arXiv:2512.04690v1 Announce Type: new 
Abstract: We present a novel recurrent neural network architecture designed explicitly for day-ahead electricity price forecasting, aimed at improving short-term decision-making and operational management in energy systems. Our combined forecasting model embeds linear structures, such as expert models and Kalman filters, into recurrent networks, enabling efficient computation and enhanced interpretability. The design leverages the strengths of both linear and non-linear model structures, allowing it to capture all relevant stylised price characteristics in power markets, including calendar and autoregressive effects, as well as influences from load, renewable energy, and related fuel and carbon markets. For empirical testing, we use hourly data from the largest European electricity market spanning 2018 to 2025 in a comprehensive forecasting study, comparing our model against state-of-the-art approaches, particularly high-dimensional linear and neural network models. The proposed model achieves approximately 12% higher accuracy than leading benchmarks. We evaluate the contributions of the interpretable model components and conclude on the impact of combining linear and non-linear structures.

تم تقديم بنية جديدة لشبكة عصبية متكررة لتوقع أسعار الكهرباء ليوم واحد، مما يعزز اتخاذ القرار في أنظمة الطاقة. يدمج هذا النموذج هياكل خطية مثل نماذج الخبراء ومرشحات كالمان في الشبكات المتكررة، مما يحسن الكفاءة الحسابية وقابلية التفسير مع التقاط الخصائص الأساسية للأسعار في أسواق الطاقة.

Se ha presentado una nueva arquitectura de red neuronal recurrente para la previsión de precios de electricidad a un día, mejorando la toma de decisiones en los sistemas energéticos. Este modelo integra estructuras lineales como modelos de expertos y filtros de Kalman en redes recurrentes, lo que mejora la eficiencia computacional y la interpretabilidad, capturando características esenciales de precios en los mercados de energía.

Une nouvelle architecture de réseau de neurones récurrents a été introduite pour la prévision des prix de l'électricité à un jour, améliorant la prise de décision dans les systèmes énergétiques. Ce modèle intègre des structures linéaires telles que des modèles d'experts et des filtres de Kalman dans des réseaux récurrents, améliorant l'efficacité computationnelle et l'interprétabilité tout en capturant les caractéristiques essentielles des prix sur les marchés de l'électricité.

A novel recurrent neural network architecture has been introduced for day-ahead electricity price forecasting, enhancing decision-making in energy systems. This model integrates linear structures like expert models and Kalman filters into recurrent networks, improving computational efficiency and interpretability while capturing essential price characteristics in power markets.

Recurrent Neural Networks with Linear Structures for Electricity Price Forecasting

arXiv:2510.03814v3 Announce Type: replace 
Abstract: Recurrent Neural Networks (RNNs) have found widespread applications in machine learning for time series prediction and dynamical systems reconstruction, and experienced a recent renaissance with improved training algorithms and architectural designs. Understanding why and how trained RNNs produce their behavior is important for scientific and medical applications, and explainable AI more generally. An RNN's dynamical repertoire depends on the topological and geometrical properties of its state space. Stable and unstable manifolds of periodic points play a particularly important role: They dissect a dynamical system's state space into different basins of attraction, and their intersections lead to chaotic dynamics with fractal geometry. Here we introduce a novel algorithm for detecting these manifolds, with a focus on piecewise-linear RNNs (PLRNNs) employing rectified linear units (ReLUs) as their activation function. We demonstrate how the algorithm can be used to trace the boundaries between different basins of attraction, and hence to characterize multistability, a computationally important property. We further show its utility in finding so-called homoclinic points, the intersections between stable and unstable manifolds, and thus establish the existence of chaos in PLRNNs. Finally we show for an empirical example, electrophysiological recordings from a cortical neuron, how insights into the underlying dynamics could be gained through our method.

تم تقديم خوارزمية جديدة لاكتشاف الأشكال الثابتة في الشبكات العصبية المتكررة (RNN) المعتمدة على وحدات خطية م rectified (ReLU)، والتي تعتبر ضرورية لفهم السلوك الديناميكي لهذه الشبكات في تطبيقات مثل التنبؤ بالسلاسل الزمنية وإعادة بناء الأنظمة الديناميكية. يركز هذا التقدم على الشبكات العصبية المتكررة الخطية القطعية (PLRNN) وخصائصها الطوبولوجية، والتي تعتبر حاسمة لتقسيم فضاءات الحالة إلى أحواض جذب مختلفة.

Se ha introducido un nuevo algoritmo para detectar variedades invariantes en redes neuronales recurrentes (RNN) basadas en unidades lineales rectificadas (ReLU), que son esenciales para comprender el comportamiento dinámico de estas redes en aplicaciones como la predicción de series temporales y la reconstrucción de sistemas dinámicos. Este avance se centra en las RNN lineales por tramos (PLRNN) y sus propiedades topológicas, que son cruciales para descomponer los espacios de estado en diferentes cuencas de atracción.

Un nouvel algorithme a été introduit pour détecter les variétés invariantes dans les réseaux de neurones récurrents (RNN) basés sur des unités linéaires rectifiées (ReLU), qui sont essentielles pour comprendre le comportement dynamique de ces réseaux dans des applications telles que la prévision de séries temporelles et la reconstruction de systèmes dynamiques. Cette avancée se concentre sur les RNN linéaires par morceaux (PLRNN) et leurs propriétés topologiques, cruciales pour diviser les espaces d'état en différents bassins d'attraction.

A novel algorithm has been introduced for detecting invariant manifolds in ReLU-based Recurrent Neural Networks (RNNs), which are essential for understanding the dynamical behavior of these networks in applications such as time series prediction and dynamical systems reconstruction. This advancement focuses on piecewise-linear RNNs (PLRNNs) and their topological properties, which are crucial for dissecting state spaces into various basins of attraction.

Recurrent Neural Networks with Linear Structures for Electricity Price Forecasting

Was this article worth reading? Share it

LucidQuery AI

Betron.io

Rekla.ai