Elimination Methods Of Intensive Noise In Industrial Field For On-line Sonic Detection

Sonic testing technology is a non-destructive testing method that canidentify inner defects of workpieces quickly, and thorough research has been madeon its mechanism in recent30years, but strong environment noise has been a seriousissue which badly blocks its widely use in industry production. Intensiveenvironment noise made sonic testing technology unreliable, since misjudging caneasily happen when active impulse and frequency spectrum analysis are used. Tosolve this problem, this paper presented a semi-enclosed sound cover used in on-linedetection and used the combination of microphone and vibration transducer toacquire workpieces’ signals based on the differences of noise in differentcommunication channels, and then through the variable-step Least Mean Square(LMS) algorithm and the natural gradient blind separation algorithm to eliminatenoise and improve resolution of traditional sonic testing. Experiments showed thatthe algorithms can efficiently eliminate noise in workpieces’ signals.Following are the contents of this paper:Firstly, in-depth research on the sonic testing technology and noiseelimination technology at home and abroad was made, noise sources inindustrial field and the coupled modes between noise and sensor wereanalyzed, the signal acquisition method combining microphone and vibrationtransducer was researched, and the hardware de-noising through semi-enclosedsound cover designing, sensor selection and installing layout was realized.Secondly, Frequency-domain Block algorithm was researched, and the best stepfunction was analyzed. Simulation is used to verify its de-noising effects. Based onthe analysis of several algorithms’ characteristics and application limitations, theimproved algorithm of natural gradient blind separation and its best step functionwere analyzed, and the simulation was used to verify its de-noising effects.Lastly, various experiments were made to verify the de-noising effects, and theresults showed that the de-noising method could reduce the intensive noise in industry field by20dB, it also could eliminate the high frequency noise in specificenvironment and it could reduce the fault rate of sonic detection from10%to1%.