Search

Short vehicle speed prediction is important in predictive energy management strategies, and the accuracy of the prediction is beneficial for energy-saving performance. However, the nonlinear feature of the speed series hinders the improvement of prediction accuracy. In this study, a novel hybrid model that combines an autoregressive integrated moving average (ARIMA) and a long short-term memory (LSTM) model is proposed to handle the nonlinear part efficiently. Generally, the ARIMA component filters out linear trends from the speed series data, and the parameters of the ARIMA are determined with the analysis. Then the LSTM handles the residual normalized nonlinear items, which is the residual of ARIMA. Finally, the two parts of the prediction results are superimposed to obtain the final speed prediction results. To assess the performance of the hybrid model (ARIMA-LSTM), two tested driving cycles and two typical driving scenarios are subjected to rigorous analysis. The results demonstrate that the combined prediction model outperforms individual methods ARIMA and LSTM in dealing with complex, nonlinear variations, and exhibits significantly improved performance metrics, including root mean square error (RMSE), mean absolute error (MAE), and mean percentage error (MAPE). The proposed hybrid model provides a further improvement for the accuracy prediction of vehicle traveling processes. [ABSTRACT FROM AUTHOR]

Copyright of Journal of South China University of Technology (Natural Science Edition) is the property of South China University of Technology and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

A motor resistance parameter identification scheme was proposed for the speed prediction of a direct-torque-controlled (DTC) induction motor (IM) drive. The DTC IM drive was established on the basis of the stator's current and flux, with the stator current acquired from an IM using the Hall effect current sensor. Rotor speed prediction was achieved using the electromagnetic torque and rotor flux. The stator resistance parameter identification scheme was developed using the model reference adaptive system based on the motor's active power, and the adaptation mechanism was designed using the modified particle swarm optimization algorithm. The MATLAB\Simulink® toolbox was utilized to simulate this system, and all the control algorithms were realized using a TI DSP 6713 and F2812 micro-control card to validate this approach. Simulation and experimental results confirmed the effectiveness of the proposed approach. [ABSTRACT FROM AUTHOR]

The eco‐driving strategy is of great significance in driving cost for plug‐in hybrid electric vehicles in driving trips, especially at signalized intersections. To address the issue of further energy saving, this study proposes an ecological approach and departure‐driving strategy by using syncretic learning with trapezoidal collocation algorithm. First, a syncretic learning‐based speed predictor is built by merging back propagation neural networks and radial basis function neural networks. Second, the syncretic learning‐based speed predictor and trapezoidal collocation algorithm are combined to optimize the speed trajectory. Third, the torque between the engine and the motor is distributed by the dynamic programming algorithm. Then, model predictive control optimizes torque output in the control time domain. Finally, the driving interval optimization method is designed to avoid mixed‐integer programming problems and redundant constraints, which make vehicles cross intersections without stopping. The numerical verification results show that the trapezoidal collocation algorithm with syncretic learning has more advantages than other methods in speed trajectory planning. Compared with the original trajectory, the driving time through the intersection is reduced and the total driving cost is lowered by 19.82%. Validation results confirm the effectiveness of the proposed strategy in energy consumption management at signalized intersections. [ABSTRACT FROM AUTHOR]

Intelligent Vehicle–Infrastructure Collaboration Systems (i-VICS) put forward higher requirements for the real-time security of dynamic traffic information interaction. It is difficult to ensure the safety of dynamic traffic information interaction by means of traditional static information security. In this study, a method was proposed through machine learning-based lane-changing (LC) behavior recognition and information credibility discrimination, based on the utilization and exploitation of traffic business characteristics. The method consisted of three stages: LC behavior recognition based on Support Vector Machine (SVM), LC speed prediction based on Recurrent Neural Network (RNN), and credibility discrimination of speed information under LC states. Firstly, the labeling rules of vehicle LC behavior and the input/output of each stage model were determined, and the raw NGSIM data were processed to obtain data sets for LC behavior identification and LC speed prediction. Both the SVM classification and RNN prediction models were trained and tested, respectively. Afterwards, a model of credibility discrimination speed information under an LC state was constructed, and the real vehicle speed data were processed for model verification. The results showed that the overall accuracy of vehicle status recognition by the SVM model was 99.18%, and the precision of the RNN model was on the order of magnitude of cm/s. Considering transverse and longitudinal abnormal velocity, the accuracy credibility discrimination of LC velocity was more than 97% in most experimental groups. The model can effectively identify the abnormal speed data of LC vehicles and provide support for the real-time identification of LC vehicle speed information under i-VICS. [ABSTRACT FROM AUTHOR]

At present, the design of modern vehicles requires improving driving performance while meeting emission standards, leading to increasingly complex power systems. In autonomous driving systems, accurate, real-time vehicle speed prediction is one of the key factors in achieving automated driving. Accurate prediction and optimal control based on future vehicle speeds are key strategies for dealing with ever-changing and complex actual driving environments. However, predicting driver behavior is uncertain and may be influenced by the surrounding driving environment, such as weather and road conditions. To overcome these limitations, we propose a real-time vehicle speed prediction method based on a lightweight deep learning model driven by big temporal data. Firstly, the temporal data collected by automotive sensors are decomposed into a feature matrix through empirical mode decomposition (EMD). Then, an informer model based on the attention mechanism is designed to extract key information for learning and prediction. During the iterative training process of the informer, redundant parameters are removed through importance measurement criteria to achieve real-time inference. Finally, experimental results demonstrate that the proposed method achieves superior speed prediction performance through comparing it with state-of-the-art statistical modelling methods and deep learning models. Tests on edge computing devices also confirmed that the designed model can meet the requirements of actual tasks. [ABSTRACT FROM AUTHOR]

In order to improve the punctuality and safety of urban rail transit trains during operation and achieve accurate parking, it is necessary to track and predict the speed curve during the train operation. This paper firstly calculates the instantaneous power of the train based on the measured data, and then uses the random forest model to classify the interval according to the power curve, and then establishes a real-time prediction method for the speed curve of urban rail transit trains based on neural network for different intervals. The train speed prediction model is tested. The results of model testing on the simulation data and actual line data show that the proposed algorithm can effectively predict the speed curve of the train in real time, improve the accuracy of speed tracking control. The error is reduced by 57.7% compared with the traditional neural network model, and the error is reduced by 73.9% compared with the random forest regression model.

