Study On Harmonic Detection Methods In Power System Based On FFT And Wavelet Transform

This paper mainly introduces the basic concept of the power system harmonic, the principle of the harmonic production, hazard of harmonic, harmonic source and deficiency, of the traditional harmonic detection methods as well. And also the paper has studied continuous Wavelet Transform, discrete Wavelet Transform in wavelet theory and their time-frequency analysis.Currently, it is hard to analyze and detect dynamic harmonic quickly and exactly using the traditional analysis methods for power system harmonic. The FFT has the advantage to analyze steady harmonic, while the Wavelet Transformation has the advantage to analyze fluctuant and fast changing harmonic, and which can compensate with each other well. This paper has studied a new method that combines the Fourier Transform with the Wavelet Transform so that we can analyze any harmonic well.The algorithm that combines Fourier Transform with Wavelet Transform is as follows: For the observed signal, firstly, the signal is decomposed into the high frequency part (in detail) and the low frequency part (approximate) by the discrete Wavelet Transform. Non-steady harmonic and singularity signal part such as mutation and discontinuous point which are generally in the high frequency part after Wavelet Transform are analyzed by modulus maximum analysis; while the low-frequency part is analyzed by the Fourier Transform to get steady harmonics. This algorithm can not only get transient component but also get exact steady harmonic component. According to this method, we can realize that following any harmonic and supply the theory basis of the real-time compensation to harmonic effectively. In this paper, the method is simulated, that combines the Fourier Transform with the Wavelet Transform, and the simulation shows that the Wavelet Transform has double resolutions in both time & frequency domains, which can solve the problem that the Fourier Transform can't do well. We can not only get various sub harmonics, but also abstract correlative signal transient component, location of the transient component original time; discontinuity, protruding, concave of voltage & current wave and detection of transient component adopting wavelet analysis algorithm to analyze the power system harmonic.This paper designs a new filter bank structure, and it can correspond the node decomposed from the Wave Packet Transform with the decomposed frequency segments by the Wave Packet Transform and find the array rule of this filter bank structure, with which we can construct the filter bank structure which is decomposed several times.From the perspectives of the relations between stationary process and its spectral distribution, and between the entropy and the variance of a random variable, the application area of FFT is further clarified and the so called uncertain relation is equivalent to that the entropy distance between the distributions based on time and frequency is always positive.The hardware part and the software part of the harmonic detection system are designed in this paper. To satisfy the need of the great operation of the real-time harmonic analysis algorithm, we adopt the design method of DSP+ARM in hardware part. This design can make full use of the digital processing advantage of DSP and the control ability of ARM to realize the complex software algorithm in system, and the executing speed has improved. There are two parts of software design of system; one is the program design of DSP, including data sampling, data processing and etc, the other is control program design of DSP, including subprogram of human computer interface, remote communication etc.