Search

Cities of all sizes around the world are facing increasing levels of congestion, which leads to increasing travel time and emissions, and ultimately affects the quality of life. Relevant research suggests that adaptive traffic light control systems can improve the traffic flow, but their impact on energy-efficiency of vehicle propulsion systems is not well understood. In this study, we use Proximal Policy Optimization, a Deep Reinforcement Learning algorithm, to develop an optimized adaptive traffic light control systems that controls three traffic lights simultaneously. For this purpose, we have created a microscopic traffic simulation of the city of Aachen, Germany, calibrated on the basis of traffic measurements, where the actual traffic light schedule, a green-wave, fixed-time control scheme, serves as a reference. The traffic simulation is coupled with detailed, physics-based powertrain models of both conventional and electric vehicles, which are validated against chassis dynamometer measurements. By analyzing the complex interactions between traffic light control, the resulting vehicle trajectories and the powertrain components, we show that Reinforcement Learning-based adaptive control can significantly improve the traffic flow, with a 41% increase in average velocity, without any drawbacks in CO2 emission (−1%). Furthermore, we find that maximizing traffic flow and minimizing CO2 emissions are not necessarily contradictory objectives, and identify an increased energy saving potential at low traffic densities. Thus, we prove that adaptive traffic light control can make traffic not only more time-efficient, but also more sustainable.

A variety of longitudinal control systems employing predictive information provide a long-term velocity profile that minimizes the cost term consisting of the superposition of energy, ride comfort and travel time. The profile can be determined by discrete dynamic programming (DDP) with a three-dimensional state space, resulting in unacceptable high computation times for real-time applications. Previous studies have used DDP with two dimensions to reduce computation time. This approach, however, leads to suboptimal solutions when the maximal velocity varies. We present an iterative algorithm called waterfilling that finds the same solution as three-dimensional DDP for arbitrary velocity limitations, while its computational complexity is orders of magnitude lower that that of DDP with two dimensions. The waterfilling algorithm minimizes the energy at the wheels and reaches the upcoming traffic light at a given point in time. For vehicles powered by an internal combustion engine, the waterfilling algorithm is extended by an efficient traffic light approach unit that avoids low vehicle velocities to increase engine efficiency. The long-term planning unit is succeeded by a short-term planning unit that handles dynamic traffic and incorporates the fuel-saving pulse and glide driving strategy. The longitudinal control system developed in this paper is validated within a dynamic simulation environment with variable traffic density and compared with a longitudinal control system employing model-predictive control. For mild-hybrid electric vehicles, the developed system achieves fuel savings of $20\,\%$ at low traffic densities. Activation of pulse and glide adds between $2.4\,\%$ to $3.7\,\%$ to the savings, depending on traffic density.

Predictive cruise control functions designed to reduce the energy consumption of intelligent and automated vehicles require an accurate prediction of the upcoming traffic situation in general and the preceding vehicle in particular. This article presents the implementation of prediction models which do not rely on explicit information of the entire preceding vehicle queue. Using this assumption, prediction algorithms based on Conditional Linear Gauss (CLG) models and Deep Neural Network (DNNs) were trained using real-world measurements in an urban setting. The training was conducted with 6.6 hours of driving data specifically collected for this study. The results show that both approaches can provide accurate estimates of a preceding vehicle’s motion. For the CLG models, a queue-estimation logic was implemented to improve the prediction accuracy. The most accurate prediction could be obtained when DNNs are set up with Long Short-Term Memory (LSTM) layers, capable of modeling time dependencies. The results also show that accurate short and mid term predictions of a vehicle’s trajectory can be made based primarily on sensor data, even when no or only little data from vehicle-to-vehicle (V2V) communication is available.

