基于距离相关系数和支持向量机回归的PM2.5浓度滚动统计预报方案

针对目前空气质量统计预报方法存在的主要缺陷,本文提出了距离相关系数和支持向量机回归相结合的统计预报方案DC-SVR.利用淮安市2013年1—12月PM_(2.5)观测资料和常规气象观测资料,首先在选入预报当日气象要素的基础上,增加选取前期污染物和气象要素作为预报因子,再采用距离相关系数分季节从预报因子中筛选出重要预报因子,最后采用支持向量机回归对PM_(2.5)浓度值进行逐日滚动统计预报.研究发现,淮安地区气温和气压对PM_(2.5)的距离相关性要高于其他气象要素,夏秋季PM_(2.5)与气象要素的距离相关性较春冬季好.基于距离相关系数和支持向量机回归建立DC-SVR模型,PM_(2.5)的试预报值和实测值的全年相关系数高达0.76,平均偏差仅为1.13μg·m~(-3),平均绝对误差为23.47μg·m~(-3).通过与支持向量机回归、人工神经网络的统计预报效果对比,DC-SVR模型有效降低预报因子维数且能自适应选取最佳参数,预报精度显著优于其他3种统计预报方案,可为业务化预报提供参考.     

基于GIS和LUR模型的西安市PM2.5浓度空间分布模拟研究

基于地理信息系统GIS和土地利用回归LUR模型,模拟西安市PM2.5浓度空间动态分布,结果表明:与PM2.5浓度相关性最高的分别为缓冲区为2 km的水域面积、人口密度和距离水域距离,R 2分别为0.501,0.393和0.280;与PM2.5浓度相关性最低的分别为缓冲区为4 km的水域面积、未利用地面积和耕地面积,R 2分别为0.039、0.021和0.017.未考虑风速建立的LUR模型多元回归的相关系数为0.856,R 2为0.733,考虑风速的相关系数为0.892,R 2为0.796,表明风速对于污染物的分布影响较大,LUR模型模拟效果较好.模拟的PM2.5年均浓度高风险区分布于中部,中风险区分布于中西部,低风险区分布于东南部和西部.     

ABC算法优化SVR的磨损故障预测模型

目的为了提高故障预测的精度,针对支持向量回归SVR(Support vector machine for regression,SVR)参数选择困难的问题,提出一种采用人工蜂群(artificial bee colony,ABC)算法优化支持向量回归(SVR)的故障预测模型(ABC-SVR)。方法该模型先对样本数据进行重构,然后将故障预测误差(适应度)作为优化目标,通过ABC算法寻优找到最优的SVR参数,建立故障预测模型。最后通过实例仿真验证模型的优越性。结果采用ABC算法优化的SVR故障预测模型进行时间序列预测,能够较好地跟踪发动机滑油金属元素浓度的变化过程,并且能够提前2个取样时间预测异常情况的出现。结论 ABC-SVR模型有效解决了SVR参数选择难题,能够更加准确地表现故障变化规律,提高了故障预测精度。     

基于增强回归树的城市PM2.5日均值变化分析:以常州为例

利用2014年12月至2015年11月常州市区6个国控监测站空气污染物浓度逐时数据,分析了PM_(2.5)浓度季节变化特征,采用增强回归树模拟分析了PM10、4种气态污染物和7个气象因子对ρ(PM_(2.5))日变化的贡献.结果表明,常州市区PM_(2.5)污染季节差异明显,冬季污染严重且持续时间长,夏季污染较轻.四季ρ(PM_(2.5))空间分布特征存在一定差异,但各季内不同监测站差异较小.增强回归树对ρ(PM_(2.5))日均值进行模拟和验证得到,训练数据的相关性为0.981,交叉验证的相关性为0.957.此外,模拟值与实测值的标准化平均偏差为1.80%,标准化平均误差为10.41%,可见模型拟合效果较好.PM10、气态污染物、气象因子和区域输送及扩散这4种影响类型对全年ρ(PM_(2.5))日均值差异的贡献率分别为23.4%、28%、36.2%和12.6%,表明在对ρ(PM_(2.5))日均值差异的影响上,气象因子二次形成一次源区域输送及扩散.在对ρ(PM_(2.5))日均值差异贡献率大于5%的因子中,ρ(PM_(2.5))日均值与PM10、相对湿度、CO和O3正相关,与温度、SO2和混合层高度负相关,与大气压和NO2关系较复杂.区域输送及扩散方面,东南风向、偏西风向和偏北风向等上风向周边城市的污染物输送对常州市区PM_(2.5)污染存在较大的负面影响.     

川东北低山丘陵区细颗粒物主要相关因素和季节变化分析——以南充主城区为例

基于中国空气质量在线监测分析平台和全球天气精准预报网的大气质量和气象数据,以四川盆地东北低山丘陵区典型城市南充市主城区为例,检验了细颗粒物(PM_(2.5))浓度的概率密度分布,发现其接近对数正态分布,由相关分析确定了PM_(2.5)浓度的主要相关因素为CO、NO_2(相关系数r分别为0.76、0.55,P0.01),再通过对2014年1月—2016年6月的日数据的逐步回归筛选出最优的回归指标和模拟方程(决定系数R_(adj)~2为0.68,P0.05),2016年7月—2017年6月的数据验证表明模拟效果较好(拟合优度为0.64,相对误差15.48%);最后根据时序插值、浓度和IAQI(PM_(2.5))的时段均值发现PM_(2.5)浓度在年际上有降低趋势;在季节上由高到低依次为冬季、春季、秋季、夏季;PM_(2.5)浓度在1月和6月分别呈现出年内的峰值和谷值,5、10月出现了阶段性峰值,尤其是5月;IAQI(PM_(2.5))的季节变化与浓度变化规律相似;且PM_(2.5)与PM_(10)比值的均值为0.67,表明现阶段南充市主城区大气污染物中细颗粒物占有较大比重。     

基于地理加权回归模型评估土地利用对地表水质的影响

针对传统线性回归模型大多忽视空间数据局部变化特征这一缺陷,引入地理加权回归模型(GWR)用于评估土地利用对地表水质的影响,分析了不同子流域内两者关系出现空间变化的规律并阐释了原因.同时,对比了GWR模型与普通最小二乘模型(OLS)的校正R2、Akaike信息准则(AICc)及残差的空间自相关指数(Moran's I),验证了GWR模型在预测精度和处理空间自相关过程中是否优于OLS模型.结果表明,同一土地利用类型对水质的影响随空间位置的改变而发生方向或大小的变化.以温瑞塘河流域总氮(TN)与农用地的关系为例,从GWR模型局部回归系数的方向分析,两者关系表现为农村正、城区负的现象,从大小分析,旧城区TN与农用地回归系数的绝对值高于其它区域;在溶解氧(DO)与人口密度所构建的GWR模型中,两者关系在整个研究区域内均表现为负值,与OLS结果吻合,从回归系数的大小分析,人口密度对DO的作用在郊区及农村更为显著.针对此类关系出现空间变化的原因分析表明,相邻子流域土地利用百分比的改变及水体主要污染源的不同,是导致土地利用对水质作用发生变化的根本因素.最后,对比所构建的80个GWR与OLS模型校正R2、AICc指标,验证了GWR作为一种局部统计模型,其预测精度优于OLS等传统全局模型且更能反映实际空间特征.     

回采工作面瓦斯涌出量遗传投影寻踪回归预测

为快速、准确预测回采工作面瓦斯涌出量,基于投影降维思想,建立一种遗传算法(GA)投影寻踪回归预测方法。选取煤层瓦斯原始含量、埋藏深度、煤层厚度、煤层倾角、工作面长度、推进速度、采出率、临近层瓦斯含量、临近层厚度、临近层层间距、岩层岩性、开采深度作为评价因子,对某矿15个学习样本进行训练,建立GA投影寻踪回归预测模型。利用该矿3个实测样本对模型进行检验,并与主成分分析和BP神经网络方法结果进行对比。研究表明:利用GA投影寻踪回归预测回采工作面瓦斯涌出量,平均误差为3.43%,最大误差为5.7%,精度优于其他2种方法。     

On-Scene Injury Severity Prediction (OSISP) Algorithm for Truck Occupants

Objective: The aim of this study is to develop an on-scene injury severity prediction (OSISP) algorithm for truck occupants using only accident characteristics that are feasible to assess at the scene of the accident. The purpose of developing this algorithm is to use it as a basis for a field triage tool used in traffic accidents involving trucks. In addition, the model can be valuable for recognizing important factors for improving triage protocols used in Sweden and possibly in other countries with similar traffic environments and prehospital procedures.

