Research On Threshold Strategy And Fast Denoising Algorithm Of Local Discharge Signal Of Cable

With the popularity of electric power equipment,including remote areas and mountainous areas are electrified,greatly meet the needs of people's work and life,but in the long-distance cable laying,harsh environment and cable aging and other reasons may lead to power failure or voltage instability and partial discharge.Partial discharge is one of the causes of cable insulation failure due to cable aging or deterioration.In order to ensure the power supply at any time and not affect people's work and life,it is necessary to detect the cable equipment in engineering.In fact,the cable equipment detection signal(partial discharge signal)is often drowned by noise.In order to obtain accurate cable state,the de-noising of partial discharge signal is a research hotspot at home and abroad.For the denoising of partial discharge signals,we adopt the method of extracting characteristic values to study them,namely Singular Value Decomposition(SVD),but with low accuracy.People have a growing demand for power,and the more data generated by partial discharge,the current single machine denoising can no longer meet people's needs,has no real-time,and the processing efficiency is low.With the rapid increase of information and data,people have gradually stepped from the information age into the era of big data.Based on distributed platforms Hadoop and Spark,people have made great progress in data storage and processing compared with the information age.For the above partial discharge signal accuracy and low efficiency,the main work of this paper includes the following three aspects:(1)an akaike information criterion(AIC)based singular value denoising algorithm median threshold estimation method(AIC-SVD)is proposed and applied to partial discharge signal denoising.In fact,the selection of singular values for noisy pd signals can be approximated to the estimation of the dominant component(subspace dimension)in information theory.Based on the existing research strategy STSVD,this paper denoising partial discharge signals to improve the accuracy of partial discharge signals and achieve better results in detecting the insulation state of cable equipment.In this paper,the feasibility and accuracy of the algorithm are verified by comparing the experiment with the existing algorithm,and the desired effect is achieved.(2)A threshold estimation method based on optimal singular value threshold(OSVT)and the above mentioned AIC-SVD is proposed,which is applied to the denoising of partial discharge signals.After the above experiments,considering that the effect of AICSVD algorithm is not as good as general singular value improvement algorithm when SNR is low,the threshold optimization algorithm is applied in this paper when SNR is low,and AIV-SVD is applied when SNR is high.Through the comparison and analysis between the experiment and AIC-SVD,this paper verifies that the algorithm is more effective and accurate than the previous algorithm,and achieves the desired effect.(3)A singular value decomposition(Sp-svd)algorithm based on Spark is proposed and applied to de-noising partial discharge signals.At present,the singular value denoising algorithm of partial discharge signal does not solve the problem of too much data,so this paper proposes a new algorithm to solve this problem.Since the partial discharge signal is a two-dimensional signal,the data is preprocessed,partitioned and then transformed into a matrix for denoising.In this paper,the experiment is compared with the existing SVD algorithm and the experimental results are analyzed to verify the feasibility and efficiency of the algorithm and achieve the desired effect.