Ensuring the stability of frequency in power systems is paramount to their reliable operation. However, in the unfortunate event of an outage, the delicate balance of frequency can be disrupted, leading to a downward spiral due to power mismatches. This phenomenon, known as under-frequency load shedding (UFLS), becomes crucial in maintaining system stability. The complexity of calculating UFLS lies in its step-wise activation scheme and the intricate dynamics of frequency response, typically modeled by differential equations. Given the intricacy involved, data-driven methods have emerged as promising solutions to tackle such challenging problems. In line with this, our proposal aims to harness the power of deep neural networks to effectively estimate the extent of UFLS following an outage. Our focus is on the La Palma power system, where we intend to leverage the wealth of data available to train and fine-tune the neural network for accurate predictions. No prior knowledge of power systems is required for this project. I will provide all the necessary datasets, already labeled, to facilitate the training of neural networks. The main task for the student will be to train a deep neural network model capable of accurately estimating the amount of UFLS.
