Research On Power System Harmonic Estimation Using Improved Kalman Filter

The presence of harmonics in power systems has been sole contribution of ever-growing utilization of non-linear loads as well as power electronic equipment.Harmonic components generated by these non-linear loads degenerate power systems and can be titanic in consequence in terms of operational characteristics of controlling and communicating devices.Further,the detection time for harmonic filters to operate is very crucial as a slow detection may as well affect the real time of the communication and control systems.Therefore,this research work seeks to propose the fastest non-linear state estimation approach which is based on an Improved Kalman Filter(IKF)to measure harmonics from a distorted signal in power systems.Like many traditional algorithms,Kalman Filter(KF)has a trade-off that improving the speed during the transient period may require high overshoot insertion otherwise estimation without overshoot can be carried out naturally,but however,the consequence will be its slower response.With or without combining KF to other intelligences,the aforementioned shortcomings of the filter can be overcome and enhanced by properly tuning of its parameters.As much as to improve operation of power systems by maintaining stable,reliable and quality power supply,the harmonic detection should be very fast as favourable when the recommended harmonic content parameters have been exceeded to enable the compensation device to take appropriate measures and also,there should be an appropriate method to estimate accurately,the presence of the harmonic pollution in these systems.It must be worth noting that the selected approach should be precise while being computationally effectual.In line with this research work,the principles and mathematical modeling of harmonic generation,effects and mitigatory remedies have been studied.Further,highlights of the most commonly adopted traditional intelligences for the harmonic estimation including Fast Fourier Transform(FFT),Recursive Least Mean Square(RLMS)and KF have been analyzed by comparing the merits and demerits of each in terms of speed-convergence characteristics,presences of artificial noise,decaying direct current(DC)component,sampling rate and effect of frequency tracking.Touching on the problem,the fastest detection time of estimation characteristics,without compromising on its accuracy,is accomplished by improving estimation algorithm and it is achieved by the proper tuning of Kalman filtering parameters.Finally,by using MATLAB/Simulink tools,the proposed estimator is used to attempt a virtual estimation of distorted signal.The proposed technique exhibits high performance of superiority and present real time estimation on face of quantities variation of the distorted source.Further,the proposed technique vis-á-vis other conventional methods is simple and robust,hence less hardware requirements and rock-bottom computational cost.