Optimal control strategy of multiple chiller system based on background knowledge graph

The multiple chiller system is a key issue for energy saving of heating, ventilation, and air conditioning (HVAC) system. Optimal control is an effective way to improve the operation efficiency of the multiple chiller system. The optimal control strategy based on model prediction usually possesses two problems. One is that it is difficult to integrate into empirical knowledge, and another is that an optimization algorithm, which usually takes huge computation cost, should be involved. Therefore, a novel optimal control strategy for multiple chiller system based on knowledge graph is presented in this research. The historical data knowledge graph (HDKG) and the expert knowledge graph (EKG) are constructed involving the knowledge of operation performance and empirical rules respectively. The approximate groups of the optimal control sets are inferred directly from the background knowledge graph (BKG), which combined HDKG and EKG by the matrix factorization with pattern-oriented feedback (MF-POF) method, and the optimal control sets are finally obtained by the identifying mechanism with evaluation of the plausibility and perceptibility. The proposed strategy is validated by simulation with the operation data from a multiple chiller system of a manufactory building. The results show that it consumes 12.32% less energy than that of the original strategy of field control. Compared with other optimal strategies, which are based on the prediction of data-driven model, the proposed strategy can improve the coefficient of performance (COP) of the system significantly.