Research On High Precision Power Harmonic Detection Algorithm

With the access of a large number of renewable energy sources and the widespread use of non-linear loads in the power grid,the structure of the power grid is extremely complex and a large number of harmonics are generated,which will inevitably bring great harm to the safe operation of the power grid.The most direct impact is to cause a large amount of power loss.Therefore,how to effectively manage harmonics has become an important concern for both power suppliers and consumers,and the premise is to continuously improve the detection accuracy of power harmonic signals.At present,power harmonics can be generally divided into steady-state harmonics and dynamic harmonics,corresponding to the steady-state and dynamic harmonic detection methods,among which the fast Fourier transform（FFT）is easy to implement and has high accuracy is widely used in steady-state harmonic detection,but in the process of processing due to asynchronous sampling and other reasons will lead to spectral leakage and the occurrence of the fence effect.So,this paper proposes a hybrid convolutional window and multispectral interpolation FFT algorithm in order to improve the precision.In addition,under some operating conditions of the power grid,the changes of harmonics are dynamic,and there are also various time-varying noise interference.At this time,the traditional algorithm is no longer capable.To solve this problem,this paper proposes a dynamic harmonic detection algorithm based on adaptive cubature Kalman filter（ACKF）algorithm.Finally verified the performance of the above two algorithms through simulation,the details are as follows:Firstly,the basic theory of Fourier transform algorithm is introduced for steady-state harmonic detection.After analyzing the problems existing in FFT algorithm,the effectiveness of window function and interpolated spectral line in suppressing spectral leakage and fence effect is discussed and verified.Based on the spectral characteristics of the traditional cosine window function,a hybrid convolution window is constructed,and a six-spectral interpolation algorithm is introduced on the basis of two-,three-and four-spectral interpolation.Through simulation analysis and comparison,the hybrid convolutional window and multispectral interpolation FFT algorithm is verified for steady-state harmonics detection.Secondly,in the dynamic harmonic detection,the basic principle and implementation steps of Kalman filter（KF）are shown through formula derivation.To address the limitations of the traditional KF algorithm to deal with dynamic harmonics and vulnerable to system timevarying noise interference and other problems,the use of the state noise covariance matrix is optimized based on the fading memory exponential weighting method,and a dynamic harmonic detection algorithm is proposed based on ACKF.The algorithm is verified by comparative experiments for dynamic harmonic signals,and it is also applied to harmonic detection in the active filter.Then a harmonic detection system is designed with EP4CE30F23C8 produced by Altera as the main control chip,and the relevant hardware and software design are completed,and the sampled signal is transmitted to the host computer for harmonic analysis to verify the function of the system.Finally,through the analysis and summary,the following aspects to be improved are proposed.