Research On Welding Structure Health Monitoring Using Artificial Neural Network And Piezoelectric Impedance Technology

Piezoelectric impedance technology is a new method used to detecting partial damage based onglobal vibration which is growing up in the recent ten years and it has many incomparable advantagesthat traditional nondestructive testing methods don’t have. So this technology has a good applicationprospect in structural health monitoring area. Artificial neural network as a information processingmathematical model imitating biological neural network show the huge application potential in healthmonitoring area which has developed to intelligent stage.In this paper, combined two methods and used in welding structure health monitoring research.According to theoretical analysis we build a experimental platform to do experimental analysis andlocate the welding crack accurately by the genetic neural network. The main contents are as follows:1. The structural mechanical impedance and electromechanical coupling property of PZT isanalyzed, and the coupling admittance of PZT is deduced in detail. The theoretical basis ofpiezoelectric impedance technology applied to structural health monitoring is obtained.2. A welding aluminum plate with several PZT sensors pasted on the surface is used asexperimental object. Collecting and analysis admittance to make qualitative judgment for thedamage of structure, proving the piezoelectric impedance technology is a useful method ofstructure health monitoring.3. Using genetic algorithm to optimize the traditional BP neural network, and design a new geneticneural network. By using this genetic neural network successfully locate the welding crack.Comparing the location effect of BP neural network constructed by MATLAB toolbox, thegenetic neural network show big advantage.4. Design a small impedance test system using a high integrated chip AD5933as the core part. Thissystem is small, light and chip, can measure impedance instead of traditional impedanceanalyzer.