Research On Adaptive Harmonic Current Detection Method For Active Power Filter

At present, active power filter (APF) has became the most effective metod to suppress harmonic pollution, due to it can track and compensate harmonic and reactive currents in real-time and dynamic manner. However, to improve the performance of the APF, it needs to rapidly and accurately extract harmonic currents or distorted currents in advance. By reviewing presented harmonic current detection methods as well as on the basis of the Fourier series and adaptive filtering algorithms, this paper theoretically studys adaptive detection algorithms of the harmonic currents or distortion currents.providedFirst, the basic principle of two kinds of adaptive harmonic current detection models are described, which are based on adaptive noise cancellation technology (ANCT) and Adaline, respectively. The detection performance, defects and the scope of applications of the two models under the condition of the same adaptive algorithm are analyzed, and the impact on the detection performance of different adaptive algorithms with respect to the same model is also provided. Analysis results show that under the conditions of the same algorithm, the convergence rate of the two models is consistent, and the steady state error of the latter is small (theoretical value may be0) but computationally intensive, the former is only used in APF while the latter has other applications such as power quality assessment; the impact of different adaptive algorithms on the detection performance of the same model is determined by the algorithm itself. In addition, on the basis of ANCT, adaptive harmonic current detection principle for three phase cirucuit is introduced.Then, by analyzing the shortcomings of the least mean square (LMS) and the exsiting variable step-size harmonic current detection algorithms, this paper proposes a novel variable step-size LMS harmonic current detection algorithm based on the modified versiera function. The algorithm eliminates the interference of the distorted currents using the time mean estimate of the instantaneous cross-correlation between the reference input signal and the error signal (which is difference between the load current and the filter output). The modified versiera function is used to adjust the algorithm’s step-size, thereby ensuring that the dynamic response time is not increased. This method has not only a smaller steady state error and a better tracking capability of the jump system. Finally, aiming to the shortcomings of the LMS and recursive least square (RLS) algorithms for harmonic detection, an adaptive weighted LMS algorithm to detect distorted current is developed in this paper. The algorithm has a faster convergence rate than the LMS algorithm for guaranteeing the same steady-state error, and also has a better tracking capability than the RLS algorithm when the load current has a sudden change, due to the fact that the updating of the weights is not subject to disturbance of the distorted currents. Moreover, the computational complexity of the algorithm is between the LMS and RLS algorithms.The computer simulations of the adaptive distorted current detection of the single-phase and three-phase circuits that verifiy these works in this paper.