SafeEngine: Fault Detection with Severity Prediction for Diesel Engine

Diesel Engine plays a vital role in the working of various machinery, and the disruption in working of such machines due to some faults leads to a large number of losses. The time and expense to fix those faulty machines are also considerable. To avoid the chances of fault, we can use Predictive Health Maintenance to monitor the health status of devices and act before the fault occurs. In this paper, we have proposed a data-driven model to classify the type and severity of faults that occur in a diesel engine in a real-time scenario. In our proposed system, vibration sensors are used, and the data is sent to the model. The spectrum analysis of the received signal is done before classifying the signal into faults using the Random forest algorithm. The model has very high accuracy of 99.7% for the classification of the different fault and their severity. It will also detect whether multiple faults are present or not. The model will alert for any significant defects in the system, which can help avoid the disruption of work during production.