Audio Signal Enhancement Technology For Nonstationary Background Noise

With the rapid development of urban construction,high altitude cables are gradually replaced by buried cables.Although buried cables have the advantages of beautifying the urban environment,improving the value of land use and enhancing the reliability of the urban power grid,with the social and economic growth,the cable load increases,the working cycle increases,and the fault is easy to occur.And the buried cable has a large span of space,when cable fails,it is very difficult to find fault points.If the failure point cannot be repaired in time,it will cause great inconvenience to people's lives and huge losses in social economy.Therefore,locating the cable fault point quickly and accurately is of urgent social significance.This thesis is based on the current status and difficulties of cable fault detection at home and abroad,in order to study the filtering and enhancement of the collected signals in the accurate positioning of the cable fault,to extract a useful cable fault point shock discharge sound signal,so as to locate the fault point accurately.The main tasks of the thesis include the following parts,First of all,the development of cable fault detection and sound signal enhancemen t technology at home and abroad are studied,and the difficulty of the cable fault detection is the accurate positioning of the fault point is clarified.However,the difficulty of precise positioning is the extraction of the impulse discharge signal from the fault point in the mixed signal.In view of the problem that the traditional filtering method based on statistical theory is difficult to remove the random noise of non-stationary,spectrum distribution on the whole frequency axis,and overlapped with the spectral distribution of the effective shock discharge sound signal in sound signals,an independent component analysis(ICA)impact discharge sound signal enhancement algorithm based on principal component analysis(PCA)is proposed to detect sound signals effectively.The algorithm is simulated in MATLAB.It is proved that it can suppress noise well,and can effectively separate impulse discharge sound signal from mixed signal in low SNR environment.Then,in view of the problem that the discharge sound of the cable fault point is not obvious,and the impulse discharge sound signals detected by the cable fault detection fixed-point instrument are completely submerged by the noi se under the complex background noise detection environment,an impulse discharge enhancement algorithm based on wavelet packet analysis and genetic optimization is proposed.It takes advantage of the characteristics of the wavelet packet transform can detect whether the normal signal contains a transient anomaly,the adaptive filter needs less for prior knowledge of the signal and noise,and genetic algorithm does not depend on the specific problems to extract the sound signals submerged in the background noise.The simulation in MATLAB proves the effectiveness of the algorithm,and the algorithm can accurately extract the impulse discharge sound signal.Finally,based on the ADSP-BF525 processor,a shock discharge sound signal detection system for the accurate location of the cable fault points is built.It consists of magnetic field and audio signal acquisition module,amplification filter module,LCD display module,data processing module,data communication and output module,etc.And through the shock discharge sound signal of the fault point enhancement experiment in the actual project,it is verified that the system can effectively extract the shock discharge sound signal in the low SNR environment,so then locate the fault location quickly and accurately.