🏆 People’s Choice In-person | 🏆 Gangnam Stylish
The objective of this project is to develop a method for evaluating the optimal vehicle configuration for fuel efficiency in a given customer application, using available historical data.
Sponsor
Volvo Group Trucks
Team Members
Xinru Yang Kaiming Cui Chaewan Chun Nada Alaskary
Project Poster
Click on any image to enlarge.
Project Video
Project Summary
Overview
A task had been given by Pennsylvania State University (PSU) and Volvo Group Trucks to propose a design and application to help the customer pick the best optimal truck configuration that will fulfill their needs. This project targets the optimization of fuel efficiency in Heavy-Duty trucks using a large dataset provided from Volvo Group Trucks. Minor improvements in fuel efficiency for customers operating large truck fleets that accumulate millions of miles per year lead to significant operating cost reductions. Therefore, a tool that helps customers and dealers find the best optimal configuration based on their needs is required. An Android mobile application, with the support of a data processing program, was constructed to reach the goal.
Objectives
– Our objective is to examine the relationship between customer application, vehicle configuration, and fuel efficiency and develop a method and easy-to-use tool for evaluating the optimal vehicle configuration for customers buying current products using available historical data.
Approach
– Weekly meeting with our sponsor Volvo Group Trucks team to discuss our project and get their feedback.
– Cleaned and analyzed the given dataset of customers’ truck information.
– Filtered the given dataset with matching values based on the input from the customer.
– Ranked the filtered data based on the fuel efficiency of trucks and generated the top 3 trucks with the highest fuel efficiency.
– Chose a machine learning model that fits our dataset: XGBoost (Gradient Boosting.)
– Trained the machine learning model to learn the relationship between fuel efficiency and fuel efficiency-related truck configurations.
– Generated importance feature plot based on the result from the machine learning model, representing which truck configuration affects the fuel efficiency the most.
– Build an application using the Flutter framework as an easy-to-use user interface for customers and dealers.
– Build an API with Flask to receive data from the application and run the recommendation algorithm.
Outcomes
– Volvo Group Trucks will receive codes for a working Android application and a data processing algorithm.
– The Android application significantly reduced the time of manually selecting truck configurations for customers out of 100,000+ trucks with 100+ configurations, hence outputting three recommendations with the optimal vehicle configurations for fuel efficiency.