Robust Calibration of Industrial HVAC and Battery Systems

Explore robust calibration techniques for industrial HVAC and battery systems in this informative conference talk. Delve into the use of ModelingToolkit.jl for building large-scale models of industrial systems, leveraging its advanced symbolic manipulation techniques and acausal nature. Learn about the essential process of fine-tuning these models to align with real-world industrial systems, incorporating design constraints and experimentally measured data. Discover how to overcome challenges in model calibration, such as behavioral complexity, noise, partial observability, and sparse measurements. Examine the JuliaSim Model Library's high-performance, composable tools for industrial systems, including JuliaSim-HVAC for refrigeration and air conditioning systems, and JuliaSim-Batteries for electrochemical models of large battery packs. Investigate techniques like Single Shooting, Multiple Shooting, Collocation, and Prediction Error Method, demonstrated on HVAC and Battery systems to avoid local minima during calibration. Compare predictions against validation data sets to assess result robustness, and explore the effects of parameter unidentifiability using Bayesian priors.