摘要 结冰风险路段的路面抗滑能力大幅度降低，行车条件恶化，给高速公路交通安全和公路网的正常运行带来极大的安全隐患。在交通气象预报（包括当前交通和气象跨部门合作开展的公路交通气象预报）和高速公路运营管理与控制方面，关于路面结冰状态监测预警，特别是不易被人肉眼所察觉的路面黑冰状态，尚缺少有效的技术手段。虽然某些高速公路已经布设了路面状态传感器，用以实时监测固定位置的路面温度和路面状态，鉴于传感器数量有限，不可能针对路段上可能存在的结冰风险进行长距离精细化的预测预警。本文以实现未来短时间段内，针对高速公路连续路段中潜在结冰点段进行准确预测、研判和预警为目标，围绕高速公路路面结冰状态关键控制因子确定、移动式交通气象环境监测系统研发、连续路面温度观测与空间分布特征分析、定点路面温度时间序列预测、结冰风险路段动态短临预警等关键内容开展研究。主要研究内容包括以下几个方面： （1）高速公路路面结冰状态关键控制因子确定 分析高速公路路面结冰特征，通过室内外试验确定路面结冰状态的关键控制因子，即导致路面结冰的关键参数。 （2）移动式交通气象环境监测系统研发 研发移动式交通气象环境监测系统，针对高速公路沿线现有固定式交通气象监测站所在路段，实现路面结冰关键参数的连续观测。 （3）高速公路路面温度观测与空间分布特征分析 提取研究路面结冰核心关键参数——路面温度的观测方法，研究不同天气条件下高速公路路面温度数据的时空修正处理方法，挖掘移动监测路段范围内连续路面温度空间分布特征，分析路段中不同位置的相对路面温度差异。 （4）高速公路定点路面温度时间序列预测建模 基于固定式交通气象监测站的历史监测数据，分析路面温度的主要影响因素，应用三种不同的预测方法建模分析，对未来短时间内的路面温度进行预测，综合评判选取最优的定点路面温度时间序列预测模型。 （5）高速公路路面结冰风险路段动态短临预警 形成高速公路路面温度与状态的时空维预测技术，对固定交通气象监测站位置以及路段上其它位置的路面状态及最早结冰时刻进行预测，提出结冰风险路段预警分级标准，提高高速公路沿线路面结冰状况早期预测预警的精细化水平。 论文研究取得的主要成果如下： （1）研发了高速公路移动式交通气象环境监测系统，实现了由有限的固定式交通气象监测站拓展到高速公路全路段的连续监测方法；其综合成本低、安装简单、携带方便，能够有效解决高速公路沿线路面结冰状态关键参数的连续采集、分析和工程应用等问题。 （2）建立了高速公路空间路面温度数据采集、融合处理、描述展现等步骤在内的一整套流程和方法。基于沈阳三环高速公路的实测数据，获得了冬季三种典型天气条件下（极端、中间、抑制）20米间隔分辨率的不同位置与固定交通气象站点的相对路面温度差异分布。 （3）确定了路面温度的主要影响因素为空气温度、露点温度、相对湿度和风速。基于这些外部变量，应用ARIMA、DECISION TREE和ANFIS等方法建立路面温度时间序列预测模型。误差分析表明基于ARIMA的预测模型最优，路面温度允许绝对误差在±0.5℃和±1℃范围内，未来3小时的平均预测准确率分别达到81.25%和99.65%，对应的平均绝对误差分别为0.21℃和0.26℃。其中在±0.5℃允许绝对误差范围内，未来1小时的预测准确率和平均绝对误差分别达到92.5%和0.15℃。 （4）提出了耦合连续路段路面温度空间分布特征和定点短时路面温度时间序列预测结果的连续路面温度短临预测方法，结合潜在路面结冰状态判断条件，可实现24小时滚动周期预测未来1小时的每分钟路面结冰状态和早期预警输出。 （5）在以上研究成果的基础上，建立了我国高速公路结冰风险路段动态短临预警技术，从观测仪器参数设置、数据采集和分析、结冰风险路段预警等方面给出一套工程可实际应用的技术流程，实现高速公路路段上黑冰、冰霜等路面结冰状态预测的整体准确率达到80%以上，验证了本文提出的高速公路路面结冰风险路段动态短临预警方法能够满足实际推理预测的要求。 本论文的研究成果可在我国现有高速公路路面状态传感器稀缺，并对黑冰、冰霜等路面结冰状况检测能力不足的条件下，推动交通气象环境监测的智能化发展和应用，提高高速公路结冰风险路段短时间内连续滚动预测预警的时空精细化水平，为提升冬季结冰风险路段下的高速公路运营管理、养护决策和应急服务水平提供了强有力的理论依据和技术支持。 关键词：高速公路，结冰风险路段，路面温度，路面状态，移动交通气象监测系统，相对路面温度空间分布，时间序列预测模型，预警，动态短临 ABSTRACT Pavement anti-slide capacity of icing risk sections is greatly reduced, which leads to driving conditions deterioration, and brings great hidden danger to the traffic safety and normal operation of expressway network. Regardless of the traffic weather forecast (including highway traffic weather forcast carried out by transport and meteorological departments) and expressway operation management and control, all have not effective technical means to monitor or early warning on icy pavement state, especially a black ice condition. Although some expressways have been set up few pavement state sensors, used for real-time monitoring pavement temperature and conditions of fixed positions, it is impossible that the icing risks on road sections are forecast finely over long distances. In order to achieve accurate prediction, analyzing and warning on potential icing sections in a short period of time in the future, this paper is focus on determining key controlling factors of pavement icing state, research and development of mobile traffic meteorological monitoring system, continuous pavement temperature observation and spatial distribution characteristics, fixed-point time series prediction of pavement temperature, and dynamic short-impending pre-warning of icing risk sections. The main research contents include the following aspects: (1) Key control factors of expressway pavement icing state By analyzing pavement icing characteristics, key control factors of the icing state are determined through indoor and outdoor tests. (2) Development of mobile traffic meteorological monitoring system Through the mobile traffic meteorological monitoring system, the continuous observation of key parameters of the icing state were realized. (3) Expressway pavement temperature observation and spatial distribution analysis Studying a pavement temperature observation method, and a temporal and spatial data processing method under different weather conditions, this paper mined the spatial distributions characteristics of continuous pavement temperature within mobile monitoring sections, and analyzed