An Integrated Processing Method for Fatigue Damage Identification in a Steel Structure Based on Acoustic Emission Signals
This paper presents an integrated processing method that applies principal component analysis (PCA), artificial neural network (ANN), information entropy and information fusion technique to analyze acoustic emission signals for identifying fatigue damage in a steel structure. Firstly, PCA is used to build different data spaces based on the damage patterns. Input information from each sensor is diagnosed locally through ANN in the data space. The output of the ANNs is used for basic probability assignment. Secondly, the first fusion operation adopts Dempster-Shafer (D-S) evidence theory to combine the basic probability assignment value of ANNs in the different data space of a sensor. Finally, the fusion results of each sensor are combined by D-S evidence theory for the second fusion operation. In this paper, information entropy is used to calculate the uncertainty and construct basic probability assignment function. The damage identification method is verified through four-point bending fatigue tests of Q345 steel. Validation results show that the damage identification method can reduce the uncertainty of the system and has a certain extent of fault tolerance. Compared with ANN and ANN combined with information fusion methods, the proposed method shows a higher fatigue damage identification accuracy and is a potential for fatigue damage identification.
Keywordsacoustic emission construction damage identification steel
This work was financially supported by National Key Technology R&D Program of China (2015BAG20B04 and 2015BAF06B01-3), the National Key Research and Development Program of China (2016YFC0801903) and National Natural Science Foundation of China (Nos. 51175023 and U1537212).
- 2.M. Arrington, Acoustic Emission-A Review, Non-Destructive testing, 1989, p 429–434Google Scholar
- 13.D.D.L. Hall, S.A.H. McMullen, Mathematical Techniques in Multisensor Data Fusion 2nd ed., Artech House Publishers, Norwood, 2004Google Scholar
- 17.H. Wang, H. Luo, Z. Han, and Q. Zhong, Investigation of Damage Identification of 16Mn Steel Based on Artificial Neural Networks in Tensile Test, Reliability, Maintainability and Safety, 2009, 8th International Conference on ICRMS 2009, IEEE, 2009, p 1057–1061Google Scholar
- 18.H. Wang, H. Luo, Z. Han, and Q. Zhong, Investigation of Damage Identification of 16Mn Steel Based on Artificial Neural Networks and Data Fusion Techniques in Tensile Test, Advanced Data Mining and Applications, 2009, p 696–703Google Scholar