Driving Range Estimation and Energy Consumption Rate Deviation Classification in Electric Vehicles using Machine Learning Methods

Description

Using a dataset collected from https://spritmonitor.de/ driving range for electric vehicles is predicted via input features, such as driving_style, avg_speed and route_type.

• Regressors:

1. Linear Regression
2. Multilayer Perceptron (MLP)
3. Random Forest
4. AdaBoost
5. Deep Multilayer Perceptron (Deep MLP)

• Classifiers:

1. Support Vector Machines (SVM)
2. Multilayer Perceptron (MLP)
3. Random Forest
4. Deep Multilayer Perceptron (Deep MLP)

Citation

Find the related published conference paper here.

@inproceedings{amirkhani2019electric,
  title={Electric Vehicles Driving Range and Energy Consumption Investigation: A Comparative 
  Study of Machine Learning Techniques},
  author={Amirkhani, Abdollah and Haghanifar, Arman and Mosavi, Mohammad R},
  booktitle={2019 5th Iranian Conference on Signal Processing and Intelligent Systems (ICSPIS)},
  pages={1--6},
  year={2019},
  organization={IEEE}
}

Input data

Dataset crawler (vehicle_crawler.py) and an example result (volkswagen_e_golf.csv) in csv file can be found here: https://github.com/armiro/crawlers/tree/master/SpritMonitor-Crawler

Run the code

First, change the dataset path in both files. Then,

• run the driving_range_prediction.py file to predict the trip distance of the electric vehicle; how long this vehicle can go in the next trip.
• run the ECR_deviation_classification.py file to classify the ECR deviation from the manufacturer; whether in this trip ECR is higher (more consumption) or lower (less consumption) than the factory-defined ECR.