Improvement of power quality in distribution utility using X-LMS based adaptive algorithm

This study presents the design and functional performance of Cross Least Mean Square (X-LMS) neural network based distributed static compensator for shunt compensation. First, considering the adaptive least mean square (ALMS) technique, harmonics correction, power factor (p.f) upgrading, load balancing, and voltage regulation at the supply bus are attained. Second, based on the X-LMS principle, active and reactive components of load current by updating their respective weights are mathematically analyzed. The main objective is to extract the tuned weighted values of fundamental active and reactive power components of distorted load currents using the suggested adaptive principle. The X-LMS exhibits better convergence speed provides a significant computational saving and avoids the additional filtering needed with the ALMS structure. Finally, the X-LMS model is built using MATLAB/Simulink for simulation study and validated experimentally by d-SPACE laboratory setup results. The parameter design of X-LMS can be used to improve the harmonic reduction with other compensation performances to showcase better power quality of the distribution utility.