Truck drivers often steal fuel from trucks, without their employer companies being able to determine for sure if there was stealing.
The project aims to solve the problem by monitoring data from various sensors around the truck. By analyzing this data we can learn the behavior of each individual truck and determine various events including frauds.
The aim of the project is to build a program that can detect fuel fraud based on various sensors reading in trucks, without the need of a physical device to be implemented in the truck.
The first part of the project was to analyze the data of trucks using Matlab, and build a basic flow that can achieve the wanted goal.
The Second part is to build a program (using java) that runs on servers of Tactile Mobility company (the company than offered the project).
The program scans the incoming data from the traveling trucks, analyze this data and save calibration data used to determine fraud cases in future checks.
Fraud detecting occurs by scanning the travels and calculating the approximate fuel consumption in a travel and comparing this approximation to the real engine consumption taken from the truck, and thus determines if there were any frauds.
