Research On Pressure Vessel Wireless Acoustic Emission Monitoring Technology And Instrument Design

The pressure vessel has been widely used in petrochemical industry which has therisk of explosion. When the accident occurred, it will bring disastrous consequences. Inorder to solve the actual difficulties of wired acoustic emission detection on pressurevessel and achieve long-term structural health monitoring, a wireless acoustic emissioninstrument was developed on the basis of existing AE acquisition card. The main workincludes the development of modules such as the hardware circuit, the Wi-Fi wirelesstransmission, GPS clock synchronization, CF card, etc. In addition, the acquisition andwireless transmission of AE waveform and parameters also wireless location of theacoustic emission source were realized for the first time.Because the suspected defect of on-line pressure vessel detected by AE method cannot be retested in most cases, the identification and judgment of acoustic emissionsource are particularly necessary. Therefore, this paper puts forward extraction methodsof four new feature parameters which were used to express the characteristics ofacoustic emission source. The four parameters are the time domain waveformcharacteristic parameters based on the wavelet packet decomposition and reconstruction,the relative energy proportion of each band based on wavelet packet decomposition,peak value of each band based on FFT and mean value of each band based on FFT. Bycomparison, the following conclusions can be drawn.(1) For the four new extractioncharacteristic parameters, the effects of their neural network pattern recognition arebetter than the traditional parameter method in various circumstances (no noise, lightnoise, strong noise).(2) In more serious noise, the advantage of new feature parametersidentification is more obvious. The last two methods have the highest accuracy rate.(3)In strong noise, recognition accuracy rate of all characteristic parameters decreasedseriously, not exceed65%.In order to improve the wireless transmission efficiency of data and the wirelesssignal monitoring and recognition accuracy, the compression and dimensionalityreduction technique have been used to deal with the above five kinds of parameters (35groups) including traditional characteristic parameters in this paper. The mostrepresentative four groups of the filter results are obtained as follows:(1) energy peakvalue of the FFT frequency band (150-200kHz),(2) wavelet packet energy ratio of each band (125-156.25kHz),(3) energy mean value of the FFT band (150-200kHz),(4)thecharacteristic parameters of wavelet packet decomposition and reconstruction(amplitude). Then, the four parameters above are regarded as the optimal featureparameters for identification of standard acoustic emission signals of broken lead.Under the disturbance of strong noise, pencil percussion, bamboo percussion, metalpercussion and gravel impact, wireless acoustic emission monitoring of broken leadbehavior on pressure vessel has been conducted. The results show that using the fourgroup characteristic parameters to identify will get high accuracy rate (90.9%). Withless number of parameters, the purpose of high efficiency wireless transmission and thehigh accuracy wireless monitoring has been achieved.