An efficient energy management strategy (EMS) is significant to improve the economy of hybrid electric vehicles (HEVs). Herein, a power‐split HEV model is built and validated against test results, and then the EMS is proposed for this model based on vehicle speed prediction and deep reinforcement learning (DRL) algorithms. The rule‐based local controller and global optimal empirical knowledge are introduced to enhance the convergence speed. It is shown in the results that the twin delayed deep deterministic policy gradient algorithm (TD3) achieves more satisfactory performance on converge speed and energy efficiency. The networks of the DRL algorithm with continuous control update more robustly during iterations, in contrast to the discrete ones. Although the power‐split HEV with lower control dimension can reduce the learning burden for DRL EMS; however, the multidimensional control space shows greater optimization potential. As a result, the equivalent fuel consumption of TD3‐based EMS with multidimensional continuous control differences from the global optimal algorithm only by 4.92%. Herein, it is demonstrated in the results that long short‐term memory recurrent neural network (LSTM RNN) performs better for vehicle speed prediction than classical RNN and BP neural network, and the predictive vehicle speed feature helps improve fuel economy by 0.55%. [ABSTRACT FROM AUTHOR]

Smart cars rely on sensors like LIDAR and high-precision map-based perception for driving environment sensing. However, they can't detect low-speed vehicles beyond visual range, affecting safety and comfort. Manual vehicles face similar challenges. Low-speed driving contributes to expressway accidents due to limited visibility, road design, and equipment performance. To enhance safety, an over-the-horizon potential safety threat vehicle identification method using ETC big data is proposed. It consists of three layers. The first layer is the vehicle section travel speed sensing layer based on the wlp-XGBoost algorithm. The second layer is the in-transit vehicle position estimation layer based on the DR-HMM algorithm. The third layer is the Multi-information fusion of potential safety threat vehicle identification layer. Dynamic real-time detection and identification of potential safety threats in expressway sections were achieved, and simulations were conducted using real-time ETC data from Quanxia section on an ETC platform. Results show accurate prediction of vehicle speed and position in different road sections and traffic situations, with over 95% accuracy and recall in identifying potential safety threat vehicles. It perceives changes in the traffic conditions of road sections in real-time based on the changing trend of potential safety threat vehicle numbers, providing a vital reference for speed planning and risk avoidance.

This work proposes a predictive equivalent consumption minimization (P-ECMS) strategy for a power-split hybrid electric vehicle (HEV) using predicted driving cycle speed based on Monte Carlo (MC) algorithm. The proposed P-ECMS fully takes advantage of the predicted speed profiles by the MC algorithm to optimally determine the power split among energy sources. In this study, to validate the workings of the MCbased P-ECMS scheme, a series of simulations under a total of seven replicated driving cycles including New European Driving Cycle (NEDC), Worldwide Harmonised Light Vehicles Test Procedure (WLTP), Urban Dynamometer Driving Schedule (UDDS), Highway Fuel Economy Test (HWFET), New York City Cycle (NYCC), California Unified Cycle (LA-92), and a combination of all (ALL-CYC) are conducted. The MC-based P-ECMS strategy is compared with a baseline ECMS in terms of fuel-saving. The fuel economy saving up to 6.01% under NEDC, 9.09% under WLTP, 6.33% under UDDS, 5.14% under HWFET, 1.96% under NYCC, 11.47% under LA-92, and 7.92% under ALL-CYC is achieved. The results in this article put forward that the proposed strategy delivers competitive fuel savings compared to the widely used baseline method. [ABSTRACT FROM AUTHOR]

Due to the continuous high traction power impact on the energy storage medium, it is easy to cause many safety risks during the driving process, such as triggering the aging mechanism, causing rapid deterioration of the battery performance during the driving process and even triggering thermal runaway. Hybrid energy storage is an effective way to solve this problem. The ultracapacitor is an energy storage device that has high power density, which can withstand high instantaneous currents and can be charged and discharged quickly. By combining batteries and ultracapacitors in a hybrid energy storage system, energy sources with different characteristics can be combined to take advantage of their respective strengths and increase the efficiency and lifetime of the system. The energy management strategy plays an important role in the performance of hybrid energy storage systems. Traditional optimization algorithms have difficulty improving the flexibility and practicality of applications. In this paper, an energy management strategy based on reinforcement learning is proposed. The results indicate that the proposed reinforcement method can effectively distribute the charging and discharging conditions of the power supply and maintain the SOC of the battery and, at the same time, meet the power demand of working conditions at the cost of less energy loss and effectively realize the goal of optimizing the overall efficiency and effective energy management strategy. [ABSTRACT FROM AUTHOR]

Copyright of Journal of Southeast University / Dongnan Daxue Xuebao is the property of Journal of Southeast University Editorial Office and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Urban road speed prediction is helpful to guide drivers to choose unimpeded routes, reduce waiting time and improve travel efficiency. Urban traffic conditions are affected by many factors. Based on the consideration of various traffic flow characteristic data and weather data, a combined model of road travel speed prediction based on long short-term memory (LSTM) cyclic neural network is established. The Didi floating car data in the area around the South Second Ring Road of Xi'an city are selected to predict the road travel speed by extracting the traffic flow characteristics (speed, flow, acceleration and stopping times) and weather characteristics (temperature, humidity, weather and wind speed) of the data set. The results show that compared with LSTM model, BP neural network model and SVR model without external features, the mean absolute error, mean square error and determination coefficient of the combined model with multi-feature data are 2. 695, 13. 838 and 0. 771, and the confidence interval of (-1. 235, 1. 795) is better than other models. The combined model has higher accuracy and stability. [ABSTRACT FROM AUTHOR]

The implementation of an energy management strategy plays a key role in improving the fuel economy of plug-in hybrid electric vehicles (PHEVs). In this article, a bi-level energy management strategy with a novel speed prediction method leveraged by reinforcement learning is proposed to construct the optimization scheme for the inner energy allocation of PHEVs. First, the powertrain transmission model of the PHEV in a power-split type is analyzed in detail to obtain the energy routing and its crucial characteristics. Second, a Q-learning (QL) algorithm is applied to establish the speed predictor. Third, the double QL algorithm is introduced to train an effective controller offline that realizes the optimal power distribution. Finally, given a reference battery's state of charge (SOC), a model predictive control framework solved by the reinforcement learning agent with a novel speed predictor is proposed to build the bi-level energy management strategy. The simulation results show that the proposed method performs with a satisfying fuel economy in different driving scenarios while tracking the corresponding SOC references. Moreover, the calculation performance also implies the potential online capability of the proposed method.

