Research On Flicker And Interharmonic Detection Technique Based On Blind Source Separation Theory

With the continuous application of large power electronic devices of the non-linear and impact loads, which cause the voltage flicker and interharmonics, the power grid is polluted and its quality is declined. People's daily life and production is affected. So detecting the flicker and interharmonics accurately is a basis of evaluating power quality, and a premise of improving power quality. Currently wavelet analysis and various fast Fourier transform methods are used to detect flicker, harmonics, and interharmonics. However, these methods are vulnerable to spectrum leakage and grid effect. In order to improve the precision of detecting flicker, harmonics, and interharmonics, some research have done. Firstly, this dissertation researches blind source separation (BSS) algorithm to separate waveform of flicker, harmonics, and interhamonics from input signals and calculate the frequency of flicker and interharmonics. Secondly, this dissertation deeply studies the methods for calculating the amplitude and phase of flicker and interharmonics. Finally, the testing system of flicker and interharmonics has been developed to validate the proposed detection method and resolve some key technologies in flicker and interharmonics detection. The main contributions of this dissertation are as follows:This dissertation proposes a method of constructing a delay matrix using the output signal sequence of single voltage transformer instead of the multiple sensor output signal arrays. The matrix satisfies the basic requirement of BSS so that each frequency component is separated from flicker, harmonics and interharmonics which contain a number of frequencies.To accurately obtain the frequency and fluctuation of the flicker and avoid spectrum aliasing and spectrum leakage, This dissertation uses fast independent component analysis (Fast-ICA) based on minimum mutual information algorithm to separate the flicker envelop curve from sampling signals. Then frequency and fluctuation of the flicker is calculated. In order to improve the precision, BSS and particle swarm optimization are selected to detect flicker, and implement the frequency and amplitude calculation of carrier signal and flicker envelope curve respectively. It improves detection precision compared with Hilbert - Huang transform. With the application of db24 wavelet function for decomposing voltage sampling signals, the flicker envelope curve can be separated in low band. Also, Fast-ICA is compared with this db24 wavelet function and the former results show that detection error is lower than wavelet transform. In the noisy case, we research the BSS method to separate the flicker curve. As the noises, flicker signal and carrier signal are independent of each other, the robust pre-whitening method has been used in BSS and the flicker envelope curve is separated smoothly from the signal. This method eliminates the denoising process, lessens the costs of hardware and software and further improves the calculating efficiency.Harmonic is an important part of power quality. To identify the fundamental and interharmonics parameters, this paper proposes the simultaneous diagonalization second order blind identification algorithm to separate each frequency component. Meanwhile, every frequency value is computed. Then the amplitude and phase are calculated by using the immune particle swarm optimization (IPSO) algorithm. Theexperiments results show that the precision of this method is higher than FFT and Blackman-Harris window function interpolation FFT at interharmonics, especially for the phase precision. IPSO avoids the premature phenomena. This dissertation makes good use of these methods and the precision is also higher. Experiment results approve that it is an effective interharmonics detection method.The flicker and interharmonics testing hardware platform to validate the proposed voltage flicker and interharmonics detection methods is implemented. It accomplishes the flicker envelope curve detection using the frequency and amplitude calculating algorithm. The fundamental and interharmonics frequency, amplitude and phase calculating algorithm are implemented on this platform, too. The experiments are done including voltage sensor calibration and testing in order to evaluate the reliability and the real-time ability of the algorithm and the function of the testing system.