Center of Inertia Frequency Estimation Using Deep Learning Algorithm

Abstract

Abstract Increasing the number of generation units connected to the grid via power electronic devices potentially implies negative impacts on the power system frequency stability and, depending on the power system inertia value, implies the necessary contribution of wind power plants to inertial response of the system. An alternative approach to the active power control of wind power plants, without the impact of local frequency deviation on the output power, is the application of a control strategies based on the center of inertia frequency. Since control schemes based on the input variable of the center of inertia frequency require a satisfactory level of signal transmission capacity in real time and the advanced telecommunication infrastructure of the power system, the paper considers an alternative approach to estimate the input signal value. According to the developed long short-term memory recurrent neural network, paper presents the idea of center of inertia frequency estimation by monitoring the speed of several generators in the system and passing the sequence of input data for a certain time interval, after the occurrence of imbalance, to the artificial intelligence module.