A Non-intrusive IoT System for the Detection of Faults in Internal Combustion Engines

Abstract

The purpose of this work is to develop a non-intrusive low-cost system capable of detecting faults and register working hours of Internal Combustion (IC) engines. The system should allow the monitoring of combustion engines usage, detect faults remotely, and is focused on IC engines for maritime applications. It consists of two parts, as explained in the following. The first part involves the development of the device, which should be capable of performing a continuous monitoring of an engine’s activity and register its working periods. This decision is made based on the vibration analysis. As a requirement, the developed device must be low-cost, portable and have a low energy consumption to provide at least two years of autonomy. The second part involves fault’s detection and, like the first part, it requires information from the engine’s vibrations. A low power accelerometer will be used to monitor the engine’s activity, and a piezoelectric polyvinylidene fluoride film is used to record the engine’s vibration, for fault detection. The data is then sent via Bluetooth to the user’s smartphone, which forwards it to the Internet, for further analysis. Bluetooth Low Energy is used to support the required autonomy. The developed technique for fault detection includes data acquisition, analog and digital pre-processing and digital processing. The most important part of the digital processing, in this case, is a Time-Frequency Analysis. Several Time-Frequency Analysis algorithms were tested with real data to select the best solution to detect faults in such engines.