Methodology for developing models to estimate vehicle instantaneous energy consumption based on hub-type dyno test data

This paper describes a methodology to develop simple energy consumption models of road vehicles exploiting transient experimental datasets obtained from a vehicle/powertrain four-dyno testbed available at the Center for Automotive Research and Sustainable mobility (CARS@POLITO) of Politecnico di Torino. These models, based on a locally weighted linear regression method, can serve as a simpler alternative to more conventional methods based, for example, on engine maps obtained by steady-state characterization at engine testbeds, and combined with powertrain subsystem models. The present methodology was applied to a conventional diesel-powered vehicle. Three different modeling approaches are proposed: vehicle-based (VB), engine-based (EB) and engine-based modified (EB*). The VB approach is the simplest, being able to estimate the vehicle fuel consumption by only using, as inputs, wheel torque and speed, while the EB and EB* approaches enhance modeling accuracy by using engine speed and torque, as inputs, along with transmission-related parameters and/or by considering the moments of inertia of the powertrain rotating parts. The manuscript describes, in full, the process used to develop these models, providing significant guidance for researchers who may want to replicate the procedure with their own experimental data. These energy consumption models can be useful tools for the development and assessment of eco-driving or ADAS functions or for energy consumption comparison between different vehicles that were not tested on the same driving cycle. They can also support the estimation of the total energy consumption of vehicles along different traffic conditions or routes, based on a limited number of experiments and low computational effort.