Yiran Liu
Research Subjects
Yiran Liu has a total of 13 co-authors. Here are the top co-authors by number of shared publications:
- Jin Li4 shared publications
- Bo Feng
Huazhong University of Science and Technology
2 shared publications Hongbao Ma
2 shared publications- Yanting Chen
School of Mechanical Science and Engineering, Huazhong University of Science and Technology, Wuhan, China
2 shared publications Yihua Kang
2 shared publications- Hongbao Ma1 shared publication
Latest Publications
- Open Access
- 8 September 2026
- Published by Elsevier in SSRN Electronic Journal
Speed-Robust Eddy-Current Inspection of Thickly Insulated Coupler Cables Using Spatially Matched Dual Receivers
Broken-strand inspection of electric multiple unit (EMU) coupler cables is hindered by approximately 8-mm insulation, large lift-off, confined access, and variable manual scanning speeds. This study proposes a single-excitation integrated dual-receiver eddy-current method that combines parametric excitation design, spatially differentiated sensing, and statistically calibrated decisions. A second-order vector potential model determines excitation-coil geometry and operating parameters from impedance sensitivity, skin depth, and portable-instrument constraints. To distinguish the localized high-gradient disturbance of broken strands from the broader disturbance of shielding wires, a narrow racetrack receiver (S1) and a circular receiver (S2) use asymmetric differential reception to suppress common backgrounds. The receivers share one excitation and move synchronously. After encoder-based spatial registration, their amplitude ratio suppresses correlated speed-dependent drift while retaining disturbance-specific spatial responses. Configuration-dependent thresholds are calibrated through 50 repetitions of stratified five-fold cross-validation and defined from the distribution of fold-specific thresholds. Experiments on 40 broken-wire and 10 shielding-wire specimens produced 250 responses at five speeds from 10 to 50 mm/s. The ratio reduced the mean coefficient of variation from 48.028% to 9.754%, a 79.691% reduction in speed sensitivity, and increased the Fisher separability index from 1.519 to 56.334 (37.09-fold). Broken-wire and shielding-wire disturbances were also correctly identified in a five-year-old service cable. The method therefore provides speed robustness, disturbance-source separability, and practical feasibility for the specified cable-probe configuration, offering a calibratable approach to portable inspection of thickly insulated coupler cables.
- 8 September 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessSpeed-Robust Eddy-Current Inspection of Thickly Insulated Coupler Cables Using Spatially Matched Dual Receivers
Broken-strand inspection of electric multiple unit (EMU) coupler cables is hindered by approximately 8-mm insulation, large lift-off, confined access, and variable manual scanning speeds. This study proposes a single-excitation integrated dual-receiver eddy-current method that combines parametric excitation design, spatially differentiated sensing, and statistically calibrated decisions. A second-order vector potential model determines excitation-coil geometry and operating parameters from impedance sensitivity, skin depth, and portable-instrument constraints. To distinguish the localized high-gradient disturbance of broken strands from the broader disturbance of shielding wires, a narrow racetrack receiver (S1) and a circular receiver (S2) use asymmetric differential reception to suppress common backgrounds. The receivers share one excitation and move synchronously. After encoder-based spatial registration, their amplitude ratio suppresses correlated speed-dependent drift while retaining disturbance-specific spatial responses. Configuration-dependent thresholds are calibrated through 50 repetitions of stratified five-fold cross-validation and defined from the distribution of fold-specific thresholds. Experiments on 40 broken-wire and 10 shielding-wire specimens produced 250 responses at five speeds from 10 to 50 mm/s. The ratio reduced the mean coefficient of variation from 48.028% to 9.754%, a 79.691% reduction in speed sensitivity, and increased the Fisher separability index from 1.519 to 56.334 (37.09-fold). Broken-wire and shielding-wire disturbances were also correctly identified in a five-year-old service cable. The method therefore provides speed robustness, disturbance-source separability, and practical feasibility for the specified cable-probe configuration, offering a calibratable approach to portable inspection of thickly insulated coupler cables.
