Determination of international roughness index with PVDF sensors based on tire-road interaction

In this thesis, the International Roughness Index (IRI), one of the comfort indicators of road pavement, is tried to be determined for the first time with a new method based on tire-road interaction. For this, a mountain bike is chosen as the test vehicle. Six polyvinylidene fluoride (PVDF) based piezoelectric sensors are mounted in the front wheel of this bike. The data of the PVDF sensors and the positions obtained by GPS are simultaneously recorded in the data acquisition system (VTS) designed on the test vehicle. The roughness information of the 660 m long dual carriageway in Bursa Uludağ University campus in both directions (1320 m) is measured with a laser profilometer device as the reference international roughness index (IRI) data. With the newly constructed test tool, the same road is tested 13 times at different times. Support Vector Regression (DVR) algorithm is used to estimate the IRI value of the test road with the obtained sensor data. With these sensor data, 32 feature data sets are extracted from each 10m, 30m, 60m, and 100m sections of the 1320 m long test road, and these feature datasets are used as input data in the DVR algorithm.The algorithm's performance is tested, with the prepared feature subset %85 as training data and 15% as test data. At the same time, the IRI values of the test road are tried to be estimated on the same road by using traditional methods such as a quarter vehicle model with a 3-axis accelerometer in the VTS. Thus, the performance of the traditional method and the new method are compared. The performance of both methods is evaluated with mean absolute percentile error (MAPE), mean absolute error (MAE), mean square error (MSE), and root mean square error (RMSE) metrics. According to this, For the traditional method, the best MAPE results are found to be 23.79, 17.58, 17.13, and 12.65 at the 10m, 30m, 60m, and 100m sections of the test road. The MAPE values measured according to the new method at the same road intervals are found to be 18.08, 13.64, 10.73, and 5.34, respectively. These results revealed that the proposed method gives satisfactory results in determining the roughness of the roads. Thus, this study offers a cheaper, easier, and alternative method to road authorities in determining the IRI values of highways compared to traditional methods and tools.