• Acta Photonica Sinica
  • Vol. 50, Issue 9, 0930004 (2021)
Meihui JIA, Lijuan LI, Jiaojiao REN, Jian GU, Dandan ZHANG, Jiyang ZHANG, and Weihua XIONG
Author Affiliations
  • Key Laboratory of Optoelectronic Measurement and Control and Optical Information Transmission Technology of Ministry of Education, College of Optoelectronic Engineering, Changchun University of Science and Technology, Changchun130022, China
  • show less
    DOI: 10.3788/gzxb20215009.0930004 Cite this Article
    Meihui JIA, Lijuan LI, Jiaojiao REN, Jian GU, Dandan ZHANG, Jiyang ZHANG, Weihua XIONG. Terahertz Nondestructive Testing Signal Recognition Based on PSO-BP Neural Network[J]. Acta Photonica Sinica, 2021, 50(9): 0930004 Copy Citation Text show less
    Defect sample drawing of high temperature resistant composite material with multi-adhesive structure
    Fig. 1. Defect sample drawing of high temperature resistant composite material with multi-adhesive structure
    Terahertz time domain waveforms of different defect areas
    Fig. 2. Terahertz time domain waveforms of different defect areas
    Operating principle diagram of BP neural network
    Fig. 3. Operating principle diagram of BP neural network
    Flow chart of optimization process
    Fig. 4. Flow chart of optimization process
    Schematic diagram of high temperature resistant composite bonding sample
    Fig. 5. Schematic diagram of high temperature resistant composite bonding sample
    Schematic diagram of reflective THZ-TDS
    Fig. 6. Schematic diagram of reflective THZ-TDS
    Comparison of mean square errors of training results between BP neural network and PSO-BP neural network.
    Fig. 7. Comparison of mean square errors of training results between BP neural network and PSO-BP neural network.
    Defect recognition results of BP neural network.
    Fig. 8. Defect recognition results of BP neural network.
    Defect recognition results of PSO-BP neural network
    Fig. 9. Defect recognition results of PSO-BP neural network
    Recognition results of different degree of debonding defects by PSO-BP neural network
    Fig. 10. Recognition results of different degree of debonding defects by PSO-BP neural network
    Error comparison of PSO-BP neural network in recognizing defects of different degrees
    Fig. 11. Error comparison of PSO-BP neural network in recognizing defects of different degrees
    Serial numberCharacteristics of the nameCharacteristic expression
    1KurtosisKmin=E[(Xi-i=1nXi/n)4]{E[(Xi-i=1nXi/n)2]}2
    2SkewnessSmin=E[(Xi-i=1nXi/n)3]{E[(Xi-i=1nXi/n)2]}3/2
    3Waveform factorXpp-up=i=1nXi2/ni=1nXi/n
    4Absolute mean of amplitudeXmean=i=1nXi/n
    5In peak valueXpv=Xma-Xmi
    6Minimum amplitude valueXimin
    Table 1. Time domain characteristics and their expressions
    The serial numberUpper defect levelLower defect level
    1150 μm350 μm
    2100 μmNo defect
    3No defect250 μm
    4No defect100 μm
    Table 2. Defects of the sample
    Meihui JIA, Lijuan LI, Jiaojiao REN, Jian GU, Dandan ZHANG, Jiyang ZHANG, Weihua XIONG. Terahertz Nondestructive Testing Signal Recognition Based on PSO-BP Neural Network[J]. Acta Photonica Sinica, 2021, 50(9): 0930004
    Download Citation