EmuCast: A python package for generating time series forecasts with tunable error in model predictive control design and testing
Abstract
EmuCast is a lightweight tool for generating synthetic time-series forecasts with tuneable error levels. It is intended for researchers and engineers to test predictive control strategies without needing forecasting expertise. It uses Markov Chain Monte Carlo (MCMC) and a reshaping process to create realistic hourly to sub-hourly forecasts. It is simple with only two main parameters, and do not require any calibration to adapt to any type of data (e.g. electricity demand and generation). It is fast and realistic with errors naturally increasing along the horizon. The package includes examples and datasets for predictive management in the energy sector.
Fields of Study & Subjects
Advanced Control Systems Optimization
Model Reduction and Neural Networks
Control Systems and Identification
Publication details
- Journal / source
- SoftwareX
- Publisher
- Elsevier BV
- Volume
- 34
- Pages
- 102574–102574
- Publication date
- 28 February 2026
- Language
- EN