Automotive Powertrain Control Research Articles

Systematic hyperparameter selection in Machine Learning-based engine control to minimize calibration effort

For automotive powertrain control systems, the calibration effort is exploding due to growing system complexity and increasingly strict legal requirements for greenhouse gas and real-world pollutant emissions. These powertrain systems are characterized by their highly dynamic operation, so transient performance is key. Currently applied control methods require tuning of an increasing number of look-up tables and of parameters in the applied models. Especially for transient control this state-of-the-art calibration process is unsystematic and requires a large development effort. Also, embedding models in a controller can set challenging requirements to production control hardware. In this work, we assess the potential of Machine Learning to dramatically reduce the calibration effort in transient air path control development. This is not only done for the existing benchmark controller, but also for a new preview controller. In order to efficiently realize preview, a strategy is proposed where the existing reference signal is shifted in time. These reference signals are then modeled as a function of engine torque demand using a Long Short-Term Memory (LSTM) neural network, which can capture the dynamic input–output relationship. A multi-objective optimization problem is defined to systematically select hyperparameters that optimize the trade-off between model accuracy, system performance, calibration effort and computational requirements. This problem is solved using an exhaustive search approach. The control system performance is validated over a transient driving cycle. For the LSTM-based controllers, the proposed calibration approach achieves a significant reduction of 71% in the control calibration effort compared to the benchmark process. The expert effort and turbocharger experiments used in calibrating transient compensation maps in physics-based feedforward controller are replaced by little simulation time and parametrization effort in ML-based controller, which requires significantly less expert effort and system knowledge compared to benchmark process. The best trade-off between multi-objective cost terms is achieved with one layer and 32 cells LSTM neural network for both non-preview and preview control. For non-preview control, a comparable control system performance is achieved with the LSTM-based controller, while 5% reduction in cumulative NOx emissions and similar fuel consumption is achieved with preview controller.

Regulation of Uncertain Markov Jump Linear Systems With Application on Automotive Powertrain Control

Modeling and control of dynamic systems that experience abrupt changes are challenging and fundamental tasks in applications of different areas of engineering. Among these, control synthesis for autonomous heavy-duty vehicles has the potential to highly benefit transportation systems. That said, we propose a robust recursive regulator for discrete-time Markov jump linear systems with polytopic uncertain data. We assume the uncertainties affect the state-space parameters as well as the transition probability matrix. We formulate an optimization problem using the penalty function method and design a recursive solution, whereby we attain the robust state feedback gains. We impose no restrictions on how fast the uncertainties can change between two consecutive iterations. Conditions for convergence and mean-square stability are well defined in terms of Riccati recursive equations. We also set up a procedure for powertrain polytopic model identification based on CAN bus data of an autonomous heavy-duty ground vehicle. We provide two numerical examples to verify the effectiveness of the robust recursive regulator. Moreover, in a third example, we validate the obtained powertrain model on a trajectory tracking problem and compute acceleration and braking inputs using the robust regulator.

Anti-Windup Recursive Least Squares Method for Adaptive Lookup Tables with Application to Automotive Powertrain Control Systems

Anti-Windup Recursive Least Squares Method for Adaptive Lookup Tables with Application to Automotive Powertrain Control Systems

Design of an eco-gearshift control strategy under a logic system framework

Good access to traffic information provides enormous potential for automotive powertrain control. We propose a logical control approach for the gearshift strategy, aimed at improving the fuel efficiency of vehicles. The driver power demand in a specific position usually exhibits stochastic features and can be statistically analyzed in accordance with historical driving data and instant traffic conditions; therefore, it offers opportunities for the design of a gearshift control scheme. Due to the discrete characteristics of a gearshift, the control design of the gearshift strategy can be formulated under a logic system framework. To this end, vehicle dynamics are discretized with several logic states, and then modeled as a logic system with the Markov process model. The fuel optimization problem is constructed as a receding-horizon optimal control problem under the logic system framework, and a dynamic programming algorithm with algebraic operations is applied to determine the optimal strategy online. Simulation results demonstrate that the proposed control design has better potential for fuel efficiency improvement than the conventional method.

