Neural Network-Based Prediction of Vehicle Fuel Consumption Based on Driving Cycle Data

This paper deals with fuel consumption prediction based on vehicle velocity, acceleration, and road slope time series inputs. Several data-driven models are considered for this purpose, including linear regression models and neural network-based ones. The emphasis is on accounting for the road slope impact when forming the model inputs, in order to improve the prediction accuracy. A particular focus is devoted to conversion of length-varying driving cycles into fixed dimension inputs suitable for neural networks. The proposed prediction algorithms are parameterized and tested based on GPS- and CAN-based tracking data recorded on a number of city buses during their regular operation. The test results demonstrate that a proposed neural network-based approach provides a favorable prediction accuracy and reasonable execution speed, thus making it suitable for various applications such as vehicle routing optimization, synthetic driving cycle validation, transport planning and similar.

Assessment of potential sediment disasters and resilience management of mountain roads using environmental indicators

In recent years, extreme rainfall events occur frequently, causing serious watershed sediment disasters, destroying mountain roads, and endangering the safety of residents' lives and property. This study aims to deal with the spatial change of potential sediment movement on the road slope pre-disaster and to screen disaster hot spots for early warning and control system. The conceptual model is used to simulate the distribution of primary and/or derived disasters on a watershed scale to assess the impact of sediment disasters caused by heavy rain event. Correlation analysis shows that the models in assessment of primary disaster and derived disaster are significantly correlated with the collapse ratio and disaster ratio, respectively. Since the primary disaster has been considered when calculating the derived disaster risk, the terrain subdivision along Provincial Highway 21 (Tai-21) is extracted to understand the derived sediment disaster on the road slope. The model can effectively evaluate the road sections prone to disasters. According to the risk level, the hot spot of road slope disasters and the management of disaster resilience are determined and can be the reference for disaster prevention and control.

Adaptive Cruise Control for Intelligent City Bus Based on Vehicle Mass and Road Slope Estimation

Adaptive cruise control (ACC), as a driver assistant system for vehicles, not only relieves the burden of drivers, but also improves driving safety. This paper takes the intelligent pure electric city bus as the research platform, presenting a novel ACC control strategy that could comprehensively address issues of tracking capability, driving safety, energy saving, and driving comfort during vehicle following. A hierarchical control architecture is utilized in this paper. The lower controller is based on the nonlinear vehicle dynamics model and adjusts vehicle acceleration with consideration to the changes of bus mass and road slope by extended Kalman filter (EKF). The upper controller adapts Model Predictive Control (MPC) theory to solve the multi-objective optimal problem in ACC process. Cost functions are developed to balance the tracking distance, driving safety, energy consumption, and driving comfort. The simulations and Hardware-in-the-Loop (HIL) test are implemented; results show that the proposed control strategy ensured the driving safety and tracking ability of the bus, and reduced the vehicle’s maximum impact to 5 m/s3 and the State of Charge (SoC) consumption by 10%. Vehicle comfort and energy economy are improved obviously.

Fuel Consumption of Wheeled Vehicle and Transportation Costs during Highway Construction/Reconstruction

A method is proposed for determining the fuel consumption of a wheeled vehicle depending on its speed, road surface flatness and road slope in the longitudinal direction. The purpose of the research is to derive mathematical relationships for calculating the fuel consumption of vehicles, which is one of the transport cost factors during the construction/reconstruction or overhaul of a highway. The proposed polynomial dependencies for calculating fuel in addition to vehicle speed, road surface flatness and its longitudinal slope take into account the mass-dimensional parameters of vehicles involved in road traffic. New mathematical relationships between the speed of wheeled vehicles, road surface flatness and longitudinal road slope allow to simulate the change in the value of fuel consumption of a wheeled vehicle when the speed of traffic flow or the slope of the road surface changes in the forward or reverse direction of the vehicle. In a graphic way, the influence of the pavement slope on the value of fuel consumption, both loaded and unloaded wheeled vehicle is presented. When determining transport costs associated with the highway construction, reconstruction or overhaul it is proposed to use empirical mathematical relationships, which make it possible to obtain fuel consumption with an accuracy of 5 % and save up to 15 % of budget (private) investments. The analysis of scientific publications of the existing approach determine the transport costs associated with highway construction, reconstruction or overhaul. The presented method for determining the fuel consumption of wheeled vehicles with small and large loading capacity increases the accuracy of determining transportation costs and reduces the level of financial costs for highway construction, reconstruction or overhaul.

