Research On Detection And Identification Of Power Quality Disturbance Signals

With the rapid development of modern intelligent information technology,the operation of the power system is becoming more and more intelligent.Many advanced nonlinear power electronic devices and precision equipment are put into the power grid.Many advanced non-linear power electronic devices and precision equipment are put into the power grid,while providing people with high-quality experience,it also makes the power in the power system polluted.Therefore,while enjoying the convenience of modern electronic equipment,we should also focus on the improvement of power quality.In order to ensure the efficient operation of power grid and the safe use of users,relevant workers need to quickly and accurately analyze the start and end time and duration of power quality signal disturbance,and correctly judge the specific type of disturbance signal.However,the existing detection and recognition methods still have the problems of insufficient accuracy and less types.Therefore,this study focuses on the analysis of transient power quality disturbance signal,and introduces some algorithms,such as variational mode decomposition,adaptive noise full set empirical mode decomposition,gray wolf optimization support vector machine and so on.In this paper,firstly,a method of combining variational mode decomposition(VMD)to eliminate noise in power quality disturbance signal is proposed.Referring to the relevant standards at home and abroad and combining with MATLAB 2019 software,the waveform map is simulated.At the same time,random noise is added to the disturbance signal to ensure that the simulated waveform map is as close to the disturbance signal collected in the real power grid as possible.Then,the method of variational mode decomposition combined with permutation entropy is used to denoise the noise signal.Then,the full set empirical mode decomposition method based on adaptive noise is used to detect and analyze the waveform.It can adaptively split the signal,and add Gaussian white noise in each process of splitting to obtain the intrinsic mode function components.It can solve the mode aliasing phenomenon in the process of splitting,and has higher efficiency and stronger adaptability.The waveform is decomposed by this method,and then transformed to frequency domain by Hilbert transform.Finally,the recognition model based on gray wolf optimized support vector machine is designed.Firstly,nine different types of waveform signals including single and compound disturbance signals and normal signals are simulated by simulation software,which are used as the sample data of identification model.Then the energy entropy theory is used to extract the features of the data to improve the accuracy and efficiency of the model classification.Then the recognition model is built for training and testing.In order to verify the effect of the proposed method,the advantages of the proposed denoising method are highlighted by comparing the experimental results with the common S-G convolution smoothing and wavelet denoising methods.In the signal disturbance detection,the simulation experiment is used to test,and compared with the actual value,the detection result is more accurate.Finally,while building the classification model of this study,the BP neural network and extreme learning machine classification model are constructed for comparison.The experiment shows that the classification model of this study has higher recognition accuracy.Figure[31]table[14]reference[65]...