Predicting Prior Engine Failure with Classification Algorithms and web-based IoT Sensors

Machine learning classification techniques play a significant role in engine failure issues and machinery maintenance. With the help of Internet of Things, IoT industry, connected sensors have a considerable impact on data collection and remote engine monitoring. Mechanical engineers and professionals have difficulties determining when an engine is going to have a malfunction. So, engine maintenance requires an adequate strategy to predict the closest time in which an incident would likely to occur. This research investigates a perfect solution so that engineers will have an earlier alert about the potential incident which might exist. This study gives a visualized time left for how long an engine lifetime is present, accordingly, the system notifies the engineers of the best time to implement the maintenance. The methodology that we follow is setting up an appropriate mechanism by collecting data with IoT, and analyzing such data with classification algorithms. These algorithms categorize the status of an engine into particular conditions, so they indicate how far an engine going to work in an optimal state. Experiments have proved that K-Near Neighbor is the best algorithm for this kind of work in between others like; decision tree and linear discriminant with accuracy 82.9%, 51.0%, and 64.9% respectively. Consequently, classification techniques confidently distinguish the engine condition and warning for necessity of maintenance at the right time and right status.