Figure 1.
Asphalt pavement design.
Figure 1.
Asphalt pavement design.
Figure 2.
Workflow diagram in ABAQUS.
Figure 2.
Workflow diagram in ABAQUS.
Figure 3.
Diagram illustrating machine learning models, depicting input and output variables.
Figure 3.
Diagram illustrating machine learning models, depicting input and output variables.
Figure 4.
RF regression model.
Figure 4.
RF regression model.
Figure 5.
Tree expansion in LGBM.
Figure 5.
Tree expansion in LGBM.
Figure 6.
SHAP Attributes.
Figure 6.
SHAP Attributes.
Figure 7.
Illustration of field test CCTV.
Figure 7.
Illustration of field test CCTV.
Figure 8.
Video analysis flowchart methodology.
Figure 8.
Video analysis flowchart methodology.
Figure 9.
Rutting at 100 km/h.
Figure 9.
Rutting at 100 km/h.
Figure 10.
Scenario 1: With speed at 100 km/h (a) temp vs. rutting at 100,000 action times, (b) temp vs. rutting at 200,000 action times, (c) temp vs. rutting at 300,000 action times, (d) temp vs. rutting at 400,000 action times, and (e) temp vs. rutting at 500,000 action times.
Figure 10.
Scenario 1: With speed at 100 km/h (a) temp vs. rutting at 100,000 action times, (b) temp vs. rutting at 200,000 action times, (c) temp vs. rutting at 300,000 action times, (d) temp vs. rutting at 400,000 action times, and (e) temp vs. rutting at 500,000 action times.
Figure 11.
Rutting at 80 km/h.
Figure 11.
Rutting at 80 km/h.
Figure 12.
Scenario 2: With speed at 80 km/h (a) temp vs. rutting at 100,000 action times, (b) temp vs. rutting at 200,000 action times, (c) temp vs. rutting at 300,000 action times, (d) temp vs. rutting at 400,000 action times, and (e) temp vs. rutting at 500,000 action times.
Figure 12.
Scenario 2: With speed at 80 km/h (a) temp vs. rutting at 100,000 action times, (b) temp vs. rutting at 200,000 action times, (c) temp vs. rutting at 300,000 action times, (d) temp vs. rutting at 400,000 action times, and (e) temp vs. rutting at 500,000 action times.
Figure 13.
Rutting at 60 km/h.
Figure 13.
Rutting at 60 km/h.
Figure 14.
Scenario 3: With speed at 60 km/h (a) temp vs. rutting at 100,000 action times, (b) temp vs. rutting at 200,000 action times, (c) temp vs. rutting at 300,000 action times, (d) temp vs. rutting at 400,000 action times, and (e) temp vs. rutting at 500,000 action times.
Figure 14.
Scenario 3: With speed at 60 km/h (a) temp vs. rutting at 100,000 action times, (b) temp vs. rutting at 200,000 action times, (c) temp vs. rutting at 300,000 action times, (d) temp vs. rutting at 400,000 action times, and (e) temp vs. rutting at 500,000 action times.
Figure 15.
Analysis of (a) AdaBoost model, (b) LGBM model, (c) RF model, (d) LR model.
Figure 15.
Analysis of (a) AdaBoost model, (b) LGBM model, (c) RF model, (d) LR model.
Figure 16.
Global Factor Interpretation; (a) factor importance plot; (b) factor bee swarm plot.
Figure 16.
Global Factor Interpretation; (a) factor importance plot; (b) factor bee swarm plot.
Figure 17.
Single factor analysis: (a) effect of temperature, (b) effect of number of load, (c) effect of number of wheels.
Figure 17.
Single factor analysis: (a) effect of temperature, (b) effect of number of load, (c) effect of number of wheels.
Figure 18.
Interaction analysis: (a) effect of temperature and number of loads, (b) effect of number of wheels and temperature, (c) effect of number of wheels and number of loads.
Figure 18.
Interaction analysis: (a) effect of temperature and number of loads, (b) effect of number of wheels and temperature, (c) effect of number of wheels and number of loads.
Figure 19.
The distribution of wheels track of small cars in (a) driving lane, (b) passing lane.
Figure 19.
The distribution of wheels track of small cars in (a) driving lane, (b) passing lane.
Figure 20.
Wheel track distribution of medium sized truck in (a) driving lane, (b) passing lane.
Figure 20.
Wheel track distribution of medium sized truck in (a) driving lane, (b) passing lane.
Figure 21.
Wheel track distribution of large buses in (a) driving lane, (b) passing lane.