The "white hole effect" alters the driving environment during a tunnel's exit phase, making it more difficult and uncertain for drivers to access information and control their behavior, thereby endangering traffic safety. Consequently, the driving risk at the exit of a long spiral tunnel served as the subject of this study, and the Jinjiazhuang spiral tunnel served as the object of the natural vehicle driving experiment. Following the theory of a non-linear autoregressive dynamic neural network, a vehicle speed prediction model based on driver characteristics was developed for the exit phase of the tunnel, taking driver expectations and behavioral changes into account. It also classifies the driver's behavior during the tunnel's exit phase to assess the risk posed by the driver's behavior during the tunnel's exit phase and determine a dynamic and safe comfort speed. The study's results indicate that the driver's behavioral load changed significantly as the vehicle approached the tunnel exit. At the exit of the spiral tunnel, the vehicle's actual speed was 71 km/h, which is below the speed limit of 80 km/h. This demonstrates that the expected change in the driver's behavior in the tunnel exit phase was substantial. Therefore, setting the emotional safety and comfort speed so that the driver maintains a smooth comfort level in the tunnel exit phase can reduce the tunnel exit driving risk. The results of this study provide a benchmark for tunnel traffic safety and lay the groundwork for further development of vehicle risk warning settings for the tunnel's exit phase. [ABSTRACT FROM AUTHOR]

Inclement weather and environmental factors impact traffic operations resulting in travel delays and a reduction in travel time reliability. Precipitation is an example of an environmental factor that affects travel conditions, including traffic speed. While Intelligent Transportation Systems services aim to proactively mitigate congestion on roadways, these services are often not sensitive to weather conditions. This paper investigates the application of high-resolution weather data in improving the performance of proactive transportation management models and proposes short-term speed prediction models that fuse real-time high-resolution weather surveillance radar data with traffic stream data to conduct spatial and temporal prediction of the speed of roadway segments. Extreme gradient boosting weather-aware speed prediction models were developed for a 7-km segment of Interstate 270 in St. Louis, MO, USA. The performance of the weather-aware models was compared with the performance of weather-insensitive speed prediction models that did not take precipitation into account. The results indicated that in the majority of instances, the weather-aware models outperformed the weather-insensitive models. The extreme gradient boosting models were compared with the K-nearest neighbors algorithm and feed-forward neural network models. The extreme gradient boosting model consistently outperformed the other two methods. In addition to speed prediction models, van Aerde speed-flow traffic stream models were developed for rain and no-rain conditions to study the impact of precipitation on the traffic stream across the corridor. Results indicated that the impact of precipitation is not identical across the corridor, which was mirrored in the results obtained from weather-aware speed prediction models. [ABSTRACT FROM AUTHOR]

Traffic speed prediction is a vital part of the intelligent transportation system (ITS). Predicting accurate traffic speed is becoming an important and challenging task with the rapid development of deep learning and increasing traffic data size. In this study, we present a deep-learning-based architecture for network-wide traffic speed prediction. We propose a deep-learning-based model consisting of a fully convolutional neural network, bidirectional long short-term memory, and attention mechanism. Our design aims to consider both backward and forward dependencies of traffic data to predict multistep network-wide traffic speed. Thus, we propose a model named AttBDLTSM-FCN for multistep traffic speed prediction. We augmented the attention-based bidirectional long short-term memory recurrent neural network with the fully convolutional network to predict the network-wide traffic speed. In traffic speed prediction, this is the first time that augmentation of AttBDLSTM and FCN have been exploited to measure the backward dependency of traffic data, as a building block for a deep architecture model. We conducted comprehensive experiments, and the experimental evaluations illustrated that the proposed architecture achieved better performance compared to state-of-the-art models when considering the short and long traffic speed prediction, e.g., 15, 30, and 60 min, in multistep traffic speed prediction. [ABSTRACT FROM AUTHOR]

At present, the design of modern vehicles requires improving driving performance while meeting emission standards, leading to increasingly complex power systems. In autonomous driving systems, accurate, real-time vehicle speed prediction is one of the key factors in achieving automated driving. Accurate prediction and optimal control based on future vehicle speeds are key strategies for dealing with ever-changing and complex actual driving environments. However, predicting driver behavior is uncertain and may be influenced by the surrounding driving environment, such as weather and road conditions. To overcome these limitations, we propose a real-time vehicle speed prediction method based on a lightweight deep learning model driven by big temporal data. Firstly, the temporal data collected by automotive sensors are decomposed into a feature matrix through empirical mode decomposition (EMD). Then, an informer model based on the attention mechanism is designed to extract key information for learning and prediction. During the iterative training process of the informer, redundant parameters are removed through importance measurement criteria to achieve real-time inference. Finally, experimental results demonstrate that the proposed method achieves superior speed prediction performance through comparing it with state-of-the-art statistical modelling methods and deep learning models. Tests on edge computing devices also confirmed that the designed model can meet the requirements of actual tasks.

This research aims to improve the performance and economics of fuel cell hybrid electric vehicles (FCHEVs), validated and established by introducing an innovative energy management strategy (EMS) based on a speed-predictive fusion model. Firstly, a mixed prediction model was built based on BiLSTM, TCN, and Self-attention (SA) mechanism to accurately search, capture and fuse multi-granularity features in time series. Then, Harris-Hawk Optimization (HHO) was used to optimize the dropout rate and model learning rate of the combined BiLSTM-TCN-SA time series model to improve the prediction accuracy and generalization ability of the model. Finally, stochastic model predictive control was combined with BiLSTM-TCN-SA to form SMPC-NSGA III algorithm, which was used for multi-objective optimization of fuel economy, fuel cell durability and battery durability. In this study, the effectiveness of the proposed strategy was verified under the condition of CLTC-P driving cycle. The experimental results showed that RMSE and R2 of HHO-BiLSTM-TCN-SA velocity prediction model are 1.169 and 0.998, respectively. In addition, the output of the model is within the confidence interval of 97.5 % of the real speed, and there is no significant difference, which is statistically significant. Under the SMPC-NSGA III strategy, the average efficiency of the fuel cell was increased by 12 % and 1 % respectively. • A new energy management strategy was created to improve fuel cell durability. • HHO-based optimisation of BiLSTM-TCN-Self-Attention hyperparameter fusion model improves speed prediction performance. • NSGA-III multi-objective optimization algorithm was combined with SMPC for collaborative optimization results. • The average efficiency of the fuel cell in the SMPC strategy is higher than the DP and FLC. [ABSTRACT FROM AUTHOR]

