Browsing by Author "Manjarrés, Diana"

Now showing 1 - 2 of 2
  • No Thumbnail Available
    Item
    Online Pentane Concentration Prediction System Based on Machine Learning Techniques †
    (2023) Manjarrés, Diana; Maqueda, Erik; Landa-Torres, Itziar; IA; DIGITAL ENERGY
    Industry 4.0 has emerged together with relevant technological tools that have enabled the rise of this new industrial paradigm. One of the main employed tools is Machine Learning techniques, which allow us to extract knowledge from raw data and, therefore, devise intelligent strategies or systems to improve actual industrial processes. In this regard, this paper focuses on the development of a prediction system based on Random Forest (RF) to estimate Pentane concentration in advance. The proposed system is validated offline with more than a year of data and is also tested online in an Energy plant of the Basque Country. Validation results show acceptable outcomes for supporting the operator’s decision-making with a tool that infers Pentane concentration in Butane 400 min in advance and, therefore, the quality of the obtained product.
  • No Thumbnail Available
    Item
    SpeCluRC-NTL: Spearman's distance-based clustering Reservoir Computing solution for NTL detection in smart grids
    (2024-06) Serra, Adrià; Ortiz, Alberto; Manjarrés, Diana; Fernández, Mikel; Maqueda, Erik; Cortés, Pau Joan; Canals, Vincent; IA; DIGITAL ENERGY
    Smart grids are ushering in a transformative era for energy distribution and consumption, yet their emergence also brings forth novel security and fraud detection challenges. The intricacy of detecting fraud within smart grids demands sophisticated techniques for scrutinizing vast volumes of time series data. This work introduces a novel approach that integrates time series aggregation functions, time series clustering using Spearman's distance, and reservoir computing forecasting to effectively identify fraud within smart grid systems. Specifically, the proposed methodology employs a clustering approach based on Spearman's rank distance to summarize time series data. This enables the aggregation of similar daily patterns, providing highly descriptive power and simplifying forecasting through Reservoir Computing. The subsequent step classifies each prosumer behavior as regular or potentially fraudulent. The SpeCluRC-NTL methodology, as proposed, is designed to detect fraud almost in real-time with low operational costs. The effectiveness of our approach is confirmed through empirical findings gathered from the Parc Bit distribution grid. This grid is located near Palma (Balearic Islands), Spain. The results of our research highlight the demonstrated effectiveness of the proposed approach, revealing its promising potential as it undergoes testing at the ParcBit premises. In comparison to previous works, SpeCluRC-NTL showcases its ability to reduce the false positive rate while maintaining a high true positive ratio, resulting in an increased AUC score. This has substantial implications for mitigating financial losses and addressing the various impacts associated with fraudulent activities in smart grids.