Simulation-based sensor allocation for dynamic environment estimation in cyber-physical building system

Building environment is considered as a complex dynamics system with high uncertainty due to weather changes and human activities. Traditional sensor allocation methods are hard to generate the strategy to improve the observation performance of environmental information in buildings. But the simulation-based estimate is usually time-consuming and noisy. This paper presents a sensor allocation strategy based on the theory of ordinal optimization that the ordinal comparisons of performance measures are robust with respect to noise and modeling error. The basic idea is to use an approximate model that describes the temperature and humidity dynamics of the building. Nominal N allocations are obtained by uniform sampling with given numbers of the sensor. The ordinal optimization method is applied to isolate a good enough set S that contains some good allocations with high probability by performing rough evaluation through a neural network which is trained by clustered historical data. The best allocation is then selected by solving a model-driven building simulation for each of the allocations in S. Using this simulation-based method we are able to obtain a good enough sensor allocation strategy with reasonable computational effort. A case study is given based on a real building to demonstrate the proposed sensor location optimization method.