The development of embedded systems is a time-consuming process that relies on the calibration of the control systems by experienced engineers and the availability of prototype vehicles. Modern embedded systems must operate with non-linear functions in a wide range of applications. As a result, even experts are unable to fully comprehend all system interconnections, and optimal performance is rarely achieved. Machine Learning (ML)-based techniques can redefine the conventional approach to decision-making problems in this field. Because Reinforcement Learning (RL) algorithms are self-adaptive, near-optimal solutions can be developed with minimal human intervention, providing significant potential for cost and time savings. In this paper, it is demonstrated how RL can be used to design a control function that competes with and outperforms a state-of-the-art controller, while ensuring the stability of the system. In the Model-in-the-Loop (MiL) framework, a method of accelerated training for the control of the turbocharger is presented that considers both performance and emissions criteria. Data efficiency and training robustness were tested for the Proximal Policy Optimization (PPO) algorithm. Selected policies have been evaluated under transient conditions and are shown to enhance the overall boost pressure accuracy by up to 23%, while reducing cumulative NOxand soot emissions by 4% and 10%, respectively, compared to the reference controller. This work demonstrates another step towards the application of the PPO algorithm in the development of embedded systems for complex control problems. Due to its proven robustness and stability, the method is suitable for transfer to real hardware applications.

Homogeneous charge compression ignition (HCCI) is a promising combustion process to reducing both greenhouse gas and pollutant emissions in the transportation sector. The combustion starts when the thermodynamic state of the cylinder charge reaches the auto-ignition properties of the fuel, which is challenging to model. In addition, the strong cyclic coupling due to required exhaust gas recirculation leads to a significant dependence on the combustion of the previous cycle.

Homogeneous charge compression ignition (HCCI) is a low-temperature combustion process that can both increase efficiency and reduce pollutant raw emissions. The process is highly dependent on low-temperature kinetics of the chemical reactions that lead to auto-ignition. Due to strong coupling of consecutive combustion cycles stable control of HCCI is a major challenge. Ion current sensors that detect ionization in the combustion chamber can provide information about chemical reactions taking place inside the cylinder. In contrast, model-based HCCI control strategies, which require accurate models that describe the current cylinder state, are often based only on the cylinder pressure. However, low-temperature chemical kinetics are not sufficiently taken into account by only using the cylinder pressure signal. Therefore, the main hypothesis of this paper is that HCCI strongly depends on the chemical state in the cylinder, which cannot be adequately described by the pressure sensor in the cylinder alone. Hence, a conventional spark plug is used to apply an ion current sensor, which is used alongside the pressure sensor. The aim is to use further information about the state of the chemical mixture for the control of the engine. The features derived from the ion current are integrated into an explicit artificial neural network-based controller. The developed control strategies are then compared against a purely pressure-based controller. Using the ion current features, the standard deviation of the combustion phasing CA50 is reduced by up to 42% compared to the pressure-based approach. The experimental results of this work show that the inclusion of the ion current in the controller provides a significant advantage for low temperature combustion control.

For cost and space reasons, the electric machine (EM) and its cooling system in electrified powertrains for road vehicles are usually designed in such a way that the maximum power cannot be called up continuously. For this purpose, control systems are used that derate the power of the EM depending on its temperature. In this context, this paper presents the integration of a Recurrent Neural Network (RNN) model into a nonlinear model predictive control (MPC) to efficiently control the driving performance of an electric vehicle (EV) on a race track and reduce thermal derating effects. The discrete black-box RNN model incorporating a Long Short-Term Memory (LSTM) layer describing the thermal dynamics of the electric machine is combined with the continuous-time one-dimensional vehicle dynamics to form the hybrid MPC. For the selected example, the RNN-MPC outperforms a standard linear derating strategy by 6.23% in terms of energy consumption while achieving equal lap speed and maximum machine temperatures. Compared to an existing grey-box thermal network model, the RNN model significantly reduces the temperature prediction error by 28%. The trajectory tracking problem is formulated using acados and deployed on a dSPACE SCALEXIO embedded system to meet real-time requirements.

