AHP与GA-SVM耦合模型在滑坡预警中的应用*

为了减少滑坡造成的损失,提高滑坡预测的准确性,通过搭建灾害模拟平台获得滑坡的实验数据,在获得多组模拟实验数据后,分析各变量的特性。首先,通过层次分析(Analytic Hierarchy Process,AHP)算法,对滑坡进行危险度划分;然后,通过支持向量机(Support Vector Machine,SVM)建立模型,遗传算法(Genetic Algorithm,GA)再优化SVM参数,提出1种层次分析法与GA-SVM相耦合的模型。研究结果表明:AHP方法划分后的数据,通过GA与SVM结合建立的模型精度较好,实验预测结果与实际结果较为吻合,与单一SVM相比,精度更高,结果更好,更加适用于多变量的复杂非线性滑坡预警。     

提高GPS定位精度的群调度滤波算法*

为解决现行安全生产科技领域中全球定位系统（GPS）/惯性导航系统（INS）融合系统容易产生较大时变误差的关键问题,满足GPS定位信号的精度和信号数据处理时间的要求,提出1种群调度滤波算法,通过实验仿真,分别与模糊算法、机器学习算法及卡尔曼滤波网络算法进行比较研究。结果表明:群调度滤波算法兼顾了定位精度与数据处理时间,具有较大优势,可在航空安全、船舶安全、应急监测、灾后抢险等安全科技领域广泛应用。     

基于响应曲面法的遗煤自燃分析与研究*

为研究高瓦斯易自燃煤层不同供风量、高抽巷抽采流量、低抽巷抽采流量3因素对采空区自燃“三带”分布影响规律,选取阳煤五矿8406工作面为研究对象,在数值模拟研究基础上,采用Design Expert软件进行Box Behnken试验设计,构建采空区氧化升温带宽度在3因素、3水平条件下的二次回归响应曲面模型,并对不同条件下采空区氧化升温带宽度进行预测与分析。结果表明:二次回归方程P值为0.001 6,预测模型显著,模型的失拟项为0.606 3,不显著,回归方程具有统计学意义;当供风量为1 500~2 000 m3/min,低抽流量为450~650 m3/min,高抽流量为100~200 m3/min时,对氧化升温带宽度一次项重要度排序为C（高抽巷抽采流量）>A（供风量）>B（低抽巷抽采流量）,二次项重要度排序为AC（供风量和高抽巷抽采流量）>AB（供风量和低抽巷抽采流量）>BC（低抽巷抽采流量和高抽巷抽采流量）,且AB,AC,BC之间均无交互作用。     

分离岛式地铁站可用安全疏散时间计算与比较分析*

为研究大客流地铁车站风险因素,提出车站疏散能力分析的基本要素,结合某分离岛式地下车站,采用规范计算及精细网格模型对可用安全疏散时间进行计算与比较分析。结果表明:楼扶梯的通过效率、车站结构形式、疏散人员的数量及分布情况等因素对仿真模拟的结果产生影响;《地铁安全疏散规范》（GB/T 33668—2017）综合考虑了疏散至楼扶梯入口时间、楼扶梯上平均滞留时间及通道非均匀性偏差时间等,与仿真模拟结果的吻合度较高。研究结果可为国内类似车站的安全疏散设计及运营疏散组织提供参考。     

基于人工神经网络的气体泄爆最大超压预测研究

为解决传统经验公式在预测气体泄爆中最大超压出现时的较大偏差或过于保守的问题,提出使用人工神经网络预测气体泄爆最大超压。基于124组实验数据,采用BP与RBF神经网络,通过优化算法计算与迭代循环对泄爆样本中的影响因素进行降维与选择,并确定2类神经网络本身在学习与计算气体泄爆样本时的相关参数。结果表明:PCA（主成分分析法）在当前样本条件下的降维效果较差,而通过迭代对比确认气体泄爆样本中的5类特征全部保留时神经网络的训练模拟效果最好;通过对124组实验数据进行随机挑选训练集与测试集的训练模拟结果发现,神经网络对气体泄爆中最大超压的预测效果较好;通过对比Molkov提出的和经Fakandu等改进的NFPA 68经验公式以及2类神经网络的预测结果表明,神经网络相比于传统气体泄爆经验公式具有明显优势。     

基于蒙特卡洛模拟的多米诺事故风险量化管理*

为降低危化品相关的化工事故造成的人员伤亡和财产损失,以化工多米诺事故为研究对象,探讨由初始事故引发1个或多个次生事故的连锁反应机理与风险评估方法。提出应用蒙特卡洛模拟对多米诺事故风险进行动态量化的方法,梳理化工多米诺事故风险的识别、分析、评定、处理全周期管理流程,并以1个天然气压气站为案例,验证基于蒙卡模拟的化工多米诺事故风险量化方法的有效性。结果表明:该方法可以更准确地对化工多米诺事故风险进行定量评估。多米诺事故风险全周期管理流程的梳理能够有效指导化工企业开展安全管理、事故预防等工作。     

