Research On The Harmonic Detection Method Based On Rotating Invariant Subspace

With the improvement of industrial automation, due to the growing complexity of power load of power, the harmonic pollution is more and more serious. Because of the increasingly complicated harmonic signal, harmonic detection analysis method in power system has been called for new requirement. At the same time quickly and accurately extract harmonic signal is one of the key technologies of active filter and the reactive power compensation. So it is necessary to combine new methods to obtain the harmonic parameter. The method will provide a new analytical tool for harmonic analysis.Space spectrum estimation is an important research area of array signal processing. Subspace rotation invariant algorithm is an important method of space spectrum estimation. The application has involved in many military and national, and economics areas such as radar, communication, sonar, earthquake, exploration. There are some papers has been proposed in the application of harmonic detection. In this paper rotating space invariance algorithm will be introduced in harmonic parameter detection, which provides a new method of harmonic detection. The main research contents are followed.1. It introduces the background and situation both at home and abroad of the research based on the rotation invariant algorithm. It also gives the concept and procedure of the algorithm. We discuss the differences and characteristics between the rotational invariant algorithm and other space spectrum algorithm.2. It introduces the algorithm based on the least square method and space rotation invariant algorithm (LS-ESPRIT). And according the algorithm, it proposes to use the method which is based on Toeplitz matrix method and space rotation invariant algorithm (TAM) to detect the harmonic. The algorithm combines the advantages of rotation invariant subspace and least-square method. After calculation through the singular value decomposition it uses the singular vector instead of data covariance matrix, and constructs the new matrix by singular value diagonal matrix and eigenvector of Signal subspace. Finally, use the least square method to calculate the eigenvalue and get the frequency amplitude parameters. The simulation experiments show the method of TAM is obviously superior to the method of LS-ESPRIT.3. It introduces the algorithm based on the total least square method and space rotation invariant algorithm (TLS-ESPRIT). And according the algorithm, it proposes to use the method which is based on conjugate matrix space rotation invariant algorithm (C-SPRIT) to detect the harmonic. The algorithm inherits the advantage of the least square method and the noise immunity is better. Construct a subarray which contains the same number of sample data, then gives the singular value decomposition of the data matrix, finally, uses the conjugate of rotation matrix instead of rotation matrix in array signal processing and gets the frequency amplitude parameters of harmonic. The simulation examples demonstrate that the algorithm is capable of having higher precision and stronger anti-noise capacity than TLS-ESPRIT.