Internal combustion engines face increasingly stringent pollution and greenhouse gas emission restrictions. In this regard, the technology of low-temperature combustion is the scope of research, as it can reduce pollutant emissions while increasing engine efficiency. However, the system dynamics are very complicated and have a high sensitivity concerning the thermodynamics in the combustion chamber. Modeling with first principle models would be time-consuming and requires accurate parameter estimations. Therefore, this work presents deep learning-based models to capture the dynamics of the cylinder pressure trace for gasoline controlled auto-ignition. These models trained with experimental data can learn the internal pattern of the combustion cycle dynamics. Based on information from the current cycle, they enable the prediction of the combustion pressure trace for the following. The results show that the models precisely predict the pressure traces and ensure high prediction accuracy of engine performance parameters derived from the predicted traces. R2values of 0.95 or more are obtained for global parameters, whereas values of 0.8 or more are obtained for local parameters.

This paper presents a comprehensive evaluation of system interactions in a battery electric vehicle caused by temperature sensitivity of permanent magnet synchronous machines (PMSM). An analytical model of a PMSM considering iron losses and thermal impact is implemented on a field programmable gate array suitable for hardware-in-the-loop testing. By the presented virtual prototyping approach, different machine characteristics defined by the design are used to parameterize the analytical model. The investigated temperature effect is understood as an interacting influence between machine characteristics and control, which are investigated in terms of torque generation, voltage utilization and efficiency under closed-loop condition in a vehicle environment. In particular, using a surface permanent magnet rotor and an interior permanent magnet rotor, the performance of both machine designs is analyzed by varying temperature-adjusted feedforward control strategies on the basis of a driving cycle from a racetrack. The comparison shows that the machine design with surface-mounted magnets is associated with higher temperature sensitivity. In this case, the temperature consideration in the feedforward control provides a 14%loss reduction in closed-loop vehicle test operation. It can be summarized that the electromagnetic torque is less sensitive to a temperature variation with increasing reluctance. The presented development approach demonstrates the impact of interactions in electric powertrains without the need of real prototypes.

The high thermal efciency and reliability of the compression-ignition engine makes it the first choice for many applications. For this to continue, a reduction of the pollutant emissions is needed. One solution is the use of Machine Learning (ML) and Model Predictive Control (MPC) to minimize emissions and fuel consumption, without adding substantial computational cost to the engine controller. ML is developed in this paper for both modeling engine performance and emissions and for imitating the behaviour of a Linear Parameter Varying (LPV) MPC. Using a support vector machine-based linear parameter varying model of the engine performance and emissions, a model predictive controller is implemented for a 4.5 L Cummins diesel engine. This online optimized MPC solution offers advantages in minimizing the NOxemissions and fuel consumption compared to the baseline feedforward production controller. To reduce the computational cost of this MPC, a deep learning scheme is designed to mimic the behavior of the developed controller. The performance in reducing NOxemissions at a constant load by the imitative controller is similar to that of the online optimized MPC, however, the imitative controller requires 50 times less computation time when compared to that of the online MPC optimization.

Numerous systems engineering (SE) methods for the model‐based and textual specification of systems focus on managing complexity solely by partitioning the system based on physical structures or by defining different views of the system and therefore reach their limits in agile development. The increasing demand for an agile system development requires an agile systems engineering procedure for the model‐based and textual top‐down specification of systems. Although function‐based development, variant management, and product line development are well established in software engineering, previous work has failed to introduce methods for the agile specification of systems by combining established methods from systems and software engineering. For that purpose, this paper demonstrates a new SE methodology, which for the first time combines conventional SE methods with the agile development procedure of feature‐driven development. The methodology is systematically developed based on theoretical analyses and its suitability for the application‐specific definition of feature‐driven development processes is demonstrated using the example of reference architectures for XiL simulation models of electric vehicles. By applying feature‐driven development, the resulting CUBE methodology enhances collaboration in interdisciplinary development teams and enables companies to adapt development processes to a more agile top‐down specification of systems.

