Identification of Air Condition Refrigerant Noise by Using Kendall’s Tau Correlation Based on Copula

Abstract

Because the mechanism of air condition refrigerant noise occurring from air conditioners has correlated noise sources, accurate noise source modeled by multi-input/single output system was identified. Joint probability function by Copula model of input and output was considered to analyze contribution. A simulation calculating contribution on pure output of input by partial Kendall’s tau correlation based on the Copula model regarding data with existing correlation between inputs. The Copula models which had highest fitting on input/output by the Vuong and Clarke test of multi-input/single output system model were selected. Using the selected Copula model for calculation of ordinary Kendall’s tau correlation, contribution of pure input on output when correlation between inputs exist was calculated by partial Kendall’s tau correlation. By measuring standard position (output) and noise (input) occurring from each part in the air conditioner noise measurement test, interval where air condition refrigerant noise occurred due to jury test was sorted. The contribution between input and output in the sorted interval was calculated using Kendall’s tau correlation based on the Copula model. Also, Kendall’s tau correlation between inputs was calculated and input with high correlation was set as residue and calculated by partial Kendall’s tau correlation. Pure contribution of inputs on output was graphically expressed using Vine Copula. The order of parts contributing in air condition refrigerant noise was known through Partial Kendall’s tau correlation. Instead of denoting the part name and location of the specific air conditioner, parts 1, 2, 3 and 4 are replaced with the names. Air conditioner part 1 most contributed in air condition refrigerant noise in the sorted interval.