Data Mining: Concepts and Techniques
Data Mining: Concepts and Techniques
Watts in the basket?: Energy Analysis of a Retail Chain
Proceedings of the 5th ACM Workshop on Embedded Systems For Energy-Efficient Buildings
| Hi-index | 0.00 |
Energy consumption and room temperature can be simulated using EnergyPlus model, given the building model, weather information, and HVAC control strategy. For a given weather forecast, however, online simulation-based optimization of the HVAC control strategy is time-consuming and thus impractical. In this paper we suggest to offline learn a compact candidate list of demand response (DR) HVAC control strategies that best cover potential weather variations. Training weather data is first clustered using a specially-designed hybrid method, and the resulting clusters are then used to supervise the sampling on the training weather data. The candidate list is formed as the union of all the optimal HVAC control strategies corresponding to the selected weather patterns. The proposed method is tested on real historical data (from Berkley and Phoenix), and is able to provide a strategy list that covers 95--100% weathers using only 8% of the computational cost of a brute-force search. In addition, the fitness function can be optimized up to 99.9% of the global optimum.