1. |
- Carpatorea, Iulian, 1982-, et al.
(author)
-
APPES Maps as Tools for Quantifying Performance of Truck Drivers
- 2014
-
In: Proceedings of the 2014 International Conference on Data Mining, DMIN'14. - USA : CSREA Press. - 9781601323132 ; , s. 10-16
-
Conference paper (peer-reviewed)abstract
- Understanding and quantifying drivers’ influence on fuel consumption is an important and challenging problem. A number of commonly used approaches are based on collection of Accelerator Pedal Position - Engine Speed (APPES) maps. Up until now, however, most publicly available results are based on limited amounts of data collected in experiments performed under well-controlled conditions. Before APPES maps can be considered a reliable solution, there is a need to evaluate the usefulness of those models on a larger and more representative data.In this paper we present analysis of APPES maps that were collected, under actual operating conditions, on more than 1200 trips performed by a fleet of 5 Volvo trucks owned by a commercial transporter in Europe. We use Gaussian Mixture Models to identify areas of those maps that correspond to different types of driver behaviour, and investigate how the parameters of those models relate to variables of interest such as vehicle weight or fuel consumption.
|
|
2. |
- Carpatorea, Iulian, 1982-, et al.
(author)
-
Towards Data Driven Method for Quantifying Performance of Truck Drivers
- 2014
-
In: The SAIS Workshop 2014 Proceedings. - : Swedish Artificial Intelligence Society (SAIS). ; , s. 133-142
-
Conference paper (peer-reviewed)abstract
- Understanding factors that influence fuel consumption is a very important task both for the OEMs in the automotive industry and for their customers. There is a lot of knowledge already available concerning this topic, but it is poorly organized and often more anecdotal than rigorously verified. Nowadays, however, rich datasets from actual vehicle usage are available and a data-mining approach can be used to not only validate earlier hypotheses, but also to discover unexpected influencing factors.In this paper we particularly focus on analyzing how behavior of drivers affects fuel consumption. To this end we introduce a concept of “Base Value”, a number that incorporates many constant, unmeasured factors. We show our initial results on how it allows us to categorize driver’s performance more accurately than previously used methods. We present a detailed analysis of 32 trips by Volvo trucks that we have selected from a larger database. Those trips have a large overlap in the route traveled, of over 100 km, and at the same time exhibit different driver and fuel consumption characteristics.
|
|