Research On Welding Structure Health Monitoring Using Piezoelectric Impedance Technology

With the development of science and technology, as well as people's growing demand and increasing consciousness of safety, it is important to make health assessment of engineering structures in service effectively and conveniently. Welding is an important connection method in current engineering structures; however, fracture accident is occurred frequently in use. Therefore, health monitoring of welded structure has a broad background in engineering applications. Piezoelectric impedance technology is a brand-new method for damage detection, in this paper the PZTs are pasted as sensing elements on the surface of structure being detected and the electrical impedance was extracted when high frequency excitation is applied. The damage in structure can be identified by analysis of impedance spectrum.The basic principle of piezoelectric impedance technology is present in this paper, then experimental analysis and localization study based on artificial neural network of welded aluminium alloy sheet with crack damage is carried out.1. The structural mechanical impedance and electromechanical coupling property of PZT is analyzed, and the coupling admittance of PZT is deduced in detail. The theoretical basis of piezoelectric impedance technology applied to structural health monitoring is obtained.2. Finite element model of welded structure with PZT attached is simulated using ANSYS software, and harmonic response analysis of piezoelectric impedance is studied when PZT is away from damage at different distances. The results showed that: piezoelectric impedance is sensitive not only to the appearance of micro damage, but also to the variation of PZT position.3. Experimental board is welded by two aluminium alloy sheets with crack fabricated and an experimental platform is built. Layout scheme of PZTs is designed and possibility of damage localization using piezoelectric impedance technology is proposed after the analysis of admittance spectrum of PZT in different frequency bands or at different locations.4. BP neural network designed in MATLAB environment is applied to the localization of welding damage after reasonable selection and organization of samples, and the crack location is predicted successfully.