To meet the challenges in software testing for automated vehicles, such as increasing system complexity and an infinite number of operating scenarios, new simulation methods must be developed. Closed-loop simulations for automated driving (AD) require highly complex simulation models for multiple controlled vehicles with their perception systems as well as their surrounding context. For the realization of such models, different simulation domains must be coupled with co-simulation. However, widely supported model integration standards such as functional mock-up interface (FMI) lack native support for distributed platforms, which is a key feature for AD due to the computational intensity and platform exclusivity of certain models. The newer FMI companion standard distributed co-simulation protocol (DCP) introduces platform coupling but must still be used in conjunction with AD co-simulations. As part of an assessment framework for AD, this paper presents a DCP compliant implementation of an interoperable interface between a 3D environment and vehicle simulator and a co-simulation platform. A universal Python wrapper is implemented and connected to the simulator to allow its control as a DCP slave. A C-code-based interface enables the co-simulation platform to act as a DCP master and to realize cross-platform data exchange and time synchronization of the environment simulation with other integrated models. A model-in-the-loop use case is performed with the traffic simulator CARLA running on a Linux machine connected to the co-simulation master xMOD on a Windows computer via DCP. Several virtual vehicles are successfully controlled by cooperative adaptive cruise controllers executed outside of CARLA. The standard compliance of the implementation is verified by exemplary connection to prototypic DCP solutions from 3rd party vendors. This exemplary application demonstrates the benefits of DCP compliant tool coupling for AD simulation with increased tool interoperability, reuse potential, and performance.

48V systems enable not only mild hybrid functionalities such as recuperation or torque assist by a belt-driven starter generator (BSG), but also electrification of accessories and the engine boosting system. To maximize the powertrain efficiency, a proper layout of the electrified system and an optimized distribution of the electric power during transient operation is essential. In this study, a vehicle co-simulation of a conventional powertrain with a downsized turbocharged gasoline engine is extended by a 48V system with an electric compressor (eC) and a BSG. The control functions of the eC and BSG are based on a state-of-the-art vehicle application and calibrated for transient operating conditions. The engine model, which is built using a one-dimensional crank angle resolved approach in GT-POWER, has been validated with measurement data and is used to predict the interaction between the eC and the engine air path. The investigations using the simulation platform show that the 48V eC and the BSG can significantly improve the fuel effïciency if the electric energy consumption is initially neglected. However, when considering the electric energy consumption within the vehicle co-simulation, efficient operation is particularly depending on driver torque demand, the battery state-of-charge and charging effïciency. Hence, intelligent operating strategies are necessary to take advantage of the better torque response and improve fuel consumption at the same time.

The ongoing trend toward powertrain electrification and vehicle automation enables the exploitation of additional energy saving potential through the joint optimization of the driving trajectory and the hybrid management. To tackle this complex control task within the E-COSM 2021 Benchmark Challenge, we combine a GLOSA algorithm generating a uniform speed profile with an optimization-based hybrid powertrain controller employing an equivalent consumption minimization strategy.

This paper presents a real-time capable nonlinear model predictive control (NMPC) strategy to effectively control the driving performance of an electric vehicle (EV) while optimizing thermal utilization. The prediction model is based on an experimentally validated two-node lumped parameter thermal network (LPTN) and one-dimensional driving dynamics. An efficient solver for the trajectory tracking problem is exported using acados and deployed on a dSPACE SCALEXIO embedded system. The lap time of a high-load driving cycle compared to a state-of-the-art derating strategy improved by 2.56% with an energy consumption reduction of 2.43% while respecting the temperature constraints of the electric drive.

Shifting automotive powertrain development tasks to earlier phases (frontloading) increases efficiency by utilizingtest-benches as opposed to prototype vehicles (road-to-rig approach). The coupling of distributed test-benches by a virtualized shaft connection is required to reproduce interactions of automotive powertrain components. A coupling algorithm simulates a rigid connection by synchronizing the torque and speed of two distributed test-bench’s electric motors. System dead-times lead to limited stability and reduced bandwidth of the coupling algorithm. In this study, a method for a stable bidirectional coupling of speed and torque of both subsystems is described analytically and verified by simulation. All component models are calibrated based on measurements using state-of-the-art test-bench equipment. A distributed prediction algorithm is proposed for the dead-time compensation. Four Kalman predictors estimate the coupling values of both subsystems at wall-clock-time without measurement and communication latencies. A detailed drive cycle analysis is performed through simulation. This enables a Virtual Shaft Algorithm to achieve a higher bandwidth and an improved coupling robustness.