To exploit the energy-saving potential and optimize the battery state of charge (SOC) maintaining capability of energy management strategies for fuel cell hybrid vehicles in specific driving scenarios, this study proposes a scenario-oriented adaptive equivalent consumption minimization strategy (SA-ECMS) based on a Nanjing-oriented driving cycle (NODC) and future speeds predicted via a hybrid neural network model. The proposed strategy determines the initial value of the equivalent factor (EF) and the proportional coefficient of the adaptive increment based on the NODC. Then, it periodically adjusts the EF via local optimization process according to the predicted speed to enhance scenario-specific adaptability and energy efficiency performance. Simulation results show that the hybrid neural network model achieves an average calculation time of 0.0033 s with a root-mean-square error of 0.85 m/s for 10 s prediction horizon, outperforming existing speed prediction models. Compared with the existing SOC feedback-based ECMS, the proposed SA-ECMS effectively suppresses the battery SOC within a narrower fluctuation range of −0.12% to 0.33%, achieves a deviation of only 0.0026 from the SOC reference value, and reduces the equivalent hydrogen-fuel consumption by 2.49% to 7.06 g/km. • A scenario-oriented driving cycle construction method is proposed. • A novel hybrid neural network model with higher prediction accuracy is proposed. • The scenario-oriented adaptability of the proposed strategy is enhanced. • Both battery SOC maintaining capability and fuel economy are improved. [ABSTRACT FROM AUTHOR]

In order to improve the fuel economy of hybrid electric vehicle (HEV) with single row planetary gear and reduce the fuel consumption of the HEV, an energy management strategy based on gated recurrent unit neural network (GRU-NN) speed predictive model and adaptive differential evolution (A-DE) algorithm was proposed. The future speed of HEV was predicted under the framework of model predictive control (MPC). The energy management strategy converted the global optimization solution problem in the entire working condition into a local optimization solution in the prediction time domain. Aiming at the lowest fuel consumption of the engine and the balance of battery state of charge (SOC) during driving, the optimal control sequence in the prediction domain was solved by A-DE. The simulation results show that the energy management strategy based on the GRU-NN and A-DE reduces fuel consumption by 4.55% compared with that of equivalent consumption minimization strategy (ECMS), and the fuel economy reaches 93.04% compared with that of dynamic programming (DP) under the driving cycle collected by vehicle. [ABSTRACT FROM AUTHOR]

Given the vehicle speed during actual driving, it is possible to apply an advanced energy management strategy for achieving better efficiency and less emission. We conducted a study to predict the future speed while driving of city buses, where only a few bus driving data and bus stop IDs are used without external complex traffic information. The speed prediction models were developed based on long time short memory (LSTM) and a gated recurrent unit (GRU), and a deep neural network (DNN) is also adopted for the bus stop ID processing. The performances of the models were analyzed and compared such that we found the LSTM-based model presents remarkable and practical prediction ability in accuracy and time spent. Adopting the proposed speed prediction model would make it a reality sooner, application of the optimal energy control strategy in the real world. [ABSTRACT FROM AUTHOR]

The future motion prediction of vehicles in the front is widely valued for its great potential to improve a vehicle's safety, fuel consumption, and efficiency. However, due to the uncertainty of a driver's driving intentions and vehicle dynamics, future motion prediction faces great challenges. In order to break the bottleneck in the prediction of leading vehicle motion, this paper proposes a prediction idea of decoupling the prediction of leading vehicle motion into vertical vehicle speed prediction based on the Gaussian process regression algorithm and horizontal heading angle prediction based on the long short-term memory method, which combines the predicted vehicle speed and heading angle to derive the future trajectory of the leading vehicle. Moreover, we propose a prediction algorithm of the leading vehicle motion based on the combination of driving intention recognition and multimodel prediction results by the Fuzzy C-means algorithm, which tries to solve the problem of the unclear driving intention of the predicted object and the nonlinearity between the future motion of the vehicle and the environment. Finally, the algorithm is validated using real vehicle data, proving that it has high prediction accuracy. [ABSTRACT FROM AUTHOR]

The aim of this work is to forecast wind energy by fitting the wind speed logged data, that have been measured over a year period (Nov. 2019–Mar. 2021), on a unique probability density function selected among a number of similar probability functions, as it is not always possible to select one distribution function that fits all wind speed regimes. The wind speed and direction data were measured at Fujairah site, which are affected by long-term fluctuation of ± 10% of wind speed, and short-term fluctuation of more than ± 20%. Based on the foregoing measurements, five different probability density functions can be fitted, namely Weibull, Rayleigh, Gamma, Lognormal and Exponential, with their associated parameters. A procedural algorithm is proposed for wind speed forecasting with best selected fitting distribution function, using a procedural forecast-check method, in which forecasting is performed with time on the most suitable distribution function that fits the foregoing data, depending on minimum errors accumulated from preceded measurements. Different error estimation methods are applied. The algorithm of selecting different distribution functions with time, makes energy prediction more accurate depending on the fluctuation of wind speed. A detailed probabilistic analysis is carried out to predict probable wind speed, and hence wind energy, based on variations of the parameters of the selected fitting distribution function. [ABSTRACT FROM AUTHOR]

In this study, an adaptive speed prediction scheme based on reactive power was established for direct torque-controlled (DTC) permanent magnet synchronous motor (PMSM) drives. The current and flux of a stator were used to establish a DTC PMSM drive. Hall effect current sensors with a non-contact sensing technique were used to detect the stator current of the PMSM. The voltage space vector pulse width modulation (VSVPWM) DTC scheme was used in place of a traditional switching table (ST) DTC scheme to reduce current and torque ripples. Model reference adaptive control (MRAC) was utilized to develop a speed prediction scheme, and its adaptation mechanism was designed using the elephant herding optimization (EHO) algorithm. The torque, flux, and speed controllers were designed using a proportional–integral (P–I)-type controller. The MATLAB/Simulink© toolbox was used to establish the simulation scheme, and all control algorithms were realized using a microprocessor control card. The simulation and experimental results confirmed the effectiveness of the proposed approach. [ABSTRACT FROM AUTHOR]

Classical approaches for modelling aircraft taxi-speed assume constant speed or use a turning rate function to approximate taxi-timings for taxiing aircraft. However, those approaches cannot predict spatio-temporal component of aircraft-taxi trajectory due to a lack of consideration of the complexity and stochasticity of airport-airside movements and interactions. This research adopts the Generative Adversarial Imitation Learning (GAIL) algorithm for aircraft taxi-speed modelling, while considering multiple operational factors including surrounding traffic on the ground and target take-off time. The proposed model can learn and reproduce the ground movement patterns in a real-world dataset under different circumstances. In addition, the characteristics of the taxi-speed model are also analyzed, especially focusing on handling conflict scenarios with surrounding traffic. Finally, the travel-time of the aircraft from starting to target positions are compared with baseline models and actual taxiing data. The proposed model outperforms all the baseline models with a significant margin. In terms of spatial completion (SC), it achieves up to 97.1% for arrivals and 88.3% for departures. The results also show significantly high performance for temporal completion. The model achieves a stable performance with low Root Mean Square Error (RMSE) (16.8 seconds for arrivals, 32.4 seconds for departures) and Mean Absolute Percentage Error (MAPE) (4.4% for arrivals and 7.6% for departures). Our model’s errors are 72% lower for arrivals and 48% lower for departures when compared to other baseline models. [ABSTRACT FROM AUTHOR]

