Research on Concrete Failure Warning Based on Acoustic Emission Non-Gaussian Statistics and Tsallis Entropy

To reveal the nonlinear dynamic characteristics and early warning indicators of crack propagation during the flexural loading process of concrete, this paper investigates notched concrete specimens. Acoustic emission (AE) technology is employed to monitor the dynamic crack propagation behavior during flexural loading, and a statistical model characterizing crack evolution is developed within the framework of non-extensive statistical mechanics. Non-Gaussian statistical analysis of the AE time series is conducted based on Tsallis entropy and the entropy index q. The “variation of return” of the acoustic emission impact count, which exhibits transient fluctuation characteristics, was selected as the analytical parameter, and a q-Gaussian distribution was used in place of the traditional Gaussian distribution to describe the statistical discipline of crack propagation. The results indicate that the q-Gaussian distribution provides a better fit to the experimental probability density function (with an adjusted Rq-G² as high as 0.985) than the classical Gaussian distribution (0.752), with only a slight deviation in the extreme value region at the tail of the distribution; a distinct characteristic peak appears in the temporal evolution of the entropy index q, corresponding to the critical stage where the structure transitions from stable damage accumulation to unstable fracture. The study indicates that characterization methods based on the non-Gaussian statistical properties of acoustic emission can effectively describe the nonlinear dynamic characteristics of the concrete fracture process, and that the characteristic peak of the entropy index q has the potential to serve as an early warning indicator of macroscopic failure.