The Application Of Neural Network Based On The Harmonic Basis Function In Harmonic Detection

With the development of modern power electronics technology, the application of various power loading which are non-linear become more and more wide. There are many time-varying harmonics caused by the non-linear instruments that convert or invert the current or voltage of the power system, which causes the harm of harmonics in electric system is much more serious. The real time and accurate detecting of harmonics is significant to the supervision and prevention of contamination.Active power filer is an advanced power electronic device, which can be used for integrated compensating harmonic. Compared with some conventional measures, active power filer has many advantages and favorable prospects. Because of the characteristics, real time and accurate compensation, it is possible to take full advantage of digital signal processing and control technologies. If so, not only can the functions of active power filers be optimized, but also the performances of active power filers can be improved significantly. But it is difficult to adjust circuit parameters when loads or power conditions are changed. And the realization of the instantaneous reactive power with analog circuit needs large numbers of multiplicative implements, which cause the cost very expensive.Above all, a neural network model named harmonic basis function (HBF) of multi-frequency periodic signals is proposed, and a harmonic analysis algorithm base on the HBF model is presented in this paper after compared a few harmonics detecting methods in existence. In the supposed algorithm, the fundamental frequency and the harmonic amplitude-phase parameters are introduced as weights needed to be adjusted. The harmonic parameters are estimated through the adaptive measurement theorem. The convergence theorem of the algorithm provided theoretical guides for selecting of the learning rates. Simulations are conducted on the signals with frequency deviation as well as with white noises. The results show that the algorithm achieves high accuracy and rapid speed in convergence and is a good candidate for measuring the harmonics with asynchronous sampling and short data in power systems.High real-time precision of harmonic current detection is vital for the performance of active power filter (APF). In this paper the harmonic basis function (HBF) neural network is applied to adaptive noise cancellation technology, and the neural network is trained by least mean square (LMS) algorithm. The phase variation and frequency fluctuation of distortion currents are detected by the adaptive filter real-time. Simulation and experimental results show that the method has many advantages, such as real time, high accuracy, adaptability to load currents. And it is much more improved than the conventional measures in many aspects such as detecting distortion currents, harmonic currents, or the adaptive following ability to load currents. The algorithm of this detecting method is also simple and easy to realize and modulate.As a result, the theoretical analyses and simulation prove that the method is feasible and effective, it is indeed a good method in detecting distortion currents.