Customized path-based speed prediction is an eventful tool for congestion avoidance, route optimization and travel time prediction for navigation apps, cab-hailing companies and autonomous vehicles. Traditionally, the speed prediction algorithms are based on road segments and can only support several main roads. Path-based speed prediction is very challenging since the speed is always changing in different path locations and is jointly affected by lots of complicated factors. This article presents a novel deep learning framework for customized path-based speed prediction. A Path-based Speed Prediction Neural Network (PSPNN) is designed to achieve speed predictions for a given path and attributes information. A hierarchical Convolutional Neural Network (CNN) and deep Bidirectional Long Short-Term Memory (Bi-LSTM) structure for different kinds of feature extraction are applied for multiple levels: the path cell, sub-path and the whole path. The method narrows down the prediction unit from road segments to customized path cells (mean length: 59.52m) and achieves a mean absolute error (MAE) of 1.94 m/s and Mean Absolute Percentage Error (MAPE) of 18.14%, showing the potential of serving rigorous data-driven applications. So far, PSPNN is the first made-to-order path-based speed prediction algorithm and can help both travelers and managers to obtain large-scale bespoke paths speed information in advance. [ABSTRACT FROM AUTHOR]

Short-term road traffic speed prediction is a long-standing topic in the area of Intelligent Transportation System. Apparently, effective prediction of the traffic speed on the road can not only provide timely details for the navigation system concerned and help the drivers to make better path selection, but also greatly improve the road supervision efficiency of the traffic department. At present, some researches on speed prediction based on GPS data, by adding weather and other auxiliary information, using graph convolutional neural network to capture the temporal and spatial characteristics, have achieved excellent results. In this paper, the problem of short-term traffic speed prediction based on GPS positioning data is further studied. For the processing of time series, we innovatively introduce Dynamic Time Warping algorithm into the problem and propose a Long Short-Term Memory with Dynamic Time Warping (D-LSTM) model. D-LSTM model, which integrates Dynamic Time Warping algorithm, can fine-tune the time feature, thus adjusting the current data distribution to be close to the historical data. More importantly, the fine-tuned data can still get a distinct improvement without special treatment of holidays. Meanwhile, considering that the data under different feature distributions have different effects on the prediction results, attention mechanism is also introduced in the model. Our experiments show that our proposed model D-LSTM performs better than other basic models in many kinds of traffic speed prediction problems with different time intervals, and especially significant in the traffic speed prediction on weekends. [ABSTRACT FROM AUTHOR]

We propose an eco-driving technique for a heavy-duty ego vehicle that drives behind a leading heavy-duty vehicle. By observing a decrease in speed of the leading vehicle when driving uphill, its power capability is estimated and its future speed is predicted within a look-ahead horizon. The predicted speed is utilised in a model predictive controller (MPC) to plan the optimal speed of the ego vehicle such that its fuel consumption is minimised, while keeping a safe distance to the leading vehicle and reducing the need for braking. The effectiveness of the proposed technique is analysed in two case studies on real road topographies. By using the leading vehicle observer, fuel savings are achieved up to 8% compared to the case where the preceding vehicle is assumed to have a constant speed within the look-ahead horizon. [ABSTRACT FROM AUTHOR]

Due to the continuous high traction power impact on the energy storage medium, it is easy to cause many safety risks during the driving process, such as triggering the aging mechanism, causing rapid deterioration of the battery performance during the driving process and even triggering thermal runaway. Hybrid energy storage is an effective way to solve this problem. The ultracapacitor is an energy storage device that has high power density, which can withstand high instantaneous currents and can be charged and discharged quickly. By combining batteries and ultracapacitors in a hybrid energy storage system, energy sources with different characteristics can be combined to take advantage of their respective strengths and increase the efficiency and lifetime of the system. The energy management strategy plays an important role in the performance of hybrid energy storage systems. Traditional optimization algorithms have difficulty improving the flexibility and practicality of applications. In this paper, an energy management strategy based on reinforcement learning is proposed. The results indicate that the proposed reinforcement method can effectively distribute the charging and discharging conditions of the power supply and maintain the SOC of the battery and, at the same time, meet the power demand of working conditions at the cost of less energy loss and effectively realize the goal of optimizing the overall efficiency and effective energy management strategy.

Accurate steady state speed prediction is one of the key indexes to improve vehicle safety. This paper presents a steady state speed prediction method for expressway vehicles under snow and ice conditions. Firstly, the torque balance equation of vehicle rollover time is established, and the critical state of vehicle stability is judged by the lateral acceleration of vehicle. Secondly, the steady-state speed prediction parameters, including friction coefficient, braking distance, driver reaction time and vehicle yaw Angle, were calculated according to the judgment results of stable critical state, and the steady-state speed prediction model was built according to the predicted parameters. Experimental results show that this method can accurately detect the critical state of vehicle stability, and the steady-state speed prediction results are basically consistent with the actual speed. [ABSTRACT FROM AUTHOR]

Energy management strategies (EMSs) based on speed prediction are widely used in fuel cell hybrid electric vehicles (FCHEVs). However, two fundamental issues must be addressed: short-term speed prediction and power allocation across multi-scales. To address these issues, a hierarchical EMS for FCHEV based on multi-scale information fusion is developed in this study. In this strategy, a novel speed predictor that incorporates visual information and the motion states of a vehicle is used to predict the short-term speed. At the global level, the average power demand of each road segment is predicted based on real-time traffic information, which is used to plan the global reference state of charge (SOC) trajectory before departure. At the local level, fuzzy rules are utilized to determine the short-term reference SOC. The short-term optimal co-state of the Pontryagin minimal principle-based strategy is then updated. The results demonstrate that the proposed visual information speed predictor (VISP) enhances the prediction accuracy by 12%–32.5% and leads to a 3.36% reduction in the total cost compared with the EMS utilizing the conventional predictor. In addition, the proposed EMS makes the terminal SOC close to the target value and reduces the total cost by 20.33% compared to the benchmark strategy. • A hierarchical energy management strategy based on multi-scale information fusion is proposed. • A novel speed predictor that incorporates visual information and vehicle motion states is designed. • Global reference SOC trajectory method based on the real-time temporal and spatial traffic information of road segments is proposed. • Adaptive fuzzy rules are used to determine the shot-term reference SOC for each horizon. • Real urban driving cycle is used to verify the performance of the energy management strategies. [ABSTRACT FROM AUTHOR]

