Abstract
Nowadays, human operators at control centers analyze a large volume of alarm information during outage events and must act fast to restore the service. To assist operator decisions this work proposes novel machine learning-based functions aiming to: (a) classify the complexity of a fault occurrence (Occurrences Classifier) and its cause (Fault Cause Classifier) based on its alarm events; (b) provide fast insights to the operator on how to solve it (Data2Actions). The Occurrences Classifier takes alarm information of an occurrence and classifies it as a simpleor complexoccurrence, while the Fault Cause Classifier predicts the cause class of MV lines faults. The Data2Actions takes a sequence of alarm information from the occurrence and suggests a more adequate sequence of switching actions to isolate the fault section. These algorithms were tested on real data from a Distribution System Operator and showed: (a) an accuracy of 86% for the Data2Actions, (b) an accuracy of 68% for the Occurrences Classifier, and (c) an accuracy of 74% for the Fault Cause Classifier. It also proposes a new representation for SCADA event log data using graphs, which can help human operators identify infrequent alarm events or create new features to improve model performance.

Abstract
The present paper presents the implementation of next-generation centralized Protection, Automation, and Control (PAC) solution for Medium Voltage (MV) power grids, developed in the scope of the SCALE project [1]. The main goals of the project are the development, testing, and field pilot deployment of an innovative, fully digital PAC system for Substation Automation (SAS), centralizing in a single device the functionalities of several bay-level Intelligent Electronic Devices (IED). The envisioned system, comprised of a Centralized Protection and Control (CPC) device and Merging Units (MU)/Process Interface Units (PIU), constitutes a highly flexible, resilient, future-proof solution that relies both on modern IEC 61850 standards and on legacy industrial protocols to guarantee multi-vendor interoperability and continued integration with multi-generation devices inside and outside of the substation. Centralizing SAS functionalities in a single device provides access to a wide range of data and measurements that unlocks technologically advanced substation-centric network automation applications. © The Institution of Engineering and Technology 2023.

Abstract
In the power system decarbonization roadmap, novel grid management tools and market mechanisms are fundamental to solving technical problems concerning renewable energy forecast uncertainty. This work proposes a predictive algorithm for procurement of grid flexibility by the system operator (SO), which combines the SO flexible assets with active and reactive power short-term flexibility markets. The goal is to reduce the cognitive load of the human operator when analyzing multiple flexibility options and trajectories for the forecasted load/RES and create a human-in-the-loop approach for balancing risk, stakes, and cost. This work also formulates the decision problem into several steps where the operator must decide to book flexibility now or wait for the next forecast update (time-to-decide method), considering that flexibility (availability) price may increase with a lower notification time. Numerical results obtained for a public MV grid (Oberrhein) show that the time-to-decide method improves up to 22% a performance indicator related to a cost-loss matrix, compared to the option of booking the flexibility now at a lower price and without waiting for a forecast update.

Abstract
This paper explores the application of Deep Learning techniques to forecast electricity market prices. Three Deep Learning (DL) techniques are tested: Dense Neural Networks (DNN), Long Short-Term Memory Networks (LSTM) and Convolutional Neural Networks (CNN); and two non-DL techniques: Multiple Linear Regression and Gradient Boosting (GB). First, this work compares the forecast skill of all techniques for electricity price forecasting. The results analysis showed that CNN consistently remained among the best performers when predicting the most unusual periods such as the Covid19 pandemic one. The second study evaluates the potential application of CNN for automatic feature extraction over a dataset composed by multiple explanatory variables of different types, overcoming part of the feature selection challenges. The results showed that CNNs can be used to reduce the need for a variable selection phase.

Abstract
Nowadays, human operators at grid control centers analyze a large volume of alarm information during outage’s events, and must act fast to restore the service. Currently, after the occurrence of short-circuit faults and its isolation via feeder protection, fault location and isolation is achieved via remotely controlled switching actions defined by operator’s experience. Despite operator’s experience and knowledge, this makes the process sub-optimal and slower. This paper proposes two novel machine learning-based algorithms to assist human operator decisions, aiming to: i) classify the complexity of a fault occurrence (Occurrences Classifier) based on its alarm events; ii) provide fast insights to the operator on how to solve it (Data2Actions). The Occurrences Classifier takes the alarm information of an occurrence and classifies it as a “simple” or “complex” occurrence. The Data2Actions takes a sequence of alarm information from the occurrence and suggests to the operator the more adequate sequence of switching actions to isolate the fault section on the overhead medium voltage line. Both algorithms were tested in real data from a Distribution System Operator between 2017 and 2020, and showed i) an accuracy of 86% for the Data2Actions, and ii) the Occurrences Classifier reached 74% accuracy for “simple” occurrences and 58% for “complex” ones, leading to an overall 65% accuracy. © 2022 IET Conference Proceedings. All rights reserved.

Abstract