Failure prediction and reliability analysis of ferrocement composite structures by incorporating machine learning into acoustic emission monitoring technique

•A damage index was proposed by AE signal energy.•AE signal data was utilized to propose a hazard rate function.•Support vector regression was used to produce a prediction tool of failure by using bathtub curve data. This paper introduces suitable features and methods to define hazard rate function by acoustic emission (AE) parametric analysis to develop robust damage statement index and reliability analysis. AE signal energy was first examined to find out the relation between damage progress and AE signal energy so that a damage index based on AE signal energy could be proposed to quantify progressive damage imposed to ferrocement composite slabs. Moreover, