基于多源数据的PM2.5浓度时空分布预测与制图

随着我国经济、工业化、城市化进程迅速发展,PM_(2.5)污染在中国已经成为一个极端的环境和社会问题,并引起广泛关注.采用新技术估算的地表PM_(2.5)质量浓度,收集并处理了遥感反演的气溶胶光学厚度(AOD),气象数据,其他地理数据和污染物排放数据,采用贝叶斯最大熵(BME)结合地理加权回归(GWR)来分析2015年冬季的PM_(2.5)暴露在我国东部大范围区域的时空变异特征.结果表明,BME模型的十折交叉验证结果的决定系数R~2为0.92,均方根误差(RMSE)为8.32μg·m~(-3),平均拟合误差(MPE)为-0.042μg·m~(-3),平均绝对拟合误差(MAE)为4.60μg·m~(-3),与地理加权回归模型的结果相比(R~2=0.71,RMSE=15.68μg·m~(-3),MPE=-0.095μg·m~(-3),MAE=11.14μg·m~(-3)),BME的预测结果有极大的提高.空间上,PM_(2.5)高浓度地区主要集中在华北、长江三角洲、四川盆地,低浓度地区主要集中在中国的最南部如珠江三角洲和云南的西南部;时间上,不同月份的研究区域PM_(2.5)空间分布所有差别,2015年的12月、2016年1月PM_(2.5)污染最为严重,2015年的11月,2016年的2月污染相对较低.     

Fish forewarning of comprehensive toxicity in water environment based on Bayesian sequential method

Environmental impact of pollutants can be analyzed effectively by acquiring fish behavioral signals in water with biological behavior sensors. However, a variety of factors, such as the complexity of biological organisms themselves, the device error and the environmental noise, may compromise the accuracy and timeliness of model predictions. The current methods lack prior knowledge about the fish behavioral signals corresponding to characteristic pollutants, and in the event of a pollutant invasion, the fish behavioral signals are poorly discriminated. Therefore, we propose a novel method based on Bayesian sequential, which utilizes multi-channel prior knowledge to calculate the outlier sequence based on wavelet feature followed by calculating the anomaly probability of observed values. Furthermore, the relationship between the anomaly probability and toxicity is analyzed in order to achieve forewarning effectively. At last, our algorithm for fish toxicity detection is verified by integrating the data on laboratory acceptance of characteristic pollutants. The results show that only one false positive occurred in the six experiments, the present algorithm is effective in suppressing false positives and negatives, which increases the reliability of toxicity detections, and thereby has certain applicability and universality in engineering applications.     

基于贝叶斯网络的社区配电网系统动态风险评估

社区配电网系统的安全分析和动态风险评估是配电网系统安全运行的关键环节。传统的可靠性指标无法反映社区配电网系统中物理元器件的状态和短路故障事件等本质安全问题对风险评估的影响。针对现有风险评估方法的局限性,提出一种基于贝叶斯网络的社区配电网系统安全分析与动态风险评估框架。首先,在标准可靠性指标的基础上,创新性地将社区配电网系统中物理元器件的故障率和短路故障事件引入到风险评估中,提出一种新的社区配电网停电风险评估指标体系,该评估指标体系综合考虑了社区配电网的可靠性指标和本质安全特性,能够全面反映社区配电网系统的动态风险;其次,提出一种基于故障树-贝叶斯网络的社区配电网系统动态风险评估方法;最后,以一个实际社区的配电网系统为案例,验证了该动态风险评估框架的可行性和有效性。案例分析结果表明:所提出的框架能够充分评估社区配电网系统的停电风险,并可以给出合理的安全措施,以有效降低社区配电网系统的动态风险。     

基于贝叶斯公式的不确定性水环境容量研究——以北运河为例

针对水环境的随机不确定性,构建了基于贝叶斯公式的不确定性水环境容量核算方法,并以北运河为实证案例对所建立的核算方法进行验证：利用贝叶斯公式,通过对水环境容量核算涉及的流量、流速、背景浓度、污染物综合衰减系数这4个参数和水环境容量本身5个随机变量的分布特征进行分析,最后得到水环境容量的概率分布.结果表明：北运河流域10个河段的水环境容量都服从瘦尾偏态分布,并且河段流量与长度对其分布影响最大;与传统设计条件下的确定性水环境容量相比较,不确定性水环境容量分布不仅能够体现前者的信息,并且利用各参数的随机性以及实际监测数据进一步对前者进行了“修正”,能够更为全面准确的描述水环境.     

A method for simplifying the analysis of traffic accidents injury severity on two-lane highways using Bayesian networks

Introduction

This study describes a method for reducing the number of variables frequently considered in modeling the severity of traffic accidents. The method's efficiency is assessed by constructing Bayesian networks (BN).

Method

It is based on a two stage selection process. Several variable selection algorithms, commonly used in data mining, are applied in order to select subsets of variables. BNs are built using the selected subsets and their performance is compared with the original BN (with all the variables) using five indicators. The BNs that improve the indicators’ values are further analyzed for identifying the most significant variables (accident type, age, atmospheric factors, gender, lighting, number of injured, and occupant involved). A new BN is built using these variables, where the results of the indicators indicate, in most of the cases, a statistically significant improvement with respect to the original BN.

Conclusions

It is possible to reduce the number of variables used to model traffic accidents injury severity through BNs without reducing the performance of the model.

Impact on Industry

The study provides the safety analysts a methodology that could be used to minimize the number of variables used in order to determine efficiently the injury severity of traffic accidents without reducing the performance of the model.     

Bayesian approach to estimating margin of safety for total maximum daily load development

A Bayesian-updating approach is presented to the estimation of total uncertainty-based Margin of Safety (MOS) for Total Maximum Daily Load (TMDL) calculations. Probability distributions are presented to construct the likelihood function, the prior probability distribution, and the posterior (total uncertainty) probability distribution. The Bayesian-updating approach is demonstrated through a case study for the Lower Amite River, Louisiana. The posterior probability distribution-based on the Bayesian approach updates the standard deviation of summer dissolved oxygen in the Amite River from 1.88 mg/L to 2.10 mg/L when the total uncertainty is considered. Results from the Bayesian-updating approach are compared with two conventional methods. The dissolved oxygen reserve based on a conventional margin of safety of 20% is estimated to be 45,682.26 kg/Day. The second conventional method, where we consider the standard deviation of 1.88 mg/L, produces a dissolved oxygen reserve of 40,516.09 kg/Day. The Bayesian approach yields the dissolved oxygen reserve of 38,614.43 kg/Day with the first level (μ-σ) of MOS, producing a deficit of 5606.65 kg/Day in dissolved oxygen. The dissolved oxygen reserve deficit increases to 23,895.13 kg/Day when the second level (μ-2σ) of MOS is used, which escalated to 42,383.52 kg/Day when the highest level (μ-3σ) of MOS is used. While the total uncertainty-based Bayesian approach is demonstrated for a TMDL development on the Amite River, the overall approach could be applied in any river system with similar available data.     

基于PSR与贝叶斯网络的非常规突发事件情景分析

基于情景分析在非常规突发事件应急决策中的重要性,为了提高情景分析的有效性、全面性,在非常规突发事件"情景"界定和演变规律分析的基础上,基于PSR模型构建非常规突发事件的"压力-状态-响应"网络表达方式;利用贝叶斯网络理论,构建非常规突发事件的情景演变分析模型;以简化后的大连输油管道爆炸事件为例,示范基于PSR模型与贝叶斯网络的非常规突发事件情景分析方法的具体流程,并对推理结果进行分析。分析结果与现实情况基本一致,证明方法的可行性和有效性。实际应用中可针对具体的事件进行"多响应"、"多路径"的推理分析,以全面、系统地分析事件情景的演变情况。     

基于贝叶斯网络的尾流事故机理分析

针对事故树分析法的局限性,在尾流事故树的基础上,建立贝叶斯网络(BN)。运用推理运算对BN进行定量分析,得出:空中交通密度太大、空中交通管制(ATC)间隔判断错误和短期冲突告警(STCA)被忽略是事故的关键致因。将针对致因提出的改进措施引入到BN中,评价相关措施的有效性。应用BN进行尾流事故的机理分析,能够以比逻辑门更好的形式表达变量间的不确定性关系,从而更加方便地找到导致事故发生的关键因素。     

Value of information analysis for adequate monitoring of carbon dioxide storage in geological reservoirs under uncertainty

As monitoring is essential for the proper management of geological storage of carbon dioxide (CO₂), the ability to value information from monitoring is indispensable to adequately design a monitoring program. It is necessary to judge whether the expected improvement in management is worth the cost of monitoring. The value of information (VOI) is closely related to the possible increase in expected utility gained by gathering the information, the concept of which can be applied to such judgement. Although VOI analysis has been extensively studied in the context of decision analysis, its application to the management of carbon dioxide capture and storage (CCS) operations is rare. This paper introduces and discusses the methodology of VOI analyses in the context of monitoring CO₂ storage. A motivating problem with discrete probabilities is used to illustrate the concept of VOI. It is demonstrated that information is not always of value; for information to be worthwhile, monitoring under uncertainty must satisfy certain conditions. This concept is then extended to continuous probability distributions. The effects of prior uncertainty and information reliability on the VOI are examined. It is shown that an excessive improvement in information accuracy yields little value and that the optimal level of reliability can be inferred. VOI analyses provide quantitative insights into the value of information-gathering activities and therefore can be an objective means to adequately design and impartially justify a monitoring program.     

Skew and attribute non-attendance within the Bayesian mixed logit model

We investigate non-attendance to choice set attributes and the accommodation of preference heterogeneity within the mixed logit model. We propose a generalisation of the mixed logit enabling the degree of skew of marginal utility distributions to be estimated. The implementation is Bayesian with the marginal likelihood used as an arbiter of model performance. We find strong evidence of skew in the distributions of marginal utilities for most attributes. Models incorporating skew are preferred in all cases. The irrelevance of an attribute to significant numbers of respondents is a possible cause of such skew. We test alternative empirical accommodations of self-reported attribute non-attendance (ANA) and continue to find strong evidence of skew in the distributions of marginal utilities even having accounted for ANA. We find that, contrary to some recent findings, respondents who report having ignored an attribute typically do indeed have a zero marginal utility for that attribute.