Optimizing Control Strategy using Statistical Model Checking

This paper proposes a new efficient approach to optimize energy consumption for energy aware buildings. Our approach relies on stochastic hybrid automata for representing energy aware systems. The model is parameterized by several cost values that need to be optimized in order to minimize energy consumption. Our approach exploits a stochastic semantic together with simulation in order to estimate the best value for such parameters. Contrary to existing techniques that would estimate energy consumption for each value of the parameters, our approach relies on a new statistical engine that exploits ANOVA, a technique that can reduce the number of runs needed by the comparison algorithm to perform the estimates. Our approach has been implemented and our experiments show that we clearly outperform the naive approach.