基于云模型和随机森林的韧性城市电网风险预警模型

相比传统电网,韧性城市电网展现出了出色的适应多种扰动和灾害的能力,但其复杂性也使得韧性城市电网风险预警面临更大的挑战,亟需大数据和机器学习等先进技术的引入。首先,构建韧性城市电网风险评估指标体系,采用主客观结合的综合赋权法对指标赋权,通过大数据技术获取的实时数据流得到韧性城市电网风险评估指标的动态权重;然后,构建韧性城市电网风险评估标准云,计算韧性城市电网风险等级隶属度,确定风险等级;最后,基于随机森林构建韧性城市电网风险预警模型,并进行算例分析,通过与其他模型对比,发现所构建的模型表现出高精度的特征。所建模型具有较好的风险预警效果,从而能够及时采取有效风险管控措施,保障韧性城市电网稳定运行。     

营配调末端融合的供电综合风险预警模型

当前供电营配末端相互独立、关联度偏低,导致其综合风险预警效果较差,为此构建一种新的供电综合风险预警模型。选取供电风险预警指标,并对其进行标准化处理。计算指标间的灰度关联度,并对其进行排序,获取供电风险事件的标量时间序列,求得风险预警的最大梯度差,根据关联指向性特征得到风险预警评估的判决式,以此构建供电风险预警模型。实验结果表明,利用该模型进行供电综合风险预警,预警结果更接近实际结果。     

智能电网的风险预警校正系统设计

为了实现对智能电网运行状态的全面评估和控制,通过对智能电网主要特点进行研究,分析了其实现的主要技术构成,据此设计了一种适合于智能电网的风险预警校正系统。介绍了风险预警校正系统根据系统风险状态进行评估校正的原理,说明和解释了该系统各个模块的功能定位,并详细分析了该系统实现的技术要点。所设计的风险预警校正系统可适应大量新能源入网的未来电网发展需要。     

基于风险综合评判的设备状态检修决策优化

针对目前在设备状态检修决策中尚未将设备状况与电网运行情况综合考虑的问题,提出一种计及设备故障风险及电网运行风险的输变电设备状态检修决策优化模型,以"综合风险费用最小"为优化目标,以同时检修、互斥检修、检修资源、电网安全等限制条件为约束。在充分考虑设备状态检修前后故障概率变化情况的基础上,采用禁忌搜索算法求解所建立的优化模型,通过风险经济性量化实现设备与电网运行风险的综合评判,得到技术经济综合最优的状态检修决策方案。对华东某地区电网的算例分析表明了所提出的优化模型及算法的有效性。     

抗短路能力不足变压器实地运行风险评估方法

目前变压器抗短路能力校核所依据的GB1094.5—2008未考虑主变运行方式,所计算的短路电流与实际运行方式下存在差异,通过分析考虑多台主变并列运行、限流电抗器等多种实际运行因素的短路电流计算模型,提出了变压器实地运行风险评估方法,并给出了风险评估实例。     

基于风险约束的风电和水电联合日前调度优化模型

为解决风电随机性对电力系统稳定运行的影响,引入梯级水电机组为风电机组提供调峰备用,构建了考虑风险约束的风电和水电联合日前调度优化模型。首先,分别构建了梯级水电机组输出功率模型、风电机组输出功率模型,以系统调度净收益期望值最大化作为目标函数,综合考虑风水电协调约束、系统功率平衡约束,构建了风水电联合调度模型。进一步,为了考虑风电不确定性给系统带来的风险因素,建立了计及风险约束的风水电联合随机优化调度模型,并采用启发式算法对模型进行线性化处理。最后,选择3个风力发电场和2个流域的7个水力发电厂组成仿真系统,对风水电联合调度优化进行算例仿真模拟。结果表明：所提出的水风电联合运行策略可以用来促进风力发电,最大化系统运营期望收益的同时最小化系统运营风险,因此,在进行风电和水电联合调度时应综合考虑内部和外部不确定性因素,以获得能够兼顾运营收益和风险的最优解。     

