Comparative Evaluation of Outlier Detection Methods for the Asphalt Fracture Tests

What is it about?

In this study focusing on asphalt mixture testing, we explored four different outlier detection methods applied to the results of the Disk Shaped Compact Tension (DCT) test, a standardized procedure for assessing asphalt’s crack resistance. The methods evaluated included those based on the ASTM E178 standard and the Mahalanobis Distance technique, among others. The study aimed to identify the most effective approach for handling variations and anomalies in test data. Among the methods tested, the Robust Estimator (RE) method stood out for its superior performance. This method proved particularly robust, showing a high degree of reliability in detecting outliers without being overly sensitive to their presence. It performed consistently well across datasets with both normal and non-normal distributions, making it an excellent choice for enhancing the accuracy and dependability of asphalt cracking evaluations.

Featured Image