Artificial intelligence for traffic signal control based solely on video images

Learning-based traffic control algorithms have recently been explored as an alternative to existing traffic control logics. The reinforcement learning (RL) algorithm is being spotlighted in the field of adaptive traffic signal control. However, no report has described the implementation of an RL-based algorithm in an actual intersection. Most previous RL studies adopted conventional traffic parameters, such as delays and queue lengths to represent a traffic state, which cannot be exactly measured on-site in real time. Furthermore, the traffic parameters cannot fully account for the complexity of an actual traffic state. The present study suggests a novel artificial intelligence that uses only video images of an intersection to represent its traffic state rather than using handcrafted features. In simulation experiments using a real intersection, consecutive aerial video frames fully addressed the traffic state of an independent four-legged intersection, and an image-based RL model outperformed both the actual operation of fixed signals and a fully actuated operation.     

Morphological Change and Number-Size Distributions of Particulate Matter (PM) from a Diesel Generator Operated with Wood Pyrolysis Oil-Butanol Blended Fuel

This report details our experimental study investigating particulate matter (PM) emissions from a diesel generator fueled with wood pyrolysis oil (WPO)–butanol blended fuel for electricity generation. Particle number-size distributions and PM mass concentrations from diesel, n-butanol, and WPO-butanol blended fuels were investigated via aerosol measurements using a fast mobility particle sizer and an aerosol monitor with three generator outputs (0, 3.3, and 6.6 kWe). For the n-butanol and WPO-blended fuels, the total number concentrations of exhaust particles were higher than that of conventional diesel combustion; however, the PM mass was observed to be nearly zero for all the engine operating conditions due to the higher number concentration in the nuclei mode. The morphology of the exhaust particles was investigated by analyzing transmission electron microscopy (TEM) micrographs. The morphology of the particles was drastically changed according to the test fuels and engine loads. Two types of particles were observed, including soot and coke shaped particles. These results were directly related to the immaturity of incipient soot particles due to the different physical properties and chemical compositions of the fuels.     

Effects of altitude on combustion characteristic during cold start of heavy-duty diesel engine

Altitude has a significant effect on combustion of heavy-duty diesel engines, especially during cold start. An experimental study on a heavy-duty diesel engine operating at different altitudes was conducted. Tests were based on a direct injection (DI) turbocharged diesel engine with intake and exhaust pressure controlled by the plateau simulation test system to stimulate altitude conditions including 0 m, 1000 m, 2000 m, 3000 m and 4000 m. Results indicated that the compression and expansion resistance moment reduced and the speed increased during the cranking period. The peak pressure of several cycles was increased during the start-up period; however, the expansion pressure dropped more and the indicated mean effective pressure (IMEP) reduced as the altitude rose. While at an altitude of over 2000 m, the peak pressure fluctuated obviously during the start-up period. The higher the altitude was, the more the fluctuation amplitude and cycle number increased and combustion instability enhanced, which resulted the start-up period time increasing at high altitude. When the altitude rose, the cycle-to-cycle variation of the peak pressure and speed fluctuation increased during the idle, the ignition and CA50 were delayed and the combustion duration was shortened. The effect of altitude on combustion characteristics of the diesel engine was more significant during the start-up period than during its idle period.     

Experimental and 3D,embedded modeling for diesel engine SCR deposit

Nowadays, the selective catalytic reduction (SCR) is a promising solution to fulfill stringent nitrogen oxide (NO_X) emission standards enforced by worldwide regulation bodies for lean burning engines. However, in the practical operation, urea deposits occur under unfavorable conditions, which bring about a failure of urea dosing strategy, affect the SCR system de-NOx performance and lowering the fuel economy by increasing the engine back pressure. This paper will present 3D deposit model, which can be used to predict the deposit position. Furthermore, a model-based controls strategy and calibration are designed. The comparison test results of both engine emission tests and vehicle field tests shows there is significant deposits improvement with the embedded model.     

