Determination of Measurement based Air Conditioner Load Models

Abstract

This paper presents an air conditioner's mathematical static load model in a power distribution network. A mathematical model has been proposed to obtain a polynomial load model for air conditioner load. Mathematical analysis has been carried out by creating a large dataset experimentally by varying voltage and frequency, which has been further interpolated to obtain the final load model. Six different types (i.e., ratings and brands) of air conditioners have been used for experimental verification purposes. MATLAB software toolbox Neural Net Fitting tool has been utilized for the computation purpose of obtaining the best fit model. As the data set created is considerable, different metrics., i.e., Mean Squared Error (MSE) and Regression Value R, are used to evaluate the performance of the trained model. Finally., the obtained model has been compared with the theoretical models.