With the rapid economic development, urban areas are seeing more and more vehicles, leading to frequent urban traffic congestion. To solve this problem, the forecasting of traffic parameters is essential, in which, road operating speed (hereinafter referred to as “road speed”) is a key parameter for forecasting road congestion. This paper proposes a hybrid radial basis function (RBF) neural network algorithm for forecasting road speed. First, it proposes a fuzzy RBF neural network structure by combining the fuzzy logic system with the RBF neural network. Then, it incorporates factors such as weather, holidays and road grades into the input layer. Considering the uncertainty and sensitivity of the initial centre of the traditional membership function layer, it uses fuzzy C-means clustering to determine the centre and other parameters of the membership function layer. Then using the gradient descent method, it trains the weights between the fuzzy inference layer and the output layer. Finally, this paper trains the proposed hybrid RBF neural network with the traffic road network data and weather data of a city, and uses the trained hybrid neural network to predict the road speed and the congestion status. The prediction results show that, compared with simplex prediction methods, such as BP neural network, time series method, and RBF neural network, the hybrid RBF neural network has a higher forecasting accuracy, with the mean absolute percentage error (MAPE) being reduced to 6.4%. Experimental results verify the accurate forecasting, enhanced learning feature and mapping capability of this method in short-term road speed forecasting, indicating that it can provide reliable predicted values to help solve urban congestion problems.

Traffic speed prediction is a vital part of the intelligent transportation system (ITS). Predicting accurate traffic speed is becoming an important and challenging task with the rapid development of deep learning and increasing traffic data size. In this study, we present a deep-learning-based architecture for network-wide traffic speed prediction. We propose a deep-learning-based model consisting of a fully convolutional neural network, bidirectional long short-term memory, and attention mechanism. Our design aims to consider both backward and forward dependencies of traffic data to predict multistep network-wide traffic speed. Thus, we propose a model named AttBDLTSM-FCN for multistep traffic speed prediction. We augmented the attention-based bidirectional long short-term memory recurrent neural network with the fully convolutional network to predict the network-wide traffic speed. In traffic speed prediction, this is the first time that augmentation of AttBDLSTM and FCN have been exploited to measure the backward dependency of traffic data, as a building block for a deep architecture model. We conducted comprehensive experiments, and the experimental evaluations illustrated that the proposed architecture achieved better performance compared to state-of-the-art models when considering the short and long traffic speed prediction, e.g., 15, 30, and 60 min, in multistep traffic speed prediction.

Transport practitioners need a universally applicable speed prediction model to estimate average speeds on any road. Average annual speed is a key input to the economic assessment of transport infrastructure where reliable estimates of future average speeds are necessary to calculate economic costs and benefits. The relationship between Annual Average Daily Traffic (AADT) and average annual speed was investigated on higher-order roads across South Africa, revealing a high level of variability in this correlation at different locations. This variation is influenced by road characteristics, such as alignment and cross-section, complicating the formulation of a universal speed prediction model. Two novel speed prediction models are proposed in this article that use AADT to forecast future average annual speed. The speeds of heavy vehicles and light vehicles can be estimated separately, as well as the average speed of all vehicles simultaneously. Both models are self-calibrating, accounting for the variation in the AADT–speed relationship. This calibration step is unique to speed prediction models and increases the reliability of these models to estimate future average speeds considerably. Furthermore, self-calibrating average annual speed prediction models are universally applicable and will simplify economic assessment of transport infrastructure. [ABSTRACT FROM AUTHOR]

The future motion prediction of vehicles in the front is widely valued for its great potential to improve a vehicle’s safety, fuel consumption, and efficiency. However, due to the uncertainty of a driver’s driving intentions and vehicle dynamics, future motion prediction faces great challenges. In order to break the bottleneck in the prediction of leading vehicle motion, this paper proposes a prediction idea of decoupling the prediction of leading vehicle motion into vertical vehicle speed prediction based on the Gaussian process regression algorithm and horizontal heading angle prediction based on the long short-term memory method, which combines the predicted vehicle speed and heading angle to derive the future trajectory of the leading vehicle. Moreover, we propose a prediction algorithm of the leading vehicle motion based on the combination of driving intention recognition and multimodel prediction results by the Fuzzy C-means algorithm, which tries to solve the problem of the unclear driving intention of the predicted object and the nonlinearity between the future motion of the vehicle and the environment. Finally, the algorithm is validated using real vehicle data, proving that it has high prediction accuracy.

Heavy rainfall events are becoming more common in many areas with escalating climate change, and these events can considerably affect travel speed and road safety. It is critical to understand when and how rainfall events affect congestion in the transportation network to help improve decision making for infrastructure planning and real-time operations. This study incorporates high-resolution rainfall and wind data into a travel speed prediction model, along with other data related to weather conditions, incidents, and real-time speeds, to assess if localized rainfall data can inform travel speed prediction during light and heavy rainfall events, and how this compares with the classical method of using a single city-wide rain gauge data point. The travel speed prediction model holistically selects the most related features from a high-dimensional feature space by modeling by wind direction, correlation analysis, and the least absolute shrinkage and selection operator (LASSO) to overcome overfitting issues and is applied to two urban arterials for case studies located in Pittsburgh, Pennsylvania. The results indicate that high-resolution rainfall features in many instances are better predictors of future rainfall on the target segments, leading to overall better prediction results (in 30-min lag time), when compared with models that use a single city-wide rain gauge. This has implications for other cities that are interested in improving travel speed prediction modeling and traffic modeling under increasing impacts from climate change and extreme weather. [ABSTRACT FROM AUTHOR]

While battery electric vehicles (EVs) are on the advance, the broad customer basis is still concerned about battery electric range, a phenomenon commonly known as range anxiety. To tackle these concerns, the functionality of in-vehicle navigational systems must adapt to the new propulsion technology with a limited battery capacity. Central aim is to consider charging infrastructure in route planning. Furthermore, detailed powertrain models are required to accurately forecast an EV’s energy consumption. On the other hand, such detailed models are hardly applicable to large scale road networks that are usually handled by routing services for vehicle navigation. This study proposes a two-staged approach to compute time optimal routes for EVs. To this end, a reduced road network is obtained from a leading routing service. Subsequently, a detailed consumption model is applied and the resulting multiobjective shortest path problem is solved using an adapted Moore-Bellman-Ford algorithm. Within an experimental study, the consumption forecast is validated against measurement data and query times of the proposed methodology are assessed for generic routing problems. The former shows significant improvement of consumption forecast accuracy compared to state-of-the-art models while the latter indicates potential for application in car manufacturers vehicle backend services. [ABSTRACT FROM AUTHOR]

