Abstract: For evaluating the performance of system by using quantitative model-based method, the fractal correlation dimension technique is proposed to take the place of conventional direct-comparison method. The supply air temperature in air handling unit (AHU) is taken as an example, the temperature variations are analyzed by using fractal correlation dimension under three loads and four disturbances. The measured temperatures are obtained from transient system (TRNSYS) simulation platform and the simulated temperatures are acquired from AHU process model. The results show that the fractal correlation dimension technique can employ only one dimension value to represent the variation characteristic for parameter curve and thus the number of compared parameter pairs is considerably reduced. The noise intervals beyond certain value often have so strong impacts on dimension calculations that data filtering should be preprocessed for these disturbed measurements until measured and simulated dimension calculations are close to each other. Also, the unprocessed minimum disturbing intervals are available for the curve variation of AHU supply air temperature.

Key words: Air handling unit, Fractal correlation dimension, Noise, Parameter estimation, Supply air temperature