Research On Denoising Algorithm Of Partial Discharge Signal Based On ACEEMDAN-AWTE

Partial discharge signal is an important indicator of insulation aging condition of electrical equipment.So monitoring the partial discharge of electrical equipment is helpful for grasping the insulation situation and arranging overhaul reasonably to reduce the loss caused by equipment failure.However,limited by the low accuracy of measurement sensor and the complex operating environment,the measured partial discharge signals are always mixed with noise,which causes a great interference to observing the insulation information of electrical equipment.So it is necessary to de-noise the measured signal.In this thesis,an adaptive wavelet threshold estimation de-noising algorithm based on complete ensemble empirical mode decomposition with adaptive noise algorithm(ACEEMDAN-AWTE)is proposed for denoising the noise in partial discharge signal.The algorithm consists of adaptive wavelet threshold estimation(AWTE)algorithm and adaptive complete ensemble empirical mode decomposition with adaptive noise(ACEEMDAN)algorithm.In the AWTE algorithm,wavelet,decomposition level and threshold are obtained adaptively.When choosing the optimal wavelet,energy proportion of approximate coefficients in the sum of approximate coefficients and detail coefficients is calculated.Then it is divided by the wavelet energy entropy of approximate coefficients.Wavelet with highest value is selected as optimal wavelet.When choosing the optimal decomposition level,the noise whitening test of detail coefficients is carried out.When the whitening condition is not satisfied by detail coefficients,the decomposition is terminated and the optimal level is determined.When choosing the optimal threshold,whale optimization algorithm is used to optimize the optimal threshold searching process,which improves the accuracy,astringency and resolution of the algorithm.The de-noising experiments of bumps signal and quadchirp signal shows that signals de-noised by AWTE algorithm has the smallest maximum amplitude error and the de-noising effect is convincing.In the ACEEMDAN algorithm,adaptive complete ensemble empirical mode decomposition with adaptive noise is performed for noisy partial discharge signals.Then the AWTE algorithm is used to de-noise the intrinsic mode function.For the de-noised functions,the waveform similarity coefficient between the component and the component before de-noise is calculated.When the coefficient is low,the white noise is more and the decomposition continues.When the coefficient is high,the white noise is less and the decomposition is meaningless,the decomposition is terminated.Through this way the de-noise level is determined.The following de-noise experiment of simulation partial discharge has proved that the ACEEMDAN-AWTE algorithm can de-noise the partial discharge signal with the lowest mean square error and highest waveform similarity coefficient under different signal-to-noise ratios.Consequently,the algorithm has high accuracy and an adaptive scope.And the practicability is strong.On the basis of the above research,this paper design and develops a software system for electrical equipment partial discharge signal denoising.This system can filter the white noise in the partial discharge signal,using ACEEMDAN-AWTE algorithm.And it has the functions of user management,signal de-noising,de-noising result show,parameters adjustment and data storage.In this system,users can adjust the parameters of the de-noising algorithm adaptively and recover the partial discharge signal with high accuracy.The parameter adjustment functions in the system are operated in a graphical way,which simplified the configuration process of de-noising algorithm and has a good application value.Finally,the system is used to de-noise the measured noisy partial discharge signal of reactor,which proves the practicability of the system.