In this article, we analyze the interaction between powertrain technology, predictive driving functionalities, and inner-city traffic conditions. A model predictive velocity control algorithm is developed that utilizes dynamic traffic data as well as static route information to optimize the future trajectory of the considered ego-vehicle. This controller is then integrated into a state-of-the-art simulation environment for automated driving functionalities to calculate energy saving potentials for vehicles with a conventional gasoline engine powertrain and a P3-hybrid powertrain configuration as well as for a battery electric vehicle based on real driving measurements. The comparison of these powertrains under various traffic conditions shows that all three technologies profit from predictive driving functionalities. The determined reduction in energy demand ranges from 15% to more than 40%, but it is highly dependent on the boundary conditions and the selected powertrain technology. Specifically, it is shown that electrified powertrains can profit the most when the time-gap to the preceding vehicle is maintained at a high level. For a conventional powertrain, this effect is less pronounced and can be attributed to the efficiency characteristics of gasoline engines. It can be concluded that the development of advanced predictive driving functionalities requires microscopic simulation of inner-city traffic to achieve optimum results with regard to energy consumption.

Virtualization of powertrain components allows the front-loading of conventional vehicle calibration and validation tasks to Model-in-the-Loop (MiL) and Hardware-in-the-Loop (HiL) simulations. This approach is based on the utilization of highly accurate physics-based powertrain models that enable a seamless system validation using virtual testing methods in order to ensure cost-effective powertrain development by reducing hardware tests. Proper modelling methods target the optimum between parametrization effort, model accuracy and required computing power to grant the real-time (RT) capability of the simulation, which is mandatory for HiL simulation. In this paper two validated modelling approaches and their implementation into a MiL environment are introduced and discussed. The approaches are the MATLAB/Simulink based Mean Value Engine Model (MVEM) and the Fast-Running Modelling (FRM) of GT-Power. After the models integration in a Simulink frame, the responses of a model-based control unit with the two simulation models were evaluated using real experimental data. In transient cycles, the controller showed a different reaction to the feedback signals of the two engine models. The purposes of the conducted investigation are mainly to evaluate strong and weak points of both approaches and to propose the best-practice modelling approaches for virtual calibration and validation. A comparative rating shows the main advantage of the MVEM in the flexibility for HiL-based systems and the model training effort for the FRM.

With the goal of enhancing the efficiency of electric vehicles and improving the driving range, the appropriate dimensioning, so-called “right-sizing,” of the electric machines is a promising approach. This paper presents an efficient approach of scaling the thermal parameters in a low order lumped parameter thermal network (LPTN) model to estimate the temperature of the scaled permanent magnet synchronous machine (PMSM) by taking into account the temperature dependence of the power losses. At an early stage of development, the proposed scaling approach enables a preliminary evaluation of the thermal limit of the PMSM within the optimization of the electric powertrain and offers a notable benefit in that the low order LPTN model can easily be parameterized and scaled based on the experimental measurements. Validation for both axial scaling and radial scaling with factors ranging from 0.8 to 1.2 was carried out on a previously validated Ansys Motor-CAD model for typical automotive driving cycles. The maximum temperature error between the scaling approach and the Ansys Motor-CAD model is less than 3.5°C.

An increasing level of driving automation and a successive electrification of modern powertrains enable a higher degree of freedom to improve vehicle fuel efficiency and reduce pollutant emissions. Currently, both domains themselves, driving automation as well as powertrain electrification, face the challenge of a rising development complexity with extensive use of virtual testing environments. However, state-of-the-art virtual testing environments typically strictly focus on just one domain and neglect the other. This paper shows the results of a simulation-based case study considering both domains simultaneously. The influence of energy saving automated functionalities on a conventional, a hybrid, and a pure electric powertrain is investigated for a carefully selected inner-city driving scenario. The vehicle simulation models for the different powertrain configurations are calibrated using test bench results and vehicle measurements. A model predictive acceleration controller is developed for realizing the speed optimization function. By considering traffic conditions such as traffic light schedules and a preceding vehicle as the boundary conditions, unnecessary accelerations and decelerations are avoided to reduce the energy demand. The case study is realized by applying this function to the three powertrains variants. As a final result, a clear difference in energy demand is observed: the hybrid powertrain benefits the most in terms of energy demand reduction in the given use case. The results clearly underscore that in future vehicle development programs, the powertrain and the real-world driving functionalities have to be optimized simultaneously to minimize the energy demand during everyday vehicle operation.