Methods: The scope is adult truck occupants involved in traffic accidents on Swedish public roads registered in the Swedish Traffic Accident Data Acquisition (STRADA) database for calendar years 2003 to 2013. STRADA contains information reported by the police and medical data on injured road users treated at emergency hospitals. Using data from STRADA, 2 OSISP multivariate logistic regression models for deriving the probability of severe injury (defined here as having an Injury Severity Score [ISS] > 15) were implemented for light and heavy trucks; that is, trucks with weight up to 3,500 kg and ??16,500 kg, respectively. A 10-fold cross-validation procedure was used to estimate the performance of the OSISP algorithm in terms of the area under the receiver operating characteristic curve (AUC).

Results: The rate of belt use was low, especially for heavy truck occupants. The OSISP models developed for light and heavy trucks achieved cross-validation AUC of 0.81 and 0.74, respectively. The AUC values obtained when the models were evaluated on all data without cross-validation were 0.87 for both light and heavy trucks. The difference in the AUC values with and without use of cross-validation indicates overfitting of the model, which may be a consequence of relatively small data sets. Belt use stands out as the most valuable predictor in both types of trucks; accident type and age are important predictors for light trucks.

Conclusions: The OSISP models achieve good discriminating capability for light truck occupants and a reasonable performance for heavy truck occupants. The prediction accuracy may be increased by acquiring more data. Belt use was the strongest predictor of severe injury for both light and heavy truck occupants. There is a need for behavior-based safety programs and/or other means to encourage truck occupants to always wear a seat belt.     

Estimated Injury Risk for Specific Injuries and Body Regions in Frontal Motor Vehicle Crashes

Objective: Injury risk curves estimate motor vehicle crash (MVC) occupant injury risk from vehicle, crash, and/or occupant factors. Many vehicles are equipped with event data recorders (EDRs) that collect data including the crash speed and restraint status during a MVC. This study's goal was to use regulation-required data elements for EDRs to compute occupant injury risk for (1) specific injuries and (2) specific body regions in frontal MVCs from weighted NASS-CDS data.

Methods: Logistic regression analysis of NASS-CDS single-impact frontal MVCs involving front seat occupants with frontal airbag deployment was used to produce 23 risk curves for specific injuries and 17 risk curves for Abbreviated Injury Scale (AIS) 2+ to 5+ body region injuries. Risk curves were produced for the following body regions: head and thorax (AIS 2+, 3+, 4+, 5+), face (AIS 2+), abdomen, spine, upper extremity, and lower extremity (AIS 2+, 3+). Injury risk with 95% confidence intervals was estimated for 15–105 km/h longitudinal delta-Vs and belt status was adjusted for as a covariate.

Results: Overall, belted occupants had lower estimated risks compared to unbelted occupants and the risk of injury increased as longitudinal delta-V increased. Belt status was a significant predictor for 13 specific injuries and all body region injuries with the exception of AIS 2+ and 3+ spine injuries. Specific injuries and body region injuries that occurred more frequently in NASS-CDS also tended to carry higher risks when evaluated at a 56 km/h longitudinal delta-V. In the belted population, injury risks that ranked in the top 33% included 4 upper extremity fractures (ulna, radius, clavicle, carpus/metacarpus), 2 lower extremity fractures (fibula, metatarsal/tarsal), and a knee sprain (2.4–4.6% risk). Unbelted injury risks ranked in the top 33% included 4 lower extremity fractures (femur, fibula, metatarsal/tarsal, patella), 2 head injuries with less than one hour or unspecified prior unconsciousness, and a lung contusion (4.6–9.9% risk). The 6 body region curves with the highest risks were for AIS 2+ lower extremity, upper extremity, thorax, and head injury and AIS 3+ lower extremity and thorax injury (15.9–43.8% risk).

Conclusions: These injury risk curves can be implemented into advanced automatic crash notification (AACN) algorithms that utilize vehicle EDR measurements to predict occupant injury immediately following a MVC. Through integration with AACN, these injury risk curves can provide emergency medical services (EMS) and other patient care providers with information on suspected occupant injuries to improve injury detection and patient triage.     

基于二值逻辑的复合绝缘子状态判别策略研究

复合绝缘子是架空输电线路上广泛使用的设备,它的老化会给电力系统的正常运行带来威胁。电力相关工作人员需要一种能够对复合绝缘子的运行状态进行判别的方法,帮助他们确定复合绝缘子是否满足运行要求,而对复合绝缘子的定性表征就成为一种优选的表征策略。将运行后的复合绝缘子分成"满足运行要求"和"不满足运行要求",使用二值逻辑回归模型,选取合适的特征量,建立了复合绝缘子状态表征体系。