Intrusion detection and mitigation of attacks in microgrid using enhanced deep belief network

The convergence from the electric grid to the smart microgrid motivates the incorporation of the intrusion detection system to identify intruders and mitigate the resultant damages to ensure system stability. It is planned to employ the Deep Belief Network (DBN), which is one of the deep learning techniques with some improvement to detect the attacks in a microgrid. To improve the accuracy of the detection, a rule-based detection technique is added to enhance the detection of intruders using DBN. The proposed technique is supported with the layered micro-grid architecture that makes the system flexible and simple toward the implementation. The proposed Enhanced DBN (EDBN) performance is measured in different bus representations for identifying the higher hit rate and rejection rate, lesser miss rate and false-positive rate. Two attacks, such as False Data Injection and Denial of Service attacks, are generated by Greedy Algorithm and are detected by the proposed technique. Compared to the existing detection and control system, the proposed EDBN technique provides accuracy higher than 92%, false alarm rate less than 1%. Thus, the experimental results show that the proposed technique accuracy is higher than the existing intrusion detection techniques in a microgrid.