Key Control Based On The Recursive DFT For Active Power Filters

With the world-wide concern of global warming and energy crisis,the word-wide interest on distributed energy generation,micro-grid,flexible power transmission,et al.,which are all based power electronic devices,have received wide attention.These applications,while diversifying power transmissions and enhancing energy harvest,may,on the other hand,worsen the power quality and endanger the reliable operation of power systems due to the switching ripples and harmonic resonance.Concerning the aforementioned problems,the active power filtering(APF),as one of the most efficient tools to mitigate the harmonic pollution and improve the power quality,have been further extended and applied in the multi-function inverter,distributed harmonic compensation,active resonance damper and so on.This thesis is aimed to improve the control performances of the APF in order to better adapt it for these new applications,and to tackle the arised challenges of fast selective harmonic extraction,advanced harmonic current control and tracking,and grid synchronization under adverse grid conditions.The recursive discrete Fourier Transform(RDFT)is utilized as the mathematical tool for harmonic analysis and selective extraction.It's pointed out that RDFT is widely applied in the APF due to its advantages of simplicity,excellent real-time performances and high steady-state accuracy.The recursive form of RDFT,however,can cause potential instability issues due to the accumulation of the errors introduced by the finite-word-length effects in real practical discrete systems.In-depth research on the digital implementation of the RDFT is carried out;it is revealed that rounding errors of arithmetic operations are the key reason behind the unstable phenomenon of the RDFT.And to deal with the instability problem,two methods are proposed in this paper to eliminate the error accumulation through swapping the calculation sequences of the sliding DFT.In addition,mathematical model of APF is discussed with focus on control delays,its detrimental effects and possible mitigation measures,all of which lay the foundation for further research of the thesis.The generalized RDFT is proposed for fast and flexible selective harmonic extraction in the APF.Mathematical model of the convention RDFT is established from the transfer function views,and through which,it's revealed that RDFT relies on the mechanism of pole-zero cancellation for harmonic extraction.The conventional RDFT,however,suffers from disadvantages like slow dynamics and inflexibility.It's further proposed in the thesis to reconfigure the RDFT to adapt for frequency spectrum of the input signal,and to improve the dynamic responses and system flexibility.The proposed GDFT not only maintains the advantages of simplicity and selectively filtering properties but also features fast transients,around 0.3 fundamental cycle for typical applications,which is much shorter than the one-cycle settling time of the conventional DFT.Extensive comparisons with synchronous dq frame methods and cascaded delayed signal cancellation strategy are also performed to demonstrate the advantages of the proposed algorithms.An enhanced recursive DTT-Based controller for selective harmonic compensation in active power filter is proposed.Starting from the internal model principle,the connections between the proportional-resonant control and the repetitive control are analyzed and a general structure for selective current control is establised.And the recursive DFT is utilized as a specific instantiation.Phase leading and proportional gain are integrated in the proposed current controller for phase compensation and dyanmic adjustment.Moreover,the correction function is included in the forward path of the proposed controller to achieve zero gain and zero lagging phase.The proposed current controller features flexible structure and is able to ajust respective parameters for each harmonic according to the magnitude and phase response properties of the control plant.The proposed controller also has the advantage of selective harmonic control to minimize control interferences of different harmonic orders.Besides,zero steady-state errors can be realized through the proposed technique.Detailed parameter tuning method is further illustrated using the source-current controlled APF as an example.Extensive experiments show that high compensation accuracy and fast dynamics,i.e.,1~2 fundamental cycle are achieved.The grid frequency adaptive issues of RDFT are also approached to enhance the performances of APF under adverse grid conditions.The Lagrange interpolation method is adopted to realize fractional-order delay to tackle the digital implementations of RDFT and its frequency adaptiveness with fixed sampling period.Two grid synchronization methods of RDFT-based dq frame PLL and adaptive notch filter based FLL are proposed to accurately extract grid frequency even under adverse grid conditions.Inner connections of conventional advanced synchronization methods are established to derive the proposed methods.Small-signal modeling is also carried out for parameter optimizations.Experimental results are provided to verify the proposed algorithms.