Challenges and solutions in automotive powertrain systems

Automotive powertrain mainly consisting of combustion engine, motor and battery (i.e. special for hybrid powertrain) is a very complicated integration system, and the research on the automotive powertrain control techniques remains hot-spot in past decades. This paper proposes some challenging issues and control solutions of automotive powertrain system from the perspective of the dynamic system theory. The typical characteristics of automotive powertrain system are analysed for control development, and the several control applications using model-based and model-free control design are demonstrated with sufficient experimental validation. In addition, some open issues for future powertrain control development are summarised.

Teaching Reforms in Automotive Power train and Chassis Control System Experiment

The Automotive Power train and Chassis Control System Experiment is a supporting course for Automotive Power train Control System and Chassis Control System. This paper discusses the choice of experimental materials and how to vividly explain theoretical knowledge and conduct experiments.

Nonlinear internal model controller design for wastegate control of a turbocharged gasoline engine

Internal Model Control (IMC) has a great appeal for automotive powertrain control in reducing the control design and calibration effort. Motivated by its success in several automotive applications, this work investigates the design of nonlinear IMC for wastegate control of a turbocharged gasoline engine. The IMC design for linear time-invariant (LTI) systems is extended to nonlinear systems. To leverage the available tools for LTI IMC design, the quasi-linear parameter-varying (quasi-LPV) models are explored. IMC design through transfer function inverse of the quasi-LPV model is ruled out due to parameter variability. A new approach for nonlinear inversion, referred to as the structured quasi-LPV model inverse, is developed and validated. A fourth-order nonlinear model which sufficiently describes the dynamic behavior of the turbocharged engine is used as the design model in the IMC structure. The controller based on structured quasi-LPV model inverse is designed to achieve boost-pressure tracking. Finally, simulations on a validated high-fidelity model are carried out to show the feasibility of the proposed IMC. Its closed-loop performances are compared with a well-tuned PI controller with extensive feedforward and anti-windup built in. Robustness of the nonlinear IMC design is analyzed using simulations.

Numerically-aided Deductive Safety Proof for a Powertrain Control System

The use of deductive techniques, such as theorem provers, has several advantages in safety verification of hybrid systems. There is often a gap, however, between the type of assistance that a theorem prover requires to make progress on a proof task and the assistance that a system designer is able to provide. To address this deficiency we present an extension to the deductive verification framework of differential dynamic logic that allows the theorem prover KeYmaera to locally reason about behaviors by leveraging forward invariant sets provided by external methods, such as numerical techniques and designer insights. Our key contribution is a new inference rule, the forward invariant cut rule, introduced into the proof calculus of KeYmaera. We demonstrate the cut rule in action on an example involving an automotive powertrain control systems, in which we make use of a simulation-driven numerical technique to compute a local barrier function.

Design of observer-based feedback control for time-delay systems with application to automotive powertrain control

A new approach for observer-based feedback control of time-delay systems is developed. Time-delays in systems lead to characteristic equations of infinite dimension, making the systems difficult to control with classical control methods. In this paper, a recently developed approach, based on the Lambert W function, is used to address this difficulty by designing an observer-based state feedback controller via assignment of eigenvalues. The designed observer provides estimation of the state, which converges asymptotically to the actual state, and is then used for state feedback control. The feedback controller and the observer take simple linear forms and, thus, are easy to implement when compared to nonlinear methods. This new approach is applied, for illustration, to the control of a diesel engine to achieve improvement in fuel efficiency and reduction in emissions. The simulation results show excellent closed-loop performance.

Automotive Powertrain Control - A Survey

This paper surveys recent and historical publications on automotive powertrain control. Control-oriented models of gasoline and diesel engines and their aftertreatment systems are reviewed, and challenging control problems for conventional engines, hybrid vehicles and fuel cell powertrains are discussed. Fundamentals are revisited and advancements are highlighted. A comprehensive list of references is provided.