Bypass pipeline inspection gauges have the advantages of low cost and bringing no consumption in transportation efficiency and have been widely used in pipe cleaning, inspecting, and maintaining operations. The moving speed of bypass pipeline inspection gauges will seriously affect the results of the operations, so there are strict requirements on the moving speed of bypass pipeline inspection gauges. Because the moving speed of pipeline inspection gauge is difficult to measure or control in real time, it is important to predict it. This paper studies the influencing factors and their impact methods of pipeline inspection gauges' motion. Through the combination of computational fluid dynamics simulation and friction mathematical model, the relationship between the value of the bypass hole diameter and the pipeline inspection gauges' moving speed was studied. Under the selected research conditions, when the diameter of the bypass hole is increased from 0.1 to 0.5 m, the moving speed of pipeline inspection gauge in water and crude oil is, respectively, decreased from 2.779 to 0.589 m/s and from 2.777 to 0.373 m/s, and the relationship between them can be approximately described by a function. Based on this principle, the moving speed of pipeline inspection gauge can be predicted mathematically. The experiments also indicate that the density and dynamic viscosity of the transport medium and the deformation amount of the bypass pipeline inspection gauge sealing disk will affect the movement state of pipeline inspection gauge in the pipeline. This research has guiding significance for the design of the pipeline inspection gauges' structure size, which is beneficial to the pipeline robot to better meet the needs of cleaning, inspecting, and maintaining operations, and has reference value for related researches. [ABSTRACT FROM AUTHOR]

Real-time traffic speed prediction is an essential component of intelligent transportation systems applications on large-scale urban networks, e.g., proactive traffic management, advanced information provision, and prompt incident response. The family of traffic prediction models (e.g., convolutional neural networks) based on multi-detector speed diagrams in the time-space plane has been one of the most frequently used approaches for individual roads and the entire network. However, the predefined stacking sequence of traffic detectors along the spatial dimension of the speed diagram has a significant influence on the prediction performance, which makes network-wide speed prediction more challenging. To tackle the above challenge and better capture complicated traffic dynamics, we propose a novel speed prediction approach, named spatial–temporal deep tensor neural networks (ST-DTNN), for a large-scale urban network with mixed road types. Spatial and temporal dependencies of different road segments are simultaneously taken into account to improve the network-wide prediction accuracy. A scalable deep tensor is constructed for the ST-DTNN to eliminate the potentially negative impact caused by the manually stacking sequence of speed time series collected at different locations. Multi-step ahead traffic speeds can be simultaneously predicted based on probe data for a real-world large-scale urban network with hundreds of detectors installed on freeways, highways, and major/minor arterials. The results demonstrate the capability and effectiveness of the proposed ST-DTNN approach. Compared with the benchmark models, the ST-DTNN performs higher prediction accuracy during either peak or off-peak periods within an acceptable training time and has more stable prediction performance on the spatial scale. The proposed approach can be extended to develop network-wide traffic state monitoring, optimize routing in navigation services, and support congestion mitigation. [ABSTRACT FROM AUTHOR]

Featured Application: This work provides a tool for determining the adequate spacing between vertical traffic calming measures as a function of the target speed in the segment. Traffic calming measures (TCM) are placed in urban areas to improve road safety, and among them, vertical TCMs are widely employed. Many researches are focused on the influence of the geometry of each measure on speed reduction, but it is demonstrated that drivers forget its effect and speed up after it. Therefore, placing consecutive TCMs can help to maintain a safe area. However, scarce literature can be found about the adequate spacing between them. Hence, the aim of this paper is to analyze the adequate distance between TCMs. Various streets with variable distances and different vertical TCMs were evaluated in Poland and Spain, including raised crosswalks, raised intersections, speed humps and speed cushions. The intermediate point between two TCMs was selected as the place where the maximum speed is achieved. Results showed that there was a good correlation between the speeds at intermediate points and the distance between TCMs, with a determination coefficient around 0.80. For an 85th percentile of the speed under 50 km/h, a maximum distance of 200 m between TCMs is recommended, and for a value of 40 km/h, 75 m. [ABSTRACT FROM AUTHOR]

The digital twin, with its descriptive and predictive services provides promising prospects in the automotive field. Increased insights in customer behavior and vehicle states through descriptive models enable the predictive services of the digital twin to project these vehicles in the future. Currently, most of the predictive services of the digital twin focus on a singular dimension, the asset itself, without considering the influences of the environment and the driver, which hampers the reliability and accuracy of these models. The present work aims to provide a solution in the form of a holistic digital twin comprising of three descriptive models representing the motorcycle itself, its operating environment, and the motorcycle riding behavior. Based on data gathered during a large-scale measurement campaign, novel insights in the motorcycle riding behavior have enabled its representation into two mathematical formulations. Highlighting the capability of the digital twin to integrate data from heterogeneous sources, the environmental model is generated using geospatial data from a map provider, followed by a novel formulation of a safety driving line. Using a kinematic motorcycle model, the speed and banking angle over the defined route are predicted with high correlation to real-world motorcycle riding behavior. The insights generated by the developed digital twin can be used to enable data-driven development or as an input to provide individualized predictive services during the usage phase. • Generation of insights into real-world naturalistic motorcycle riding behavior. • Development of an environmental model using road-related topology data. • Three-dimensional digital twin of a real-world motorcycle fleet. • Validation of digital twin prediction with real-world measurements. [ABSTRACT FROM AUTHOR]

Resource scarcity, global climate change and environmental pollution are increasingly constraining the development of the automotive industry. China proposes to reach the carbon peak by 2030; to reach the carbon neutral double carbon target by 2060 and gradually promote a green and low-carbon transition in energy development. The development of new energy vehicles is an important approach for China to realize its energy structure transformation in the automobile industry. HEV, as a transitional product of automobile energy transformation, has the advantages of both internal combustion engine vehicles and electric vehicles, which can improve the fuel efficiency and the emission problem of internal combustion engine vehicles and the range is longer compared to electric vehicles. One of the important aspects of HEV research is the design of whole vehicle energy management strategy based on the model predictions. Particularly, model-based predictive control is one of the mainstream energy management strategies nowadays, and its optimization effect is mainly subject to the model prediction accuracy. In this study, we constructed the ITS environment of a local roadway through simulation, compared the speed prediction effects of different speed prediction methods in different prediction time domains, and fused the historical information of vehicles (speed of the vehicle in front, distance, signal status, distance, and remaining time). It is found that N-BEATS is more effective in predicting vehicle speed in different prediction time domains, and the prediction accuracy of the speed prediction model is effectively improved after its fusion of multivariate information. • Speed prediction based on real driving data. • Comparing the accuracy of different speed prediction methods in different prediction time domains. • Fusion of multivariate information (vehicle history, front vehicle speed, distance, signal light status, remaining time, and other factors) for speed prediction. [ABSTRACT FROM AUTHOR]