- Open Access
- 24 June 2026
- Published by Elsevier in SSRN Electronic Journal
Eddy current detection of broken wires in EMU coupler cables under large lift-off with identification of shielded strand interference
Detecting broken wires in electric multiple unit (EMU) coupler cables in-situ is challenging due to large lift off (8 mm insulation), strong interference from a metallic shielding mesh, and poor field adaptability. This paper proposes a collaborative eddy current testing (ECT) method addressing excitation, reception, and signal identification. A vertical excitation strategy induces current crowding at the broken wire cross section. A narrow racetrack shaped asymmetric differential coil cancels common mode interference while avoiding signal cancellation, effectively overcoming lift off attenuation. A dual module ratio based method suppresses scanning speed variations and robustly distinguishes broken wires from the shielding mesh. Experiments show reliable detection of minor broken wires (≈25 % cross sectional loss) beneath 8 mm insulation, with a broken wire SNR ≥ 43.79 dB. Ratio based features for broken wires and the shielding mesh fall in 5–12.2 and ≤ 3, respectively, yielding a clear classification boundary. Repeatability tests on 16 artificial cables and progressive validation on an in service cable (intact → peeled → broken wire) confirm engineering potential, offering a compact, practical solution for in service health monitoring of coupler cables.
- 24 June 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessEddy current detection of broken wires in EMU coupler cables under large lift-off with identification of shielded strand interference
Detecting broken wires in electric multiple unit (EMU) coupler cables in-situ is challenging due to large lift off (8 mm insulation), strong interference from a metallic shielding mesh, and poor field adaptability. This paper proposes a collaborative eddy current testing (ECT) method addressing excitation, reception, and signal identification. A vertical excitation strategy induces current crowding at the broken wire cross section. A narrow racetrack shaped asymmetric differential coil cancels common mode interference while avoiding signal cancellation, effectively overcoming lift off attenuation. A dual module ratio based method suppresses scanning speed variations and robustly distinguishes broken wires from the shielding mesh. Experiments show reliable detection of minor broken wires (≈25 % cross sectional loss) beneath 8 mm insulation, with a broken wire SNR ≥ 43.79 dB. Ratio based features for broken wires and the shielding mesh fall in 5–12.2 and ≤ 3, respectively, yielding a clear classification boundary. Repeatability tests on 16 artificial cables and progressive validation on an in service cable (intact → peeled → broken wire) confirm engineering potential, offering a compact, practical solution for in service health monitoring of coupler cables.
- Open Access
- 24 June 2026
- Published by Elsevier in SSRN Electronic Journal
Eddy current detection of broken wires in EMU coupler cables under large lift-off with identification of shielded strand interference
Detecting broken wires in electric multiple unit (EMU) coupler cables in-situ is challenging due to large lift off (8 mm insulation), strong interference from a metallic shielding mesh, and poor field adaptability. This paper proposes a collaborative eddy current testing (ECT) method addressing excitation, reception, and signal identification. A vertical excitation strategy induces current crowding at the broken wire cross section. A narrow racetrack shaped asymmetric differential coil cancels common mode interference while avoiding signal cancellation, effectively overcoming lift off attenuation. A dual module ratio based method suppresses scanning speed variations and robustly distinguishes broken wires from the shielding mesh. Experiments show reliable detection of minor broken wires (≈25 % cross sectional loss) beneath 8 mm insulation, with a broken wire SNR ≥ 43.79 dB. Ratio based features for broken wires and the shielding mesh fall in 5–12.2 and ≤ 3, respectively, yielding a clear classification boundary. Repeatability tests on 16 artificial cables and progressive validation on an in service cable (intact → peeled → broken wire) confirm engineering potential, offering a compact, practical solution for in service health monitoring of coupler cables.
