An operational parameter optimization method based on association rules mining for chiller plant

Abstract The traditional mechanism models of a chiller plant of HVAC are complicated with multiple variables and many constraints, so that it’s burdensome to optimize those operational parameters. Moreover, the optimization methods based on mechanism models are unpractical to be applied in the engineering projects. Therefore, an operational parameter optimization method based on the unsupervised data mining technology is proposed in this paper and verified with a large number of field operational data of the chiller plant of a shopping mall in the subtropical area. The unsupervised data mining procedure is illustrated in detail, including data preparation, data partitioning, strong association rules extraction by Apriori algorithm and so on. The definition, selection and discretization methods of external and operational parameters are described to determine the mining target and divide typical operating conditions. At last, 54 and 70 strong association rules, respectively, for the single larger chiller operating mode and the single smaller one under typical operating conditions are extracted. Simulation results shows that the energy consumption of the chiller plant is reduced by 11.60% and 13.33% after optimization, respectively, on the studied days in summer and that in winter. The analysis results mean that after optimization, the energy performance of a chiller plant was significantly improved. The strong association rules are easily utilized in the engineering projects, in terms of their simplicity and feasibility. And this method can also be used in other fields when there are enough effective operational data.