Figure 21.
Wheel track distribution of large buses in (a) driving lane, (b) passing lane.
Figure 22.
Wheel track distribution of large trucks in (a) driving lane, (b) passing lane.
Figure 22.
Wheel track distribution of large trucks in (a) driving lane, (b) passing lane.
Figure 23.
Wheel track distribution of trailers in (a) driving lane, (b) passing lane.
Figure 23.
Wheel track distribution of trailers in (a) driving lane, (b) passing lane.
Table 1.
Material properties.
Table 1.
Material properties.
Parameter | Asphalt | Cement Stabilized Base | Lime Soil Subbase | Soil Foundation |
---|
Thermal conductivity (λ) | 4680 | 5616 | 5148 | 5616 |
Density (ρ) | 2300 | 2200 | 2100 | 1800 |
Heat capacity (J/K) | 924.9 | 911.7 | 942.9 | 1040.0 |
Solar absorption rate | 0.9 |
Road surface emissivity | 0.81 |
Absolute zero (°C) | −273 |
Stefan–Boltz constant (σ) | 2.041092 × 10−4 |
Table 2.
Input Parameters.
Table 2.
Input Parameters.
No. | Input Parameter | Value |
---|
1 | Number of wheel loads, times | 1 |
2 | Vehicle axle load, kN | 100 |
3 | Wheel number of the shaft, pcs | 4 |
4 | Tire ground pressure, MPa | 0.7 |
5 | Tire contact width, cm | 18.6 |
6 | Travel speed, km/h | 100, 80, 60 |
Table 3.
Model Scenario Development.
Table 3.
Model Scenario Development.
Scenario | Description |
---|
Scenario 1 | With speed limit 100 km/h and 100,000 to 500,000 action times with 0 °C to 50 °C temperature range |
Scenario 2 | With speed limit 80 km/h and 100,000 to 500,000 action times with 0 °C to 50 °C temperature range |
Scenario 3 | With Speed limit 60 km/h and 100,000 to 500,000 action times with 0 °C to 50 °C temperature range |
Table 4.
Range of factors.
Table 4.
Range of factors.
Factors | Data Type | Range |
---|
Temperature (T) | Discrete | 0, 10, 20, 30, 40, 50 |
Speed (v) | Discrete | 60, 80, 100 |
No. of Load (N) | Discrete | 100,000, 200,000, 300,000, 400,000, 500,000 |
No. of wheels (nw) | Discrete | 4, 8, 12 |
Ground pressure (p) | Discrete | 0.7, 0.9, 1.1 |
Table 5.
Observation data of different type of vehicles from the video.
Table 5.
Observation data of different type of vehicles from the video.
Model | Small | Medium | Large Bus | Large Truck | Trailer | Total |
---|
Driving lane | 4022 | 3633 | 899 | 700 | 435 | 9689 |
Passing lane | 4877 | 3477 | 468 | 134 | 89 | 9045 |
Table 6.
ML algorithm with hyperparameters, Range and optimal values.
Table 6.
ML algorithm with hyperparameters, Range and optimal values.
Models | Hyperparameters | Range | Optimal Values |
---|
LGBM | {(learning rate), (n_estimators), (num_leaves), (reg_lambda), (reg_alpha} | {(0.001–0.20), (200–2000), (30–100), (1.10–1.50), (1.10–1.50)} | {(0.15), (1179), (47), (1.18), (1.33)} |
AdaBoost | {(learning rate), (n_estimators)} | {(0.001–0.20), (200–2000)} | (0.07), (350) |
RF | {(n_estimators), (max_depth)} | {(200–2000), (2–16)} | (940), (11) |
LR | {(learning rate)} | {(0.001–0.20)} | (0.05) |
Table 7.
Performance of training and testing dataset using ML.
Table 7.
Performance of training and testing dataset using ML.
Models | Training Dataset | Testing Dataset |
---|
MAE | MSE | RMSE | R2 | MAE | MSE | RMSE | R2 |
---|
AdaBoost | 1.695 | 4.512 | 2.124 | 0.881 | 1.489 | 3.566 | 1.888 | 0.911 |
LGBM | 0.335 | 0.481 | 0.694 | 0.987 | 0.460 | 0.903 | 0.950 | 0.977 |
RF | 1.274 | 5.326 | 2.307 | 0.859 | 1.091 | 4.247 | 2.061 | 0.894 |
LR | 1.315 | 4.420 | 2.102 | 0.882 | 1.275 | 4.211 | 2.05 | 0.894 |