Fatigue damage characterization in asphalt mixtures using acoustic emission with information entropy and the maximum Lyapunov exponent

Abstract Traditional non-destructive evaluation (NDE) methods often suffer from insufficient sensitivity in detecting early-stage microcracks and lack real-time monitoring capability. To overcome these limitations, this study conducted four-point bending fatigue tests on notched beam specimens under loading frequencies of 10 Hz and 15 Hz. During testing, the primary acoustic emission (AE) parameters—amplitude, energy, rise time, and duration—were systematically analyzed. In addition, information entropy was introduced to characterize signal randomness, while the maximum Lyapunov exponent (MLE) was employed to evaluate the chaotic characteristics of crack evolution, thereby revealing the staged damage evolution process. The results showed that amplitude was superior to other parameters in tracking fatigue damage, with its coefficient of variation (CV) remaining below 20%. Its evolution clearly corresponded to the four physical processes: pore compaction (Stage I), microcrack initiation (Stage II), stable propagation of macrocracks (Stage III), and catastrophic failure (Stage IV). Amplitude entropy exhibited distinct fluctuations across different stages, while MLE analysis further revealed the inherent disorder of crack evolution, with its variation validating the characteristics of each damage stage. Based on the coupled entropy–MLE framework, predictive thresholds (entropy > 10, MLE > 0.15, amplitude CV > 10) were proposed. Compared with conventional approaches, this framework enabled earlier identification of metastable crack propagation and effectively reduced false alarm rates. These findings demonstrate that the synergy of entropy and chaos analysis holds significant potential for NDE applications, offering new perspectives and methods for data-driven decision-making in pavement engineering.