This paper aims to incorporate the throttle handle level prediction in high speed train(HST) operation prediction problem to enable the prediction of HST drivers' activities, in which the key instructions available to HST driver are difficult to determine. Specifically, we consider an end-edge-cloud orchestration system to capture the real-time responses for driver state changes. By adding edge computing nodes, the real-time performance of data collection, transmission, and processing is improved. Our ultimate goal is to guide and regulate train drivers' activities in the same way, regardless of uncertain factors affecting HST dynamic or kinematic performance. We formulate the problem as a physical-based and data-driven deep learning-aided prediction model and solve it using a novel long short-term memory (LSTM) deep neural network which combines: (i) an off-line approximate training model to learn the time series data in the cloud layer, and (ii) an online prediction process to determine driving strategies in the real-time windows, more in general expressed as driving skill level constraints. To evaluate the performance of our approach, some case studies using the real-world railway infrastructure and HST data have been conducted. The results show that the proposed models produce higher prediction accuracy for both speed and throttle handle level prediction tasks. Compared to the conventional HST operation prediction problem, which considers speed sequences only without throttle handle level consideration, this study finds that jointly modeling speed and throttle handle level actually improves the next operation prediction performance itself, potentially because throttle handle level observations capture the information on HST control dynamics, which may affect operators' driving choices. • Structure. Edge computing and prediction scheme are innovatively combined with HST data collection to meet the delay-sensitive and sequential-aware operation requirements, and a novel HST operation structure for end-edge cloud orchestration system is built. • Modeling. A combination of the physical-based and the data-driven deep learning model is formulated for HST operation prediction under uncertainty in throttle handle level and speed, which is a new model in the literature. As discussed, the existing research has only considered speed dynamics in real-time DAS limitedly. • Practice. From the field collected data in Beijingxi–Zhengzhoudong HSR (as shown in Figure 5), we test the effectiveness of our proposed model with seven cases. Since the real-time driving advice is calculated in a few milliseconds, our two-phase solution method can be practically relevant when designing DAS or ATO systems. More importantly, most time-consuming calculations are executed off-line in the cloud layer and the amount of on-line calculations in edge layer is limited to looking up a table. [ABSTRACT FROM AUTHOR]

Fuel cell electric vehicles have earned substantial attentions in recent decades due to their high-efficiency and zero-emission features, while the high operating costs remain the major barrier towards their large-scale commercialization. In such context, this paper aims to devise an energy management strategy for an urban postal-delivery fuel cell electric vehicle for operating cost mitigation. First, a data-driven dual-loop spatial-domain battery state-of-charge reference estimator is designed to guide battery energy depletion, which is trained by real-world driving data collected in postal delivery missions. Then, a fuzzy C-means clustering enhanced Markov speed predictor is utilized to project the upcoming velocity. Lastly, combining the state-of-charge reference and the forecasted speed, a model predictive control-based cost-optimization energy management strategy is established to mitigate vehicle operating costs imposed by energy consumption and power-source degradations. Validation results have shown that 1) the proposed strategy could mitigate the operating cost by 4.43 % and 7.30 % in average versus benchmark strategies, denoting its superiority in term of cost-reduction and 2) the computation burden per step of the proposed strategy is averaged at 0.123 ms, less than the sampling time interval 1s, proving its potential of real-time applications. • A cost-optimization energy management is devised for postal-delivery FCEVs. • A data-driven dual-loop spatial-domain battery SoC reference estimator is devised. • Both energy consumption and power source durability are accounted. • A comprehensive evaluation of power-allocation and operating costs is conducted. [ABSTRACT FROM AUTHOR]

Vehicle speed prediction can obtain the future driving status of a vehicle in advance, which helps to make better decisions for energy management strategies. We propose a novel deep learning neural network architecture for vehicle speed prediction, called VSNet, by combining convolutional neural network (CNN) and long-short term memory network (LSTM). VSNet adopts a fake image composed of 15 vehicle signals in the past 15 s as model input to predict the vehicle speed in the next 5 s. Different from the traditional series or parallel structure, VSNet is structured with CNN and LSTM in series and then in parallel with two other CNNs of different convolutional kernel sizes. The unique architecture allows for better fitting of highly nonlinear relationships. The prediction performance of VSNet is first examined. The prediction results show a RMSE range of 0.519–2.681 and a R2 range of 0.997–0.929 for the future 5 s. Finally, an energy management strategy combined with VSNet and model predictive control (MPC) is simulated. The equivalent fuel consumption of the simulation increases by only 4.74% compared with DP-based energy management strategy and decreased by 2.82% compared with the speed prediction method with low accuracy.

With the development of technology, speed prediction has become an important part of intelligent vehicle control strategies. However, the time-varying and nonlinear nature of vehicle speed increases the complexity and difficulty of prediction. Therefore, a CNN-based neural network architecture with two channel input (DICNN) is proposed in this paper. With two inputs and four channels, DICNN can predict the speed changes in the next 5 s by extracting the temporal information of 10 vehicle signals and the driver’s intention. The prediction performances of DICNN are firstly examined. The best RMSE, MAE, ME and R2 are obtained compared with a Markov chain combined with Monte Carlo (MCMC) simulation, a support vector machine (SVM) and a single input CNN (SICNN). Secondly, equivalent fuel consumption minimization strategies (ECMS) combining different vehicle speed prediction methods are constructed. After verification by simulation, the equivalent fuel consumption of the simulation increases by only 4.89% compared with dynamic-programming-based energy management strategy and decreased by 5.40% compared with the speed prediction method with low accuracy.

Due to the increase of congestion on highways, providing real-time information about the traffic state has become a crucial issue. Hence, it is the aim of this research to build an accurate traffic speed prediction model using symbolic regression to generate significant information for travellers. It is built based on genetic programming using Pareto front technique. With real world data from microwave sensor, the performance of the proposed model is compared with two other widely used models. The results indicate that the symbolic regression is the most accurate among these models. Especially, after an incident occurs, the performance of the proposed model is still the best which means it is robust and suitable to predict traffic state of highway under different conditions.