Pedaling torque sensor-less power assist control of an electric bike via model-based impedance control

This paper proposes a method to assist human force acting on electric bike without using costly torque sensors via a model-based impedance control technique. In general, electric bikes are classified into two categories, i.e., pedelec electric bikes and throttle electric bikes. We focus on the system called a pedelec electric bike. It assists human pedaling force using the pedaling information, e.g., pedaling force or speed. To obtain the human’s pedaling information in real-time, it needs physical sensors such as a torque sensor and a velocity sensor. But, these sensors are expensive and weak against external loads. Also, since these sensors are fixed directly to the forced component in a bike system, there are the risks of damage. For these reasons, sensor-less control methods based on a disturbance observer have been studied so far. In this paper, we have proposed a pedaling torque sensor-less power assist method and have applied it to the experimental pedelec electric bike. A power assist control algorithm, designed by employing an impedance model, consists of a PI-type feedback controller, an inverse model-based feedforward controller, and a pedaling torque observer. Finally, we performed experiments and confirmed the effectiveness of a proposed power assist control method.     

Lateral handling improvement with dynamic curvature control for an independent rear wheel drive EV

The integrated longitudinal and lateral dynamic motion control is important for four wheel independent drive (4WID) electric vehicles. Under critical driving conditions, direct yaw moment control (DYC) has been proved as effective for vehicle handling stability and maneuverability by implementing optimized torque distribution of each wheel, especially with independent wheel drive electric vehicles. The intended vehicle path upon driver steering input is heavily depending on the instantaneous vehicle speed, body side slip and yaw rate of a vehicle, which can directly affect the steering effort of driver. In this paper, we propose a dynamic curvature controller (DCC) by applying a the dynamic curvature of the path, derived from vehicle dynamic state variables; yaw rate, side slip angle, and speed of a vehicle. The proposed controller, combined with DYC and wheel longitudinal slip control, is to utilize the dynamic curvature as a target control parameter for a feedback, avoiding estimating the vehicle side-slip angle. The effectiveness of the proposed controller, in view of stability and improved handling, has been validated with numerical simulations and a series of experiments during cornering engaging a disturbance torque driven by two rear independent in-wheel motors of a 4WD micro electric vehicle.     

The changing accuracy of traffic forecasts

Researchers have improved travel demand forecasting methods in recent decades but invested relatively little to understand their accuracy. A major barrier has been the lack of necessary data. We compiled the largest known database of traffic forecast accuracy, composed of forecast traffic, post-opening counts and project attributes for 1291 road projects in the United States and Europe. We compared measured versus forecast traffic and identified the factors associated with accuracy. We found measured traffic is on average 6% lower than forecast volumes, with a mean absolute deviation of 17% from the forecast. Higher volume roads, higher functional classes, shorter time spans, and the use of travel models all improved accuracy. Unemployment rates also affected accuracy—traffic would be 1% greater than forecast on average, rather than 6% lower, if we adjust for higher unemployment during the post-recession years (2008 to 2014). Forecast accuracy was not consistent over time: more recent forecasts were more accurate, and the mean deviation changed direction. Traffic on projects that opened from the 1980s through early 2000s was higher on average than forecast, while traffic on more recent projects was lower on average than forecast. This research provides insight into the degree of confidence that planners and policy makers can expect from traffic forecasts and suggests that we should view forecasts as a range of possible outcomes rather than a single expected outcome.

     

运营盾构隧道不同椭圆度变形情况下受力性能分析北大核心CSCD

地下结构施工不当会对邻近既有地铁盾构隧道结构产生巨大影响。文章以广州市某地铁线路下穿施工导致既有运营地铁盾构隧道产生较大变形的工程实例为背景,分析了既有隧道结构因地层损失产生不同椭圆度变形情况下管片结构的受力情况。基于工程实测数据,运用三维有限元分析软件,考虑了管片接头处的螺栓孔等细部构造,研究了管片椭圆度与结构应力状态之间的量化关系,并分析了结构的塑性变形情况及其发展趋势。结果表明:随着盾构隧道管片椭圆度的增大,结构最大主应力值与最大剪应力值均增大,且盾构隧道结构最大剪应力与椭圆度呈线性相关关系;盾构隧道结构最大主应力随椭圆度变化更加明显,与椭圆度呈非线性相关关系。