Varying sampling predictive control for time delay systems: application to automotive powertrain control

This paper presents computer control under time varying sampling period and delay and its application to automotive control. In the automotive industry context, the controller has indeed to meet the required performance, but also to offer pertinent tuning parameters and finally to fit in limited computing resources. The proposed approach is described in two stages: First, variable sampling of linear continuous model will be considered. At second, a controller design method based on a time varying observer predictor structure is presented. This method is applied to powertrain control, in order to damp driveline oscillations. The resulting predictive controller is efficient and easy to tune directly on vehicle, without redesign, as attested by experimental results on a test car with a SI engine. Copyright © 2007 IFAC

ACTL strong negation and its application to hybrid systems verification

Model checking procedures for verifying properties of hybrid dynamic systems are based on the construction of finite-state abstractions. If the property is not satisfied by the abstraction, the verification is inconclusive and the abstraction needs to be refined so that a less conservative model can be checked. If the hybrid system does not satisfy the property, this verify–refine procedure usually will not terminate. This paper introduces the concept of strong negation for ACTL formulas as an auxiliary condition that can be verified to obtain a conclusive negative verification result from a finite-state abstraction in certain cases. The concepts are illustrated with an example from automotive powertrain control.

NICELY NONLINEAR ENGINE TORQUE ESTIMATOR

In automotive powertrain control strategy, engine torque estimators are preferred to expensive sensors for obvious economic reasons. In the literature several works have been presented proposing both instantaneous engine torque and mean engine torque estimators. The first ones are based on complex models and tipically are suitable for diagnosis purposes, while the latter estimate the torque by means of static maps. In this work, a nonlinear torque estimator for a spark ignition engine has been developed. The total torque acting at the engine shaft has been obtained solving a tracking problem by means of an LQ control strategy. The novel idea is to reduce the nonlinear engine model to a second order Taylor approximation, named a nicely nonlinear model. It is then linearized via feedback and an LQ controller is designed. This approach has been tested on Mid-Size real time dSPACE simulator showing excellent results both for performance and for computational load.

A Systems Engineering Environment For Integrated Automotive Powertrain Development

Automotive powertrain control system (PCS) development faces tightening environmental requirements, shortening development cycle times and growing complexity. To cope with such an environment, PCS development is moving from a traditional evolutionary to a structured approach. The structured approach is supported by computerised structured analysis methods. However, these methods concentrate on the functional and interface requirements of the product. This paper aims to describe how a systems engineering environment (SEE) can be used for integrated PCS development by addressing not only product functional and interface requirements but also life cycle process and organisational requirements. The particular SEE used, a commercial software package, provides: · notations from different modelling paradigms, used for integrated functional, product, process and organisational modelling; · cross-references, used to link functional, product and life-cycle processes attributes; · frames to capture further information about each data items (e.g. CAD drawings). A major benefit of the application of the SEE for an integrated PCS development is the ability to investigate early in the product development/evolution process the interactions between requirements and attributes not only of the product but also of its life cycle processes and development organisation. It is demonstrated how an SEE can be used to provide better product quality, lower life cycle cost and shorter development time. The application of a systems engineering approach to a product, its life cycle processes and its development organisation in an integrated manner encompasses concurrent engineering. It provides a much broader scope for product development, in this case, the PCS subsystem.

Analysis and Control of Nonlinear Systems

This paper describes my work on nonlinear analysis and control over the last twenty years. The first part of the paper concerns the development of nonlinear analysis tools for predicting stability and forced response characteristics of high speed ground vehicles. The principal motivation was to develop an alternative to “brute force” time domain simulation. The developed tools were extensions of “describing function” or “equivalent linearization” methods for both periodic and stochastic excitation. The “statistical linearization” analysis tools were then extended and applied to design control laws for nonlinear stochastic regulators. The second part of the paper was motivated by control system design for highly nonlinear, multivariable systems, such as automotive powertrain control and aircraft flight control. For these classes of systems, statistical linearization procedures are computationally cumbersome and also provide no stability or robustness guarantees. A method which has proven extremely powerful, both theoretically and experimentally, is “sliding control.” This technique is a form of input/output linearization that directly incorporates model error information with stability and performance measures. My students and I found several difficulties in the direct application of this method to automotive and aircraft control. This paper describes our solutions to the problems of repeated model differentiation, differentiation of model error, undesirable “internal dynamics” and systems with saturating control inputs.