State Forecasting

State forecasting aims to predict the future condition of complex systems, a crucial task across diverse domains like IT operations and energy grids. Current research emphasizes leveraging the inherent relational structure within these systems, employing graph neural networks and other advanced architectures like transformers to model spatio-temporal dependencies within multivariate time series data. This focus on integrating diverse data sources, such as network topology and sensor readings, improves forecasting accuracy and enables more effective anomaly detection and predictive maintenance. Improved state forecasting ultimately enhances system reliability, optimizes resource allocation, and facilitates proactive management of critical infrastructure.

Papers

August 15, 2024

System States Forecasting of Microservices with Dynamic Spatio-Temporal Data
Yifei Xu, Jingguo Ge, Haina Tang, Shuai Ding, Tong Li, Hui Li
Microservice Architecture Time Series Forecasting Method Spatio Temporal Relationship State Forecasting

April 4, 2024

Graph Neural Networks for Electric and Hydraulic Data Fusion to Enhance Short-term Forecasting of Pumped-storage Hydroelectricity
Raffael Theiler, Olga Fink
Graph Neural Network Spectral Graph Short Term Forecasting Hydropower Plant State Forecasting

State Forecasting

Papers

System States Forecasting of Microservices with Dynamic Spatio-Temporal Data

Graph Neural Networks for Electric and Hydraulic Data Fusion to Enhance Short-term Forecasting of Pumped-storage Hydroelectricity