Study of Electromechanical Impedance Quantitative Evaluation for Pipeline Structure Based on Neural Network Technique
基于神经网络技术的管道机电阻抗健康状况定量评估研究

Electromechanical impedance method was adopted to monitor the health condition of flange and main body of pipeline structure, furthermore, BP neural network was used to evaluate structure damage quantitatively. In experiment, the influence of health condition of flange and pipeline structure main body on the impedance spectra was studied firstly, then, the variation of real parts of impedance spectra obtained from impedance analyzer can characterize the different kinds of structure damage. BP neural network technique was used to quantitatively analyze the real parts of impedance spectrum under pipeline structure different working conditions. The real part of impedance spectra were selected as input samples to train developed neural network, finally, the trained neural network can achieve the quantitative evaluation of different kinds of damage in pipeline structure. Results show that the combination of electromechanical impedance method and neural network data processing technique for structural health monitoring of complex pipeline structure not only can effectively achieve quantitative evaluation of different kinds of defect but also have high stability.