Analysis of the Embankment Settlement on Soft Soil Subgrade with a Cement Mixed Pile

The settlement of the widening of soft soil subgrade highways is typically associated with different treatment positions of cement mixed piles. In order to overcome this, in the current paper we employ the finite element method to simulate and analyze the influence of piles under an existing road slope and under an existing subgrade and new embankment on the settlement characteristics of the subgrade and foundation. In particular, we focus on the influence of the pile length and pile spacing on the subgrade and foundation settlements based on a northern high-speed reconstruction and expansion project. The subgrade and foundation soils in the finite element analysis are considered to be homogeneous, continuous, and isotropic elastoplastic materials. The Mohr–Coulomb ideal elastoplastic constitutive model is implemented as the constitutive soil model. The impact of piles under an existing subgrade and new embankment on the settlement is observed to be more significant than that of piles under the existing road slope. Moreover, the subgrade and foundation settlements increase with the pile spacing under the existing road slope and under the existing subgrade and new embankment. More specifically, an increase of the pile spacing from 200% to 400% of the pile diameter is associated with an increase in the maximum settlement of the foundation surface from 1.76 to 1.85 cm (existing road slope) and from 1.44 to 1.96 cm (existing subgrade and new embankment). In addition, the subgrade and foundation settlements decrease for increasing pile lengths under the existing road slope and under the existing subgrade and new embankment, the pile length increases from 4.7 to 9.2 m, and the maximum foundation surface settlement is reduced from 6.2 to 5.52 cm and from 9.73 to 5.43 cm, respectively. The results can provide reference for future subgrade widening projects.

Multi-parameter predictive shift schedule of automatic mechanical transmission for electric bus

Shift schedule is crucial in improving the dynamic and economic performance of electric vehicles (EVs) equipped with automatic mechanical transmission (AMT). As the driver, vehicle, and road constitute a closed-loop inseparable system, identifying the states of both vehicle and road is fundamental to realizing optimal shift schedule. However, the existing shift strategies neglect the coupling relationship of multiple parameters to the shift strategy. To minimize this gap, this paper presents a novel multi-parameter shift schedule based on model predictive control. Firstly, cubature Kalman filters (CKF) algorithm is employed to accurately estimate vehicle quality and road slope, which could improve the energy economy of EVs. Secondly, an artificial neural network (ANN) is adopted to forecast the compound future short horizon driving conditions, which contains the perdition information of vehicle velocity and road slope. Meanwhile, the AMT predictive shift schedule based on the above estimated and forecast information is constructed, which used dynamic programming to optimize in the rolling horizon. Simulation study results indicate that the ANN-based predictive approach shows better performance on accuracy and robustness than that of Markov chain, and the electricity consumption over China typical urban driving cycle (CTUDC) is further reduced by 6.79% than that of multi-parameter rule-based shift schedule.

A mass and road slope integrated estimation strategy based on the joint iteration of least square method and Sage-Husa adaptive filter for autonomous logistics vehicle

Autonomous logistics vehicles are characterised by large changes in mass and their performances are greatly influenced by slope. In addition, sensors on autonomous vehicles are expensive and difficult to be installed considering application environment. To address these problems, a novel integrated estimation strategy for vehicle mass and road slope, which is based on the joint iteration of multi-model recursive least square (MMRLS) and Sage-Husa adaptive filter with the strong tracking filter (SH-STF), is proposed by utilising information involving speed, nominal engine torque and inherent parameters of vehicles. Firstly, due to the separate slowly-changing and time-dependent characteristics, the vehicle mass and road slope are estimated by using MMRLS and SH-STF separately. Secondly, the longitudinal dynamics gain and the steering dynamics gain are calculated separately based on each model’s residual probability distribution. Then, the two estimations module are combined by employing an iterative algorithm. Finally, the proposed strategy is verified by simulation and real vehicle tests. The tests result reveals that the estimation algorithm can effective estimate vehicle mass and road slope in real-time under straight going and steering conditions.