Fuel Economy and Emission Testing for Connected and Automated Vehicles Using Real-world Driving Datasets

A new method was proposed to evaluate the fuel economy and emission level of the vehicle based on the unsupervised learning of the real-world driving data of the evaluated vehicle and typical driving primitive analysis of the naturalistic driving dataset of a large number of different vehicles. The results show that this method can successfully identify the key driving primitives, patterns, and parameters of the vehicle speed and acceleration, and couple the driving primitives from the evaluated vehicle with typical driving primitives from the large real-world driving dataset, which could enhance the evaluation method and standard of fuel economy and emission for CAV For more information see our paper "Fuel Economy and Emission Testing for Connected and Automated Vehicles Using Real-world Driving Datasets"

Getting Started

These instructions will get you a copy of the project up and running on your machine for development and testing purposes.

Prerequisites

these packages are needed:

numpy
matplotlib
os
scipy
csv
copy
time
sqlalchemy
sklearn
pyhsmm

For training purpose, we suggest that runing the code on Google Cloud Computing or other servers. We also used Jupyter Notebook for debuging and these tools all are highly recommented.

Installing

Install python packages:

For most packages you can use this easy way to install

pip install packagename

But installing package pyhsmm is a little bit triky. pyhsmm github repository

An usefull summaries as follows: 1.download pyhsmm from pyhsmm github repository](https://github.com/mattjj/pyhsmm.git)
2. install "future" and "pybasicbayes" first. Because we found that it won't work without package "future" and "pybasicbayes".

pip install future
pip install pybasicbayes

3.change to the pyhsmm-master directory and install setup.py

cd /…/pyhsmm-master
python setup.py install

Congradulations! You have installed all the packages on your machine.

Running the code

Basically, there are six parts of the code:
1)data query
2)posteriorl model
3)real data process
4)statis and results
5)constrained K means clustering
6)KL divergence
How the code arranged and worked, see comments in the ipynb files.

Data Query

Query data from server of UMICH TRI. If this sever are not avalibel for you, you can use the data published by federal government.

http://

Posterial Model

trained data by using pyhsmm package. For more introduction about how to use pyhsmm. See:

https://github.com/mattjj/pyhsmm

Real Data Process

data with fuel consumption for the validate vehicle

Statis and Results

Extracted the training results from the posteriol model

constrained K means clustering

Combined with package "conclust" in R to get the clusters with background knowledge.

KL divergence

Calculated the cluster distribution distance between valicate vehicle and big data models.

Authors

• Yan Chang - Yan Chang
• Weiqing Yang - Weiqing Yang
• Ding Zhao

See also the list of contributors who participated in this project.

License

This project is licensed under the MIT License - see the LICENSE.md file for details