Study On De-noising And Recognition Of Power Quality Disturbance Signal

At this stage,the mutual influence between the nonlinear equipment and the impact load and the interference of the surrounding environment cause interference to the power grid.This situation make the real-time collected power signal doped with a large amount of noise.It submerges the characteristic information of the disturbance signal.This caused serious interference to the analysis and recognition of the transient disturbance signal,and resulting in the complexity of power quality issues.In order to accurately analyze disturbance signal and power quality issues,this paper mainly does some work on the de-noising,detection and recognition of transient disturbance signals.Finally,based on the Lab VIEW platform,the power quality indicators are analyzed.The main tasks are:First,a wavelet de-noising algorithm based on adaptive decomposition layer number and threshold is proposed in this paper.The algorithm calculates the peak-to-sum ratio of the wavelet detail coefficients.The number of optimal wavelet decomposition levels is determined adaptively.According to the distribution characteristic of the useful signals and the noise signals in the detail coefficients of each levels and the peak-to-sum ratio of the negative and positive values of the detail coefficients.The upper and lower thresholds of the detail coefficients of each levels are dynamically adjusted.We use MATLAB platform to de-noise transient oscillation and pulse signals.The simulation results show that a wavelet de-noising algorithm based on adaptive decomposition layers and thresholds is compared with the traditional hard and soft threshold algorithm and an improved wavelet threshold algorithm.The signal to noise ratio obtained is the largest and the root mean square error is the smallest.The reconstructed signal is closer to the original signal,and better preserves the characteristic information of the signal during the disturbance period.Secondly,this paper propose a power quality disturbance signal classification method based on feature combination.The algorithm uses the de-noising algorithm proposed in this paper to process the disturbance signal.We use Hilbert-Huang Transform(HHT)to get instantaneous frequency characteristic and marginal spectrum characteristic,and extract feature data from it.Then we combine the S-transform to obtain a two-dimensional modular time-frequency matrix,extract the eigenvalues of the fundamental frequency amplitude,and combine the features extracted from the HHT transform.A method for identifying eight types disturbance signals of transient power quality is proposed.Finally,the effectiveness of the method is verified by Support Vector Machine(SVM)classifier.Finally,a power quality analysis system based on Lab VIEW platform was built.The system can dynamically display power quality indicators such as voltage unbalance,current unbalance,power grid harmonics and voltage fluctuations in real-time.On the Lab VIEW platform,we can achieve the proposed power quality disturbance signal de-noising method and feature extraction method.The test results show that the power quality analysis system is easy to operate,the error of the indicators analysis is small,and the algorithm runs quickly and easily.