基于视频分析的人群密集场所客流监控预警研究

为实现人群密集场所客流安全隐患早发现,辅助管理人员早决策,人群聚集风险区早疏散,提升对灾难的预见性和主动性。在国内外人群异常聚集监测预警现状分析基础上,对比分析得出监控视频分析技术是解决人群密集场所精准预警难题较为理想的解决方案;构建以视频智能分析的人群计数、密度估计、行人追踪、活动烈度识别为核心技术的人群密集场所风险预警技术框架;将该技术框架应用到某大型商圈的商业街区,获得监控区域内的人群总数、密度分布、行人轨迹和异常活动等特征。结果表明:提出的基于视频分析的人群密集场所风险预警技术框架可为城市大型商圈、交通枢纽、大型活动场所等城市公共场所的安全管理提供参考和借鉴。     

波浪与抛石潜堤相互作用的MPS法数值模型研究

为了研究波浪与抛石潜堤相互作用过程中大自由表面变形和堤内渗流等强非线性紊流运动问题,利用改进的MPS法,建立了模拟波浪与抛石潜堤相互作用的MPS法数值计算模型。模型将抛石潜堤假定为均质多孔介质,采用Drew的二相流运动方程描述多孔介质内外的流体运动;通过在动量方程中增加非线性阻力项,并引入亚粒子尺度紊流模型,模拟波浪与可渗结构物相互作用过程中的紊流运动。选取“U”型管中多孔介质内渗流过程和孤立波与可渗潜堤相互作用两个典型的渗流问题,通过将数值计算结果与理论解和实测值的对比分析,对所提出的MPS法紊流渗流模型的模拟精度进行验证。结果表明：基于改进的MPS法构建的垂向二维紊流渗流模型可以很好地再现 “U”型管中多孔介质内渗流以及波浪作用下可渗潜堤内外的复杂流场,显著缓解流-固界面处的压力震荡与粒子分布不均匀问题,实现了较高的模拟精度。     

Rainfall Interception Based on Indirect Methods: A Case Study in Temperate Forests in Oaxaca,Mexico

Rainfall interception represents the amount of water trapped in natural cover that is not drained directly to the ground. Intercepted rainfall may evaporate after a rain event, making it one of the main drivers of water balance and hydrologic regionalization. This process can be affected by factors such as climate, altitude, vegetation type, and topography. Here is a simple method of calculating rainfall interception in temperate forests using in Santa Maria Yavesia, Oaxaca, and Mexico as an illustrative study area. We used two rain gauges to measure net precipitation (N_p) under the canopy at each study site and one gauge outside the canopy to obtain gross precipitation (G_p). Throughfall (T_h) was indirectly measured using hemispherical photographs. Rainfall interception was obtained through a combination T_h and G_p and N_p. The mean rainfall interception was 50.6% in the Abies forests, 23%–40% in the coniferous‐mixed forests, and 27.4% in the broad‐leaved forests. We classified rainfall events by intensity to determine the effect of canopy structure and precipitation and found that 75% of the events were weak events, 24% were moderate events, and 1% were strong events. In addition, we found that rainfall interception was lower when the intensity of precipitation was higher. Our method can be replicated in different ecosystems worldwide as a tool for assessing the influence of rainfall interception in terms of ecological services.     

Sulfur cycle as an electron mediator between carbon and nitrate in a constructed wetland microcosm

• Fe(III) accepted the most electrons from organics, followed by NO₃^‒, SO₄^2‒, and O₂. • The electrons accepted by SO₄^2‒ could be stored in the solid AVS, FeS₂-S, and S⁰. • The autotrophic denitrification driven by solid S had two-phase characteristics. • A conceptual model involving electron acceptance, storage, and donation was built. • S cycle transferred electrons between organics and NO₃^‒ with an efficiency of 15%. A constructed wetland microcosm was employed to investigate the sulfur cycle-mediated electron transfer between carbon and nitrate. Sulfate accepted electrons from organics at the average rate of 0.84 mol/(m³·d) through sulfate reduction, which accounted for 20.0% of the electron input rate. The remainder of the electrons derived from organics were accepted by dissolved oxygen (2.6%), nitrate (26.8%), and iron(III) (39.9%). The sulfide produced from sulfate reduction was transformed into acid-volatile sulfide, pyrite, and elemental sulfur, which were deposited in the substratum, storing electrons in the microcosm at the average rate of 0.52 mol/(m³·d). In the presence of nitrate, the acid-volatile and elemental sulfur were oxidized to sulfate, donating electrons at the average rate of 0.14 mol/(m³·d) and driving autotrophic denitrification at the average rate of 0.30 g N/(m³·d). The overall electron transfer efficiency of the sulfur cycle for autotrophic denitrification was 15.3%. A mass balance assessment indicated that approximately 50% of the input sulfur was discharged from the microcosm, and the remainder was removed through deposition (49%) and plant uptake (1%). Dominant sulfate-reducing (i.e., Desulfovirga, Desulforhopalus, Desulfatitalea, and Desulfatirhabdium) and sulfur-oxidizing bacteria (i.e., Thiohalobacter, Thiobacillus, Sulfuritalea, and Sulfurisoma), which jointly fulfilled a sustainable sulfur cycle, were identified. These results improved understanding of electron transfers among carbon, nitrogen, and sulfur cycles in constructed wetlands, and are of engineering significance.