A Machine Learning Approach to Protect Electronic Devices from Damage Using the Concept of Outlier

Most of the appliances that we used in our everyday life are electronic devices, i.e. TV, Air Conditioner, Refrigerator, etc. Excessive voltage, current, temperature, etc. can harm the devices and in extreme cases, the devices can be completely damaged. We proposed a system to monitor the electrical behaviors of the devices in real-time. The system is trained with an unlabeled dataset and capable of identifying outliers. We have used a clustering technique, i.e. Density-Based Spatial Clustering of Applications with Noise (DBSCAN) to learn the label of a dataset and then apply machine learning algorithms, i.e. Support Vector Machine (SVM) and Decision Tree to predict the label of the new data. From the prediction, the system determines whether the device is operating in safe mode or not. In this work, we have achieved the accuracy of 98.61% in detecting outliers using SVM with `rbf' kernel. Hence, if the device operates beyond safe mode, we can shut down the device.