Research Article
Automated Detection of Microseismic Arrival Based on Convolutional Neural Networks
Table 1
Comparison of precision before and after noise adding.
| Method | Error range point | Proportion (%) | Error pickup rate (%) | Average error (μs) | Raw data | Added noise | Raw data | Added noise | Raw data | Added noise |
| AIC | 0–5 | 91.96 | 63.02 | 2.19 | 9.51 | 2.79 | 12.16 | 5–10 | 1.99 | 6.47 | 10–50 | 3.86 | 21.01 | 50+ | 2.19 | 9.51 | U-net | 0–5 | 78.97 | 73.92 | 3.34 | 4.80 | 6.38 | 11.64 | 5–10 | 7.65 | 7.70 | 10–50 | 10.03 | 13.58 | 50+ | 3.34 | 4.80 | Time window | 0–5 | 47.20 | 41.00 | 11.51 | 28.55 | 12.63 | 31.07 | 5–10 | 15.63 | 8.73 | 10–50 | 25.66 | 21.72 | 50+ | 11.51 | 28.55 |
|
|