This study investigates the control quality of an electric cam phaser system. The impact of different sensor concepts, synchronization algorithms, controller and hardware topologies on the control quality is examined by using a transient simulation covering the electric cam phaser, valve train, mechanical transmission and a wide variety of cam- and crankshaft trigger wheels. Limited angular accuracy effects are simulated by realistic sensor models and the processing of sensor signals by a real-time capable synchronization algorithm. Nonlinear friction in transmission and valve train are considered by the simulation accordingly. Furthermore, the effects of distributed controller algorithms based on conventional electronic control units are evaluated. Communication latencies have a strong impact on the control plant and are taken into account during controller definition. The effects of different layouts are compared in the time domain, and a sensitivity analysis is carried out to evaluate the effects of different parameters on the cam phasing control quality. The control quality is measured in terms of overshoot, phasing duration and energy consumption of a phasing event. Using a sensor fusion for the current cam phasing angle and an integrated controller layout – that is, an architecture without any communication delay – improves the controllability and reduces overshooting, phasing duration and electrical energy consumption under transient conditions.

New electric and hybrid vehicle propulsion architectures require the application of sophisticated control algorithms. The calibration and the validation of these control algorithms is nearly as important and time consuming as the development of the algorithms itself. Testing with the real vehicle is very expensive and only possible in late development stages. Calibration and testing tasks can be shifted to earlier phases by using precise co-simulation and Hardware-in-the-Loop approaches. This study shows the advantages of frontloading of development tasks by implementing a virtual P2-Hybrid in an Electrical-Motor-in-the-Loop test bench. As baseline, a state of the art vehicle co-simulation is used. The simulation setup is composed of a GT-Power engine model, a dSPACE vehicle dynamics model and a SimulationX power train model. This conventional model is extended by a virtual electrical motor and a hybrid control unit. With the pure Model-in-the-Loop simulation, some calibration and testing tasks are possible, but the simulation lacks of some complex physical properties, e.g. the thermal behavior of the electrical motor. In a next step, the virtual electrical motor is replaced by a real electrical motor on a test bench. The residual simulation is interacting in real-time with the real electrical motor. This enables the possibility to calibrate all relevant controllers to the real component under realistic and dynamic circumstances including mutual interaction of calibration parameter and the physical device under test.

The complexity of automobile powertrains continues to increase, leading to increased development time and effort. One method to address this challenge is to shift vehicle calibration and testing tasks to simulations with Engine-in-the-Loop (EiL) test setups. This increases the efficiency of the development process through high flexibility, low prototype costs, an early starting point, and a high degree of automation. EiL emulates the dynamic interactions between the internal combustion engine and other powertrain components by closed-loop coupling to real-time simulation models. The coupling quality strongly depends on the interface properties, the hardware layout, and the control functionality of the dynamometer test bench. This paper analyzes an existing test bench and presents the impact of altered test bench parameters on different EiL quality aspects. Thereto, the setup is analyzed and a Simulink based grey box test bench model is derived. The latter is parametrized with a full factorial approach in comparison to real test bench measurements. The test bench model is then used for simulative investigations considering different test bench parameters: Two different use cases of Model-in-the-Loop (MiL) supported EiL are investigated. First, a simulative improvement of the test bench control quality by altering the PI based dyno speed controller parameters is investigated. The virtually calibrated controller parameters are applied to the real EiL system and compared to vehicle measurements. Second, sensitivity studies are shown with respect to dynamometer inertia and system dead time in terms of control quality and energy. As a major outcome, a compromise of control accuracy and stability for different speed controller parameters, dyno inertia, and system dead time is shown. Furthermore, it is demonstrated that EiL has only a small deviation in terms of energy compared to a full vehicle, making it a valid approach when considering CO2emissions.