Detection And Noise Reduction Technology For High Voltage Excitation Flashover Signal Of Power Cable

Underground power cables are more and more widely used in urban power grids because of their small footprint,reliable operation,and improved urban environment.However,with the social and economic growth of our country and the expansion of the scale of cities,the cable load increases,the working cycle increases,and the insulation aging and other factors are prone to failure.Because cables are laid underground,it is difficult to find faults,and long-term power interruptions will have a great impact on social production and people’s lives.Therefore,timely and accurate positioning of cable fault points has important practical significance.Based on the current development of cable fault location technology and sound signal processing technology,this thesis aims at the detection and noise reduction of the flashover discharge sound signal in the fixed point of the cable fault under the complex background noise,and studies the methods of detecting and enhancing the discharge sound signal to extract effective The discharge sound signal,so as to accurately locate the cable fault,the main work includes:First,the key issue that needs to be solved based on the cable fault of sound synchronization is analyzed.The pretreatment method of cable fault point flash discharge sound signal is given,mainly including sound signal interception,pre-emphasis,Framing and adding window function,noise estimation and endpoint detection.The method first retains the sound signal of the 200 ms period after the moment of the magnetic field signal in the cable fault point flash discharge information.Secondly,the sound signal in this period is pre-emphasized and windowed to get the frame unit of sound signal processing.Then obtain an estimated noise spectrum of each frame based on the noise estimation calculation method of statistical characteristics.Finally,calculate the frequency band varying value of each frame of sound signal,and the frequency band difference value calculated by the estimated noise spectrum is used as the threshold,and the start endpoint of the sound segment is obtained.The simulation of the noise estimation algorithm and endpoint detection algorithm in matlab verifies the validity of these two algorithms.The former can obtain accurate background noise spectrum,the latter can accurately determine the starting point of the sound segment.Secondly,for the detection of cable fault point flash discharge sound signal,this thesis proposes a vector quantized discharge sound signal identification scheme characterized by spectral centroid and Meer frequency pumping factor(MFCC).Based on the short-term Fourier transform feature analysis method,the scheme is selected for the characteristic vector of the spectral centroid and MFCC to form a 25-dimensional feature vector,and the Best code book template for the vector quantifier of the LBG algorithm of the split method.The simulation carried out in MATLAB verifies the effectiveness of the scheme,and the trained codebook template can still identify discharge sounds under the signal-to-noise ratio of 0d B.Aiming at the noise reduction of cable flashover discharge sound signals at fault points,a single channel discharge sound enhancement algorithm is proposed by combining the minimum mean square error logarithmic spectrum estimation algorithm(MMSE-LSA)and the masking effect of human ears.The algorithm can obtain the optimal logarithmic spectrum estimation under the MMSE criterion.Reuse is the probability of target sound masking noise spectral component further adjust spectrum estimation,remove residual noise,using global signal-to-noise ratio and weighted band segmented signal-to-noise ratio the algorithm comparing with spectral subtraction,MMSE-LSA algorithm,simulation results show that the proposed sound enhancement algorithm can effectively eliminate the background noise,get a clear discharge sound.Finally,in order to verify the effectiveness of the cable fault point flashover discharge sound signal detection and noise reduction algorithms presented in this thesis,these algorithms in this thesis are instantiated on a fixed-point DSP processor-based fixed-point instrument.Explained in detail the processing flow of the discharge sound signal data in the fixed-point DSP and the fixed-point realization of the detection and noise reduction algorithm.Finally,carry out fault fixed-point experiments at different cable fault sites.The algorithm in this thesis can effectively detect the discharge sound and remove the background noise.Able to obtain a stable acousto-magnetic time difference,hear a clear discharge sound,and accurately locate the location of the cable fault.