Nonlinear loads model for harmonics flow prediction, using multivariate regression
PublisherInstitute of Electrical and Electronics Engineers (IEEE)
Rights accessOpen Access
This paper describes a method for obtaining a model of a single or a set of nonlinear loads (NLL) connected to a certain point of an electrical network. The basic assumption is that the network supplying the NLL has significant series impedances and is disturbed by other parallel, random, and unknown neighbor loads, sharing part of the supply system with the NLL. The main interest for obtaining the model is its further use to predict the amount and flow of harmonic currents generated by the NLL, in the case of adding a filter to reduce the harmonics distortion. The modeling technique used in the paper is based on multivariate multiple outputs regression and leads to a set of equations giving the NLL behavior (one for each of the harmonic currents). The model is obtained from data taken at measuring point and is only valid to predict the NLL behavior when new loads are connected at this point. The modeling method was first tested with V, I data coming from simulations using a MATLAB-Simulink SimPowerSystems toolbox. Finally, the method has been validated using V, I data taken in a real installation with different neighbor loads and under different load conditions.
CitationLamich, M., Balcells, J., Corbalan, M., Griful, E. Nonlinear loads model for harmonics flow prediction, using multivariate regression. "IEEE transactions on industrial electronics", Juny 2017, vol. 64, núm. 6, p. 4820-4827.
- TIEG - Grup d'Electrònica Industrial Terrassa - Articles de revista 
- L'AIRE - Laboratori Aeronàutic i Industrial de Recerca i Estudis - Articles de revista 
- Departament d'Enginyeria Electrònica - Articles de revista 
- Departament d'Estadística i Investigació Operativa - Articles de revista