基于CVaR的风电并网发电风险效益分析

为解决大规模风电并网对电力系统稳定运行的影响和分析风电并网的经济效益,把条件风险价值（CVaR）作为风险计量指标,建立了风电场并网容量的模型。该模型考虑风速的随机变化及风险,并可通过线性优化简单求解。综合考虑风电的收益及成本,建立风电效益模型,得到不同风速和风险下的并网容量及经济效益。CVaR方法避免了机会约束规划等传统方法的繁琐计算,能方便求解出并网容量从而获得可靠的经济效益。利用IEEE-14节点系统进行仿真计算并对结果进行比较,验证了该方法的可行性。     

考虑风险和收益平衡的输电断面供电方案优化

基于N-1静态安全约束校验模型的传统输电断面最大输电能力(total transfer capability,TTC)计算结果所确定的电网供电方案过于保守,为此,提出考虑供电收益和风险的电网供电方案多目标优化模型,综合考虑电网的风险、经济指标,引入风险模型和获益模型,对电网供电方案进行多目标优化,并应用多目标粒子群算法和主观权重结合熵权的余弦排序进行求解与决策。实际算例的仿真结果表明该模型和方法的有效性和可行性,技术人员可根据实际需要对方案进行分析选择。     

考虑电网多元风险的输电断面动态增容调度分配方案

针对目前电网输电断面增容容量分配决策的相关问题,提出一种考虑电网多元风险的输电断面动态增容调度分配方案优化模型。该模型考虑了电网面临的多元风险约束,包括设备健康风险、气象风险和系统异常风险。其中,断面内通道的容量裕量限额是通过三大定容设备——变压器、断路器以及输电线路的动态增容计算模型得到的。所提出的考虑电网多元风险的输电断面动态增容调度分配方案,可以较好地平衡大电网电能输送中的经济性和安全性问题,为解决输电断面内动态增容分配提出了一个新的解决思路和技术方案。     

采用条件风险方法的含风电系统安全经济调度

针对风电等新能源发电和负荷的随机性,引入条件风险方法,构建了电力系统条件风险调度模型。为了量化随机性因素所引起的不确定性,以电网安全条件风险价值(conditional value-at-risk,CVaR)作为电网安全指标,取代一般调度模型中的安全约束。由于含随机性变量概率密度的函数积分计算困难,因此将电网安全CVaR函数进行变换和离散化处理,并采用蒙特卡罗模拟和解析法相结合的方法进行计算。含风电随机出力的系统算例表明,采用条件风险约束的电力系统安全调度模型可在不同的电网安全CVaR值、不同的置信水平下,获取相应的侧重经济性或安全性的系统最优调度结果。     

基于深度学习的非机动车头盔佩戴检测方法研究

近年来,由于电动车驾驶人未佩戴头盔行车导致的交通事故频频发生,造成了较大的人身伤害与损失,调查显示事故多在交通路口发生,为此,有必要开展交通路口电动车驾驶人头盔佩戴行为的监测与管控。本文利用机器视觉传感器收集大量电动车及驾驶人目标数据,制作相应的数据集,将处理后的数据集在Pytorch框架上利用改进的Yolov5神经网络进行训练,获得最优权重参数;实验对比发现,改进后的Yolov5算法对于电动车和头盔的检测精度分别达到了92%和98%,比原始神经网络的识别准确度可提高1%至2%。最终联合使用训练改进的Yolov5模型和Sort算法,在检测电动车佩戴头盔情况的同时,对其进行跟踪标号,以此实现对交通路口违规电动车驾驶行为的有效管控。     

智能汽车拟人驾驶风险量化方法研究

驾驶风险量化评估对智能汽车拟人驾驶决策至关重要,针对复杂多任务场景下的驾驶风险量化问题,提出了一种基于 人类风险感知机理的智能汽车驾驶风险量化方法。 首先,利用传感器获取驾驶场景周围环境信息与行驶状态信息,并根据人类 驾驶经验对潜在冲突因素赋值代价,生成驾驶场景代价地图;其次,根据车辆运动状态与拟人驾驶的基本原则,利用高斯函数建 立动态风险模型;最后,结合驾驶场景代价图与动态风险模型实时计算拟人驾驶风险量化值。 仿真结果表明,提出的方法能够 基于人类驾驶经验,计算出动态变化的驾驶风险量化值,应用于智能汽车自动驾驶决策,可产生拟人驾驶行为。     

