靠近叉车准确率为85.6%,叉车举升人员准确率为83.7%,达到良好的实践效果。     

Meta-analysis of human connection to nature and proenvironmental behavior

Understanding what drives environmentally protective or destructive behavior is important to the design and implementation of effective public policies to encourage people's engagement in proenvironmental behavior (PEB). Research shows that a connection to nature is associated with greater engagement in PEB. However, the variety of instruments and methods used in these studies poses a major barrier to integrating research findings. We conducted a meta-analysis of the relationship between connection to nature and PEB. We identified studies through a systematic review of the literature and used Comprehensive Meta-Analysis software to analyze the results from 37 samples (n = 13,237) and to test for moderators. A random-effects model demonstrated a positive and significant association between connection to nature and PEB (r = 0.42, 95% CI 0.36, 0.47, p < 0.001). People who are more connected to nature reported greater engagement in PEB. Standard tests indicated little effect of publication bias in the sample. There was significant heterogeneity among the samples. Univariate categorical analyses showed that the scales used to measure connection to nature and PEB were significant moderators and explained the majority of the between-study variance. The geographic location of a study, age of participants, and the percentage of females in a study were not significant moderators. We found that a deeper connection to nature may partially explain why some people behave more proenvironmentally than others and that the relationship is ubiquitous. Facilitating a stronger connection to nature may result in greater engagement in PEB and conservation, although more longitudinal studies with randomized experiments are required to demonstrate causation.     

交通空气污染对大鼠子代发育及社交行为的影响

交通空气污染与儿童自闭症的关联时有报道,但其因果关系仍待进一步探索。为研究出生前后暴露于不同水平的交通空气污染环境对生长发育和自闭症相关社交行为的影响,将Wistar大鼠雌雄1∶1交配,受孕雌鼠根据子代接触交通污染的时间(出生后暴露或胚胎期暴露)及浓度(低、中和高3个水平)分为6组。在子代出生后21～27 d,测量子代的体重、身长及大脑脏器系数,并根据三箱实验评估社交行为,同时检测仔鼠脑组织组蛋白H3第4个赖氨酸(H3K4)的甲基化水平。结果显示,出生前后,尤其是胚胎发育阶段接触较高浓度的交通空气污染物可导致幼鼠体重及大脑湿重较低,身长较短、大脑脏器系数偏高等生长发育特征改变。同时,出生前后较高浓度的交通空气污染物暴露可导致幼鼠出现社交能力下降、社交记忆受损及社交新颖性减弱等典型自闭症样行为改变。虽然组蛋白H3K4甲基化水平与交通空气污染物浓度之间未见统计学关联,但出生后暴露组具有较高的H3K4甲基化水平,提示出生前后接触交通空气污染物可能与自闭症的发生发展有关,而不同暴露时期对脑组织H3K4甲基化水平的影响,为今后进一步探讨自闭症的表观遗传机制提供了线索。