A continually online trained impedance estimation algorithm for transmission line distance protection tolerant to system frequency deviation

- Proposal of a new method for impedance estimation based on Artificial Neural Networks.
- The method is designed to work in off-nominal frequency situations.
- The method is simulated and compared with the well-known Fourier algorithm.
- Results are presented in the form of graphs to show the performance of the proposed method.

Distance relays are protective devices which main goal is the protection of transmission lines. However, the presence of harmonics and the exponentially decaying DC offset in the system voltage and current signals negatively affects the relay performance. In this paper, an adaptive phasor estimation method based on Artificial Neural Networks is presented to reduce these components' effects, focusing on impedance estimation for distance relaying. The method uses the multilayer perceptron architecture to estimate the current and voltage signals, and then proceeds to calculate the complex apparent impedance during a continually online training process. This online training allows for adaptability regarding the system frequency, providing tolerance against its deviation. Graphical results of the test cases are presented, comparing the functionality and performance of the proposed algorithm with a Fourier-based method.