relative pavement temperature difference of different positions. (4) Prediction model of pavement temperature time series Analysis on the main affecting factors of pavement temperature, this paper predicted pavement temperature in a short period of time in the future by three different prediction methods, and selected the optimal prediction model by comprehensive evaluation. (5) Dynamic short-impending pre-warning of icing risk sections Based on the spatial and temporal dimension prediction technology of pavement temperature and state, the road surface state and the earliest freezing time were predicted in different locations. Combined with early warning grading standards of icing risk sections, the fine forecasting and warning level of road icing status could be improved. This research showed that: (1) The developed mobile traffic meteorological monitoring system realized a continuous monitoring method for the whole section by expanding those limited fixed traffic weather stations, which was low cost and simple installation and convenient carrying, and could solve those problems of continuous collection, analysis and engineering applications of pavement icing state. (2) A set of processes and methods were established for data collection, fusion processing, and descriptive presentation of the spatial pavement temperature. Based on the real data of Shenyang Ring Expressway, the relative pavement temperature distribution at 20 meters intervals between different locations and the fixed traffic weather station was obtained under three typical weather conditions (extreme, middle, damped) in winter. (3) The main affecting factors of road surface temperature were determined, including air temperature, dewpoint temperature, relative humidity and wind speed. Based on these external variables, the prediction models of pavement temperature time series were established by applying ARIMA, DECISION TREE,and ANFIS methods. The error analysis showed that ARIMA model was the most optimal, whose average prediction accuracy rate reached 81.25% and 99.65% respectively within the range of ±0.5℃ and ±1℃ absolute-error measure in the next 3 hours. The prediction accuracy and absolute error were up to 92.5% and 0.15℃ respectively within the range of ±0.5℃ absolute-error measure in the next 1 hours. (4) A short-impending precasting method of continuous pavement temperature was presented, by coupling between spatial distribution characteristics and short-term time series. Combined with judging conditions of the potential road icing state, road icing status and early warning output per minute were achieved on a 24 hours rolling forecast period in the next 1 hours. (5) Based on the above research results, a dynamic short-impending pre-warning technology of icing risk sections was established. From observation instrument parameters setting, data collection and analysis, and icing risk sections warning, a set of technical process was given for practical engineering application, which may help to realize that the overall prediction accuracy rate was more than 80% for black ice and frost. This result validated the method can meet practical predictive requirements. Under the conditions of pavement state sensors are scarce, and their detection abilities are insufficient for black ice, frost and other pavement icing states, research results can promate intelligent development and application of traffic weather monitoring, and improve the fine forecasting and warning level of road icing status in a short time, all which will provide the strong theoretical basis and technical support for enhancing the levels of operation management, maintenance decision and emergency service of icing risk sections in winter. Key words: expressway, icing risk section, pavement tempreture, pavement state, mobile traffic meteorological monitoring system, relative pavement temperature spatialdistribution, time series prediction model, pre-warning, dynamic short-impending |
博士论文