Research And The Implementation On DSP Of Power Quality Detection And Analysis Methods

With the large-scale grid connection of clean energy and the wide application of power electronic components,the power quality of the public power grid is facing more and more serious challenges.At the same time,industrial production and social life are moving towards intelligentization,making more power system loads sensitive to power quality,therefore,the users have also significantly increased their emphasis on and pursuit of power quality.As the support and basis for power quality management,the importance of power quality detection and analysis has been greatly increased.The detection and analysis of power quality implemented on embedded systems has also laid a solid foundation for the development of smart power meters.In this paper,the measurement methods of power quality parameters and the feature extraction and feature classification of power quality disturbance signals are studied,and the related algorithms of parameter measurement and disturbance classification are implemented on DSP.The details are as follows:(1)The domestic and international research status of power quality detection and analysis in parameter measurement and disturbance classification is summarized.The implementations of many power quality detection analysis on embedded systems are compared.The classical electric energy parameter measurement method,harmonic detection analysis method and flicker measurement calculation method are introduced.(2)The basic principles and properties of the S-transform are derived and analyzed.The characteristics of common single and complex disturbances in the time domain and frequency domain are analyzed by S-transform.An improved incomplete S-transform algorithm is proposed,which reduces the computational complexity and storage space requirements by performing S-transform only on the main frequency lines,and then adjusts the Gaussian window width according to the requirements of different frequency segments for time-frequency resolution.A method which extract features with improved incomplete S-transform and classify disturbances by decision trees is designed.Five eigenvalues with higher discrimination are constructed from the results of the disturbance signal passing through the incomplete S-transformation.The threshold value of each eigenvalue is determined in the statistical result of the eigenvalues of a large number of disturbance samples,and finally a decision tree classifier is formed.MATLAB simulation experiments show that the method can accurately classify 6 kinds of single disturbances and 7 kinds of complex disturbances,and the classification accuracy under various noise levels is kept at a high level.(3)The above algorithms for power quality parameter measurement and power quality disturbance classification are implemented on the DSP system.The software design reasonably allocates the physical storage space of the data,and fully utilizes the library function optimized by the DSP itself,thereby reducing the development difficulty and speeding up the running speed of the program.The test shows that the DSP can calculate various power quality parameters and identify power quality disturbances from the original sampling sequence.