The Application of Machine Learning and Deep Learning Techniques for Event Classification in Power Systems

Maintaining stability and providing an uninterrupted supply of customers’ demands are crucial aspects of power systems. With the expansion of electrical networks, the complexity of the network increases, and as a result, the possibility of faults and disturbances in the network also increases. The behavior and hardness of disturbances vary depending on different operating conditions within the power system. In addition, the severity of each contingency depends on its location, type, and duration. To minimize economic losses and blackouts, monitoring resource adequacy requires rapid and accurate detection and classification of the event and its location. Following that, based on the type and location of the event, corresponding alarm signals will be provided to the system operators so that they have a comprehensive sense and awareness of the network condition and may take suitable actions and control instructions in response to the disturbance. Recent developments in intelligent algorithms have paved the way for data-driven techniques in the power system. In addition, the widespread use of PMUs in power systems, which capture synchronized phasors of voltage, current, frequency, and ROCOF signals in real time, has made disturbance detection considerably easier. This chapter explores various categories of events within power systems and the necessity of investigating them.