Table 5 RUL prediction performance
From: Deep Spatiotemporal Convolutional-Neural-Network-Based Remaining Useful Life Estimation of Bearings
Testing bearing | Actual RUL (s) | RUL (s) / (Er (%)) | ||||||
|---|---|---|---|---|---|---|---|---|
LSTM (2019) | MSCNN (2019) | DBN-DP (2019) | RNN-AM (2020) | Hybrid (2020) | Encoder–decoder (2021) | Proposed method | ||
1–3 | 5730 | 400/(93.02) | 4731/(17.43) | 5600/(2.27) | 5294/(7.61) | 5440/(5.06) | 3650/(36.30) | 5630/(1.75) |
1–4 | 339 | −310/(191.18) | 259/(10.67) | 260/(23.53) | 874/(−157.81) | 260/(23.30) | 2500/(−637.46) | 350/(−2.94) |
1–5 | 1610 | −6490/(503.11) | 3996/(−148.19) | 2950/(−83.23) | 2778/(−72.55) | 1540/(4.35) | 1570/(2.48) | 1660/(−3.11) |
1–6 | 1460 | −6460/(542.47) | 1744/(−19.45) | 1163/(20.34) | 1446/(0.96) | 1450/(0.68) | 1640/(−12.33) | 1530/(−4.79) |
1–7 | 7570 | 7310/(3.43) | 6090/(19.55) | 5834/(22.93) | 1061/(85.98) | 10790/(−42.53) | 3210/(57.60) | 7510/(0.79) |
3–3 | 820 | 990/(−20.73) | 1060/(−29.27) | 544/(33.66) | 803/(2.07) | 850/(−3.66) | 770/(6.10) | 790/(3.66) |
SD of Er/% | Â | 247.36 | 52.97 | 43.58 | 83.24 | 21.83 | 268.80 | 3.31 |
Score | Â | 0.16 | 0.30 | 0.44 | 0.45 | 0.62 | 0.39 | 0.77 |