Research On Power System Harmonic Detection Algorithm

With the progress and development of the society,more and more new power equipment and power electronic products are constantly appearing in daily life,which not only improves the quality of life but also brings more harmonic pollution,so that people have to pay more attention to the detection and treatment of power system harmonic.Hence,more attentions have been paid to the detection and treatment of harmonics in power systems.Many scholars have conducted a lot of research on how to promote high-efficiency and high-precision harmonic detection in power system,and obtained effective research results in real-time.However,with the continuous reform of power system structure,coupled with the harmonic of power system is an unstable nonlinear signal,it is difficult to achieve better results in the application of traditional processing methods for detection and processing.The present thesis presents the real-time harmonic detection algorithm and the predi ction of harmonic content in power system.The specific contents are as follows:1.The basics related to harmonic detection theory are introduced.The traditional harmonic detection algorithm and the improved algorithm are studied,mainly including the fast Fourier transform,the wavelet transform and the instantaneous reactive power analysis.The advantages and disadvantages of various methods in the application of harmonic detection are compared and analyzed.2.Aiming at the problem of the lack of adaptive capability in the processing of power system signals by traditional ensemble empirical modal decomposition,an improved ensemble empirical modal decomposition algorithm is proposed by combining particle swarm optimization and deep belief network.The a lgorithm is trained to obtain an adaptive model,so that the set empirical mode decomposition can automatically select effective decomposition parameters according to the distribution characteristics of the harmonic content of the power signal,and the har monics of similar frequencies in the measured signal can be more reasonably separated into the corresponding eigenmode function.Experimental results have proved that the method can decompose the power signals of different loads adaptively,and has a bette r capability of harmonic separation and a higher precision,and the separation result is more conducive to the detection and processing of harmonic,further improving the efficiency and accuracy of power system harmonic detection.3.In view of the nonlinearity and unsteadiness of power system harmonics,the adaptive empirical mode decomposition is combined with the least square support vector machine to establish the harmonic combination prediction model of power system.Based on the characteristic of each eigen modulus function after the adaptive decomposition,the prediction model can be built for each component separately.It has been proved that compared with other prediction models,the model established in this paper can deal with a variety of complex power system signals more effectively,accurately predict each harmonic component by the demand of the power system,and simultaneously satisfy a variety of prediction situations,so as to provide a reliable basis for harmonic management of the power system.