• Laser & Optoelectronics Progress
  • Vol. 59, Issue 23, 2306003 (2022)
Hongquan Qu1, Zhengyi Wang1,*, Zhiyong Sheng1, Hongbin Qu2, and Ling Wang3
Author Affiliations
  • 1School of Information Science and Technology, North China University of Technology, Beijing 100144, China
  • 2International Business Department, China Petroleum Pipeline Bureau Engineering Co., Ltd., Langfang 065000, Hebei, China
  • 3Asia Pacific Branch of China Petroleum Pipeline Bureau Engineering Co., Ltd., Langfang 065000, Hebei, China
  • show less
    DOI: 10.3788/LOP202259.2306003 Cite this Article Set citation alerts
    Hongquan Qu, Zhengyi Wang, Zhiyong Sheng, Hongbin Qu, Ling Wang. Fiber Intrusion Signal Classification Based on Gradient Boosting Decision Tree Algorithm[J]. Laser & Optoelectronics Progress, 2022, 59(23): 2306003 Copy Citation Text show less
    Original signal of tapping and the decomposition result. (a) Original signal; (b) decomposition result
    Fig. 1. Original signal of tapping and the decomposition result. (a) Original signal; (b) decomposition result
    Original fiber intrusion signal and preprocessing results. (a) Original signal; (b) preprocessing result
    Fig. 2. Original fiber intrusion signal and preprocessing results. (a) Original signal; (b) preprocessing result
    Flowchart of the AdaBoost algorithm
    Fig. 3. Flowchart of the AdaBoost algorithm
    Validation curves of GBDT algorithm under different parameters. (a) n_estimators; (b) learning_rate; (c) max_depth; (d) subsample
    Fig. 4. Validation curves of GBDT algorithm under different parameters. (a) n_estimators; (b) learning_rate; (c) max_depth; (d) subsample
    AdaBoost algorithm grid search validation curve. (a) n_estimators; (b) learning_rate
    Fig. 5. AdaBoost algorithm grid search validation curve. (a) n_estimators; (b) learning_rate
    Grid search verification curve of the SVM algorithm. (a) C; (b) gamma
    Fig. 6. Grid search verification curve of the SVM algorithm. (a) C; (b) gamma
    Confusion matrices for different algorithms. (a) GBDT algorithm; (b) DT-AdaBoost algorithm; (c) SVM algorithm
    Fig. 7. Confusion matrices for different algorithms. (a) GBDT algorithm; (b) DT-AdaBoost algorithm; (c) SVM algorithm
    Recognition rate of fiber intrusion signals for different algorithms
    Fig. 8. Recognition rate of fiber intrusion signals for different algorithms
    FIBFPermutation entropy
    FIBF 10.0000
    FIBF 20.2255
    FIBF 30.2838
    FIBF 40.3253
    FIBF 70.5812
    FIBF 80.7130
    FIBF 160.4579
    FIBF 170.2181
    Table 1. Permutation entropy value of some FIBF components
    ParameterValue range
    n_estimators1-50
    learning_rate0.1-0.3
    max_depth3-10
    subsample0.5-0.8
    Table 2. Tuning range of GBDT parameters
    Hongquan Qu, Zhengyi Wang, Zhiyong Sheng, Hongbin Qu, Ling Wang. Fiber Intrusion Signal Classification Based on Gradient Boosting Decision Tree Algorithm[J]. Laser & Optoelectronics Progress, 2022, 59(23): 2306003
    Download Citation