Figures (6)  Tables (8)

    • Figure 1. 

      Technical route.

    • Figure 2. 

      Trends in the average values of the four core pavement performance indicators from 2015 to 2024. (a) PQI; (b) PCI; (c) RQI; (d) RDI.

    • Figure 3. 

      Detailed methodological workflow of the proposed framework.

    • Figure 4. 

      Pavement performance prediction model based on stacking fusion.

    • Figure 5. 

      Evolution trends of pavement performance indices (2016–2033).

    • Figure 6. 

      Asphalt pavement distress maintenance measures based on knowledge graphs.

    • Inspection year Direction Start stake End stake Road age PCI RQI RDI PQI Passenger-to-freight ratio Maintenance history code
      2020 Upbound K747 + 671 K748 + 671 17 80.76 95.59 94.65 90.56 1.68 0
      2021 Upbound K747 + 671 K748 + 671 18 83.60 93.48 92.44 90.59 1.68 3
      2023 Upbound K747 + 671 K748 + 671 20 92.78 94.45 95.89 93.78 1.68 3
      2024 Upbound K747 + 671 K748 + 671 21 88.19 93.97 94.17 92.19 1.68 3
      2020 Downbound K735 + 671 K736 + 671 17 91.08 93.49 92.89 93.28 1.68 0
      2021 Downbound K735 + 671 K736 + 671 18 82.59 92.24 92.83 89.92 1.68 3
      2023 Downbound K735 + 671 K736 + 671 20 91.04 93.79 95.93 92.89 1.68 3
      2024 Downbound K735 + 671 K736 + 671 21 86.82 93.02 94.38 91.41 1.68 3
      The passenger-to-freight ratio is reported as the annual traffic composition of the LH section. Maintenance history was encoded from 0 to 4 according to intervention type and intensity, where 0 denotes no maintenance, 1 denotes routine minor maintenance, 2 denotes milling-related treatment, 3 denotes repaving or overlay treatment, and 4 denotes major or structural rehabilitation.

      Table 1. 

      Example of processed section-level records used in the case study.

    • Index value rangeMaintenance type
      PCIRQIRDISRI
      ≥ 90≥ 90≥ 80< 75PM
      < 80−RM
      85−90−PM
      < 85−RM
      85−90≥ 85−PM
      < 85−RM
      < 85−RM

      Table 2. 

      Grading standard of pavement performance indicators for maintenance type classification.

    • Influence factor Assignment rules
      Performance index PCI The value is assigned according to the predicted value of each road segment index, and the smaller value is ranked first
      RQI
      RDI
      SRI
      Elasticity factor K The value is assigned according to the nature of the road section, and the higher value is ranked first
      Traffic volume
      (103 vehicles/d)      		 T The value is assigned according to the daily traffic volume, and the higher value is ranked first
      Road age (year) A The value is assigned according to the length of traffic time, and the highest value is ranked first
      Interval years (years) Y The value is assigned according to the interval between the current maintenance year and the last major and medium-sized maintenance, and the higher value is ranked first

      Table 3. 

      Priority ranking principle of pavement maintenance.

    • Weight a b c d
      Value range 0.35−0.45 0.25−0.35 0.15−0.25 0.10−0.20

      Table 4. 

      Index weight range.

    • ModelIndicatorPQIRQIPCIRDIAverage
      XGBoostR20.89910.81130.96770.54650.8062
      RMSE1.13090.73801.84930.94131.1649
      MAE0.86940.50401.42650.66140.8653
      RFR20.81050.69400.81110.77810.7734
      RMSE1.54210.93494.44940.65511.8954
      MAE1.19300.53853.53670.46791.4340
      OursR20.91010.91120.93140.89080.9109
      RMSE1.07150.50332.70750.45961.1855
      MAE0.82200.36022.10380.34570.9079

      Table 5. 

      Comparison of experimental results.

    • Cluster No. Core features Key indicators (average) Maintenance type ratio Distress distribution
      1 High-performance low-distress group PCI 93.60, RQI 92.85,
      RDI 95.45      		 RM 1.3%, PM 98.7% 63.6% of sections have no distress; main distress is transverse crack (21.3%)
      2 Medium-performance low-distress group PCI 87.40, RQI 87.37,
      RDI 90.57      		 RM 59.0%, PM 41.0% 39.2% of sections have no distress; main distress is longitudinal crack (39.9%)
      3 Medium-performance medium-distress group PCI 84.65, RQI 93.53,
      RDI 93.96      		 RM 43.8%, PM 56.2% Main distresses are longitudinal cracks (43.8%) and transverse cracks (56.2%); complex distress combinations accounts for 78.8%

      Table 6. 

      Core characteristics of OPTICS clustering results.

    • Road ID PCI RQI RDI SRI RSPF Maintenance
      type      		 Priority
      ranking
      G5-213.1 85.35 90.25 73.77 94.17 64.38 PM 1
      G5-204.1 86.30 85.46 93.49 94.17 67.22 PM 2
      G5-228.3 85.31 86.44 94.83 94.17 67.43 PM 3
      G5-775.4 87.93 85.58 92.22 94.17 67.57 PM 4
      G5-771.6 86.40 85.13 96.13 94.17 67.68 PM 5
      G5-745.8 86.33 89.48 91.15 94.17 67.97 PM 6
      G5-195.3 87.02 86.62 94.26 94.17 67.97 PM 7
      G5-755 85.93 90.18 91.02 94.17 68.01 PM 8
      G5-756.5 85.10 89.59 94.21 94.17 68.18 PM 9
      G5-275.7 86.63 87.44 95.02 94.17 68.23 PM 10
      G5-229.4 82.40 46.64 93.81 94.17 54.27 RM 1
      G5-756.7 83.71 54.94 85.62 94.17 55.58 RM 2
      G5-256.8 49.13 91.96 96.71 94.17 56.80 RM 3
      G5-203.1 48.68 95.89 98.04 94.17 58.09 RM 4
      G5-267.3 68.77 85.15 79.32 94.17 58.16 RM 5
      G5-215.9 55.92 89.71 95.43 94.17 58.25 RM 6
      G5-752.1 65.52 83.68 91.04 94.17 58.92 RM 7
      G5-203 50.89 96.18 98.02 94.17 58.95 RM 8
      G5-268.6 56.57 92.62 94.36 94.17 59.14 RM 9
      G5-5.8 53.25 95.34 97.15 94.17 59.35 RM 10

      Table 7. 

      Top 10 priority sections for PM and RM categories.

    • Scenario Total budget
      (10,000 CN¥)      		 Covered
      sections
      (PM + RM)      		 Average
      PCI after
      maintenance      		 Fund utilization
      rate
      Scenario 1 25 13 91.06 99.6%
      Scenario 2 35 16 97.30 99.2%
      Scenario 3 45 20 100.00 84.4%
      Traditional method 35 12 94.53 98.1%

      Table 8. 

      Comparison of fund allocation results under different scenarios.