Energy Efficient Longitudinal Control
Licentiate thesis, 2018

Vehicles are contributing to global and local environmental problems as a result of fossil fuels. A majority of the combustion engine population is driven by fossil fuels and electrified vehicles are also to a large extent dependent on electricity production from fossil fuels. Emission legislation and standardized test methods have lead the development of technology for the automotive industry. Increased efficiency, improved combustion control and aftertreatment systems have created cleaner and more fuel efficient drivetrains. Authorities and publications have highlighted an increased gap between in-use and certified vehicle consumption and emissions because of the test-cycles current design. In order to address these differences authorities have conducted changes within the test methods from 2017 and forward and a new test-cycle WLTP is introduced including real-driving-emission test procedures. Decreasing the gap of real driving emissions and consumption can also be improved outside the legislative test-cycles using forward looking sensors, map data and statistical models.

The work considers controlling the drivetrain actuators more efficiently in a vehicle with predictive information. For this, dynamic programming is used to optimize engine speed trajectories during depletion mode for a series hybrid drivetrain. The result shows that choice of state and control signals has a direct impact on the engine speed trajectory and thereby the fuel consumption. Up to 21 % lower fuel consumption could be achieved for a series hybrid drivetrain compared to a rule based engine speed demand controller (along the best efficiency line) for the drivecycle analyzed. For a parallel hybrid drivetrain a DP method was compared to a heuristic strategy in order to determine the optimal discharge rate of the battery. In the simulation study done the DP method provided the best fuel consumption results. During evaluation of the physical tests the pre-optimized DP parameter set performed worse than the heuristic strategy. In the rig tests a fuel consumption reduction of 8 % was measured with the heuristic method, compared to a non predictive controller strategy. The DP algorithm provided 4 % reduction of fuel compared to a non predictive controller.

The work has also considered different modeling methods of a high voltage battery from recorded fleet data. One individual vehicle recorded battery pack current and voltage for one year. The recorded data was used to identify battery parameters for electric equivalent circuits. The measured current was used to calculate a reference voltage from the circuit equivalent parameters that was compared to the measured voltage. The best result was obtained for a single RC circuit model which obtained the highest average goodness of fit in voltage for the entire training data set.

Autonomous Vehicles

Optimal Control

Hybrid drivetrain control

Combustion Engines

Maskinhuset Delta
Opponent: Jonas Mårtensson KTH


Rickard Arvidsson

Chalmers, Mechanics and Maritime Sciences (M2), Combustion and Propulsion Systems

Comparing Dynamic Programming Optimal Control Strategies for a Series Hybrid Drivetrain

SAE Technical Papers,; Vol. 2017-October(2017)

Journal article

An Evaluation of Discharge Strategies for Plug-In Hybrid Electric Vehicles

25th Aachen Colloquium,; (2016)

Other conference contribution

Battery parameter estimation from recorded fleet data

SAE Technical Papers,; Vol. 2016-Octobeer(2016)

Journal article

Driving Forces

Sustainable development

Areas of Advance



Subject Categories

Vehicle Engineering

Control Engineering

Other Electrical Engineering, Electronic Engineering, Information Engineering



Maskinhuset Delta

Opponent: Jonas Mårtensson KTH

More information

Latest update

3/6/2018 8