- 24 June 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessEddy current detection of broken wires in EMU coupler cables under large lift-off with identification of shielded strand interference
Detecting broken wires in electric multiple unit (EMU) coupler cables in-situ is challenging due to large lift off (8 mm insulation), strong interference from a metallic shielding mesh, and poor field adaptability. This paper proposes a collaborative eddy current testing (ECT) method addressing excitation, reception, and signal identification. A vertical excitation strategy induces current crowding at the broken wire cross section. A narrow racetrack shaped asymmetric differential coil cancels common mode interference while avoiding signal cancellation, effectively overcoming lift off attenuation. A dual module ratio based method suppresses scanning speed variations and robustly distinguishes broken wires from the shielding mesh. Experiments show reliable detection of minor broken wires (≈25 % cross sectional loss) beneath 8 mm insulation, with a broken wire SNR ≥ 43.79 dB. Ratio based features for broken wires and the shielding mesh fall in 5–12.2 and ≤ 3, respectively, yielding a clear classification boundary. Repeatability tests on 16 artificial cables and progressive validation on an in service cable (intact → peeled → broken wire) confirm engineering potential, offering a compact, practical solution for in service health monitoring of coupler cables.
- Open Access
- 11 May 2026
- Published by Elsevier in SSRN Electronic Journal
Binocular Visual Measurement for Live-Line Working Robots via Instance Mask Prediction and Contour-Constrained Threshold Segmentation
This paper proposes a binocular vision‑based visual measurement system that integrates deep learning to enhance the measurement accuracy and environmental robustness of live‑line working robots operating in complex distribution networks. The system addresses key challenges including environmental interference, lack of depth perception, and insufficient measurement precision for small targets. First, the non‑maximum suppression (NMS) algorithm of Mask R‑CNN is improved by incorporating diagonal angle discrepancy as a weighting factor for the intersection over union (IoU), effectively reducing missed detections of densely arranged cables. Second, a contour‑constrained local threshold segmentation method is developed to achieve pixel‑level segmentation accuracy. Third, a contour‑tracing based feature extraction strategy is designed to realize pixel‑level localization of cable centerlines and hook keypoints. Experimental results demonstrate that the binocular measurement error of the proposed system is less than 0.87 mm, and the overall measurement uncertainty for distribution network components is within 4 mm. Equipped with this system, the live‑line working robot achieves a success rate of 92% in 100 threading trials and 95% in 50 arrester replacement trials. In summary, the proposed method achieves millimeter‑level measurement accuracy in complex distribution network scenarios, providing an effective solution for enhancing the autonomous operation capability of live‑line working robots.
- 11 May 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessBinocular Visual Measurement for Live-Line Working Robots via Instance Mask Prediction and Contour-Constrained Threshold Segmentation
This paper proposes a binocular vision‑based visual measurement system that integrates deep learning to enhance the measurement accuracy and environmental robustness of live‑line working robots operating in complex distribution networks. The system addresses key challenges including environmental interference, lack of depth perception, and insufficient measurement precision for small targets. First, the non‑maximum suppression (NMS) algorithm of Mask R‑CNN is improved by incorporating diagonal angle discrepancy as a weighting factor for the intersection over union (IoU), effectively reducing missed detections of densely arranged cables. Second, a contour‑constrained local threshold segmentation method is developed to achieve pixel‑level segmentation accuracy. Third, a contour‑tracing based feature extraction strategy is designed to realize pixel‑level localization of cable centerlines and hook keypoints. Experimental results demonstrate that the binocular measurement error of the proposed system is less than 0.87 mm, and the overall measurement uncertainty for distribution network components is within 4 mm. Equipped with this system, the live‑line working robot achieves a success rate of 92% in 100 threading trials and 95% in 50 arrester replacement trials. In summary, the proposed method achieves millimeter‑level measurement accuracy in complex distribution network scenarios, providing an effective solution for enhancing the autonomous operation capability of live‑line working robots.