A Multilevel Collision Mitigation Approach—Its Situation Assessment, Decision Making, and Performance Tradeoffs

This paper deals with the problem of decision making in the context of forward collision mitigation system design. The authors present a multilevel collision mitigation (CM) approach that allows a flexible tradeoff between potential benefit and the risk associated with driver acceptability and product liability. Due to its practical relevance, algorithms that allow for an efficient incorporation of both sensor and prediction uncertainties are further outlined. The performance tradeoffs that come along with different parameterizations are investigated by means of stochastic simulations on three dangerous traffic situations, namely 1) rear-end collisions due to an unexpected braking, 2) cutting-in vehicles, and 3) crossing traffic at intersections. The results show that an overly conservative CM system sacrifices much of its potential benefit. However, it is pointed out that the vision of accident-free driving can be achieved only through cooperative driving strategies     

Characterization of Driving Behaviors Based on Field Observation of Intersection Left-Turn Across-Path Scenarios

There have been significant research and developments in recent years for intersection-safety solutions that are intended to alert drivers of hazardous situations by utilizing sensing, computing, and communication technologies. Since the effectiveness of intersection-safety systems depends strongly on driver perception and acceptance of the provided warning signal, the understanding of driver actions under the targeted scenario is a central research topic. One significant safety concern at intersections is the left-turn crossing-path scenarios, where a left-turning vehicle is confronted by oncoming traffic. This paper describes the analysis and synthesis of real-world data for such scenarios observed in field observations. Specifically, traffic interactions in left-turn across-path situations are evaluated to compare data from various intersections with different operation and traffic attributes. The analyzed data were characterized to gain insight into a time gap acceptance exhibited by a population of drivers. The knowledge of driving behaviors can provide the guidelines for future investigation as well as a knowledge basis for the selection of warning criteria to allow timely alerts to drivers in the intended safety applications.     

Urban traffic control structure based on hybrid Petri nets

An urban network of signalized intersections can be suitably modeled as a hybrid system, in which the vehicle flow behavior is described by means of a time-driven model and the traffic light dynamics are represented by a discrete event model. In this paper, a model of such a network via hybrid Petri nets is used to state and solve the problem of coordinating several traffic lights with the aim of improving the performance of some classes of special vehicles, i.e., public and emergency vehicles. The proposed model has been validated using real traffic data relevant to the city of Torino, Italy. Some relevant experimental results are reported and discussed.     

Real-Time Traffic Simulation With a Microscopic Model

This paper describes a microscopic model that is able to simulate traffic situations in an urban environment in real time for use in driving simulators. Two types of vehicles are considered in the simulation, namely the user-driven vehicle at the center of the simulation model and the other vehicles that interact with it and its surroundings, which configure the developed traffic model. Simulation is performed in a reduced zone, called the control zone, surrounding the user-driven vehicle. This control zone is a mobile zone centered on the user-driven vehicle. The size of the control zone depends on the maximum number of vehicles involved simultaneously, the traffic density, and the driver's limit of visibility. The other vehicles involved in the traffic simulation are created or destroyed within the limits of the control zone. The general behavior of the traffic model is based on the following theory. Vehicles have an associated driver model that establishes several control functions for them to follow the path, while the steering, acceleration, and braking maneuvers follow certain models of behavior. A traffic light regulation is also included but only in the control area. The possibility of introducing anomalous traffic situations into the simulation is also considered, such as the presence of obstacles, abnormal maneuvers, etc. The developed model is immediately applicable to large-scale driving simulators for driver training, traffic control studies, and safety studies     

