Abstract

This thesis investigates a method to save energy and thus also extend the range of a series-production battery electric vehicle by influencing the driving style automatically with the help of a of a cruise controller. An exploration of existing methods shows that the contextual consideration of the current and upcoming driving situation is necessary to realise safe and energy-efficient driving. This limits the appropriate approaches to online methods using updated predictions of the vehicle behaviour. It turns out that the most suitable method for the intended purpose is model-predictive control (MPC). The MPC generates controls for the accelerator pedal of the vehicle based on optimised predictions of the vehicle motion and energy consumption subject to the current and future road slope, curvature, speed limits and distance to an eventually preceding vehicle. The non-linear nature of the vehicle dynamics generally necessitates the use of a non-linear prediction model and solving a non-linear optimisation which goes along with difficulties in the online real-time implementation. However in this work - by exploiting and extending the tool sets of classical MPC - a controller based on a quadratic optimal control problem with linear constraints can be formulated that approximates the nonlinearities of the plant dynamics with equivalent accuracy as a non-linear formulation. A linear prediction model of the vehicle motion is derived by a change of the model domain from time to position and a change of variables to predict the kinetic energy of the moving vehicle instead of the driving speed. Further, a convex piece-wise linear energy consumption model is included in the inequality constraints of the problem according to the methodology of separable programming to capture the consumption characteristics of the vehicle in different operating points. In this form, real-time capability and the energy-saving potential of the presented control approach can be demonstrated by simulations of the closed loop and by implementing the controller for driving experiments. A Smart ED series-production battery electric vehicle is chosen for the practical tests and all models and parameters are identified and adapted to the characteristics of the car. In this application case, a significant energy-saving potential could be demonstrated compared to human drivers. To further reduce the computational burden and speed up the computation, the so-called move blocking method for input parameterisation of the MPC control trajectory is investigated and extended within this work to a flexible move blocking approach which enables a fast computation and at the same time high tracking performance.

Abstract

International audience; This paper presents a novel energy-efficient model-predictive cruise control formulation for electric vehicles. The controller and the underlying dynamic model are designed to meet the properties of a series-production electric vehicle whose characteristics are identified by measurements. A predictive eco-cruise controller involves the minimisation of a compromise between terms related to driving speed and energy consumption which are in general both described by nonlinear differential equations. In this work, a coordinate transformation is used which leads to a linear differential motion equation without loss of information. The energy consumption map is approximated by the maximum of a set of linear functions which is implicitly determined in the optimisation problem. The reformulations finally lead to a model-predictive control approach with quadratic cost function, linear prediction model and linear constraints that corresponds to a piecewise linear system behaviour and allows a fast real-time implementation with guaranteed convergence. Simulation results of the MPC controller and the simulation model in closed-loop operation finally provide a proof of concept.

Abstract

International audience; A nonlinear problem formulation of an energy saving model-predictive eco-cruise controller for electric vehicles is presented. With regard to the intended application in real world tests, the model has to include the specific properties of a serial electric vehicle such as energy-recovery and a discontinuous accelerator input giving rise to a binary control variable. These specific features and the nonlinearity of the system dynamics make it a challenging task to formulate the optimisation problem in a way that allows a fast computation in real-time application. The challenges are addressed by using a model of the vehicle dynamics that is formulated in terms of the vehicle position instead of time and by considering the kinetic energy instead of the velocity. Furthermore, various constraints on the input and state variables are introduced for a realistic representation of the vehicle characteristics. A special focus is put on the treatment of a binary input variable in the optimisation. Here, in order to avoid a mixed-integer formulation of the problem, a continuous variable is introduced which is forced to take only discrete values by a penalty term. Finally, first simulation results underline the feasibility of this control approach.