International Journal of Science, Engineering and TechnologyAuthors: Assistant Professor Sachin Suke, Astha Gawande, Sonali Nandanwar, Kartik Tipatre, Sushant Akre, Riddhi Dhawangale
Abstract: Electrical faults such as arc faults, overcurrent, short circuits and abnormal electrical conditions can cause serious damage to electrical equipment and may lead to electrical fires. Conventional protection devices mainly operate when electrical parameters cross predefined threshold values, which may not be sufficient for detecting certain intermittent and unpredictable faults at an early stage. The proposed AI Based Smart Electrical Protection System Aims to provide intelligent and real-time monitoring of electrical conditions for early fault detection and protection. The system continuously monitors electrical parameters and analyes their variations to distinguish between normal operating conditions and abnormal fault conditions. An Artificial Intelligence-based fault detection approach is used to identify abnormal patterns and improve the reliability of protection. When a potentially dangerous fault is detected, the controller generates a tripping signal to operate a relay and disconnect the electrical supply, thereby reducing the possibility of equipment damage and electrical fire. An alarm indication can also be provided to alert the user about the detected fault. The proposed system is inspired by the reference study, in which current variations and high-frequency signal energy were used as important features for arc-fault identification using a weighted least-squares support vector machine, followed by relay-based circuit interruption. The reference system demonstrated real-time fault detection and interruption of the supply before an electrical fire could occur. The proposed AI-based system extends this concept toward a smart electrical protection framework capable of intelligent fault identification, rapid isolation and improved electrical safety..
