Benchmarks for stochastic models from building automation systems

Abstract

Abstract Benchmarks Proposal: We provide benchmarks for stochastic models drawn from Building Automation Systems (BAS), specifically constructed from expertise developed on a real BAS setup. This contribution branches out of the library of general models presented in [4], specifically focussing on probabilistic models. Using this library, we generate two realistic case studies which incorporate (i) stochasticity stemming from different sources (e.g. process or observation noise on the continuous variables) and (ii) various input and output signals. We describe each model structure (syntax and semantics), identify key problems (specifications) for different analysis goals, and finally illustrate solutions for each goal.