• Laser & Optoelectronics Progress
  • Vol. 56, Issue 14, 140602 (2019)
Zhiyong Sheng, Zhiqiang Zeng*, Hongquan Qu, and Wei Li
Author Affiliations
  • School of Electronic Information Engineering, North China University of Technology, Beijing 100144, China
  • show less
    DOI: 10.3788/LOP56.140602 Cite this Article Set citation alerts
    Zhiyong Sheng, Zhiqiang Zeng, Hongquan Qu, Wei Li. Fiber Intrusion Signal Recognition Algorithm Based on Stochastic Configuration Network[J]. Laser & Optoelectronics Progress, 2019, 56(14): 140602 Copy Citation Text show less
    Principle of optical fiber pre-warning system
    Fig. 1. Principle of optical fiber pre-warning system
    Standard SCN model
    Fig. 2. Standard SCN model
    Dropout-SCN model: standard multi-input and multi-output network model is presented
    Fig. 3. Dropout-SCN model: standard multi-input and multi-output network model is presented
    Original signals. (a) Digging signal; (b) knocking signal; (c) electric drill signal
    Fig. 4. Original signals. (a) Digging signal; (b) knocking signal; (c) electric drill signal
    Scatter plot of output weight distributions of SCN, Dropout-SCN and SCN with L2 regularization
    Fig. 5. Scatter plot of output weight distributions of SCN, Dropout-SCN and SCN with L2 regularization
    Effect of Dropout probability on test error under number of hidden nodes L=70
    Fig. 6. Effect of Dropout probability on test error under number of hidden nodes L=70
    Test RMSE for three network models
    Fig. 7. Test RMSE for three network models
    Network modelHidden node numberCorrect prediction numberAccuracy /%
    SCN5021489.17
    SCN with L2 regularization5021790.42
    Dropout-SCN5022292.50
    SCN7021790.42
    SCN with L2 regularization7022192.08
    Dropout-SCN7022794.58
    Table 1. Results of optical fiber test data
    Zhiyong Sheng, Zhiqiang Zeng, Hongquan Qu, Wei Li. Fiber Intrusion Signal Recognition Algorithm Based on Stochastic Configuration Network[J]. Laser & Optoelectronics Progress, 2019, 56(14): 140602
    Download Citation