基于动态交通信息的电动汽车充电负荷时空分布预测

电动汽车充电负荷预测是研究电动汽车与电网互动的重要前提。针对交通路网信息对电动汽车行驶规律的影响,考虑电动汽车的交通工具特性和移动负荷特性,提出了一种基于动态交通信息的电动汽车充电负荷时空分布预测方法。该方法首先针对城市路网多交叉口特征,提出建立考虑路段阻抗和节点阻抗的动态路网模型。并根据路网规模确定了相应的交通网-配电网的交互模型。其次引入OD矩阵分析方法和实时Dijkstra动态路径搜索算法为电动汽车分配起止节点和规划行驶路径,模拟其动态行驶过程和充电行为。最后设计了电动汽车路径规划实验和典型区域实际路网充电负荷预测实验。结果表明,电动汽车充电负荷在不同功能区域分布存在差异且时间分布上不均匀,验证所提方法的有效性和可行性。     

DGPS-Based Vehicle-to-Vehicle Cooperative Collision Warning: Engineering Feasibility Viewpoints

The vehicle collision warning system (CWS) is an important research and application subject for vehicle safety. Most of this topic's research focuses on autonomous CWSs, where each vehicle detects potential collisions based entirely on the information measured by itself. Recently, an alternative scenario has arisen. This scenario is known as cooperative driving, where either the vehicle or the infrastructure can communicate its location, intention, or other information to surrounding vehicles or nearby infrastructure. Since installing a low-cost global-positioning-system (GPS) unit is becoming a common practice in vehicle applications, its implications in cooperative driving and vehicle safety deserve closer investigation. Furthermore, the future trajectory prediction may lead to a straightforward approach to detect potential collisions, yet its effectiveness has not been studied. This paper explores the engineering feasibility of a future-trajectory-prediction-based cooperative CWS when vehicles are equipped with a relatively simple differential GPS unit and relatively basic motion sensors. The goals of this paper are twofold: providing an engineering argument of possible functional architectures of such systems and presenting a plausible example of the proposed future-trajectory-based design, which estimates and communicates vehicle positions and predicts and processes future trajectories for collision decision making. In this paper, common GPS problems such as blockage and multipath, as well as common communication problems such as dropout and delays, are assumed. However, specific choices of GPS devices and communication protocol or systems are not the focus of this paper     

基于交通均衡的电动汽车快速充电负荷模拟

电动汽车的发展普及使得城市交通系统和电网紧密耦合,二者的交互影响日渐显现。针对传统充电负荷分析未能有效考虑交通网络特性的不足,该文建立了基于交通均衡的充电负荷模拟方法。首先,考虑用户快速充电决策过程,提出快速充电负荷模拟框架。其次,建立混合车辆流交通均衡模型,考虑用户行为相互影响,对各充电站充电负荷进行计算分析。再次,建立最优路径生成模型并采用分支定价方法进行路径生成,求解交通均衡模型,提出快速充电负荷模拟的算法流程。最后以西安市交通路网为例验证模型正确性,说明在充电负荷分析中考虑交通因素的必要性。算例也揭示了利用电价调整充电负荷空间分布的可能性。     

考虑多源信息实时交互和用户后悔心理的电动汽车充电负荷预测

文章提出了一种考虑多源信息实时交互和用户后悔心理的电动汽车充电负荷预测方法。首先,通过出行链理论和起止点(origin-destination,OD)矩阵法分别获得私家车和出租车出行的起讫点,利用Dijistra算法规划行驶路径;然后,构建基于路网实时车流量统计的速度-流量实用模型,计算路网各路段实时车速。构建考虑环境温度和车速的电动汽车单位里程耗电量模型,计算耗电量;接着考虑充电电价、时间、沿途耗电量等因素,提出基于后悔理论的电动汽车用户充电站选择模型;随后基于交通路网、车辆、公共快充站以及配电网等多源信息,建立多源信息实时交互的电动汽车充电负荷预测框架。最后采用蒙特卡洛法模拟了私家车和出租车的出行和充电过程,得到了区域内充电负荷时空分布。以某区域交通路网和典型配电网为例进行仿真,验证了所提充电负荷预测方法的有效性。仿真结果表明多源信息的及时交互以及考虑用户的后悔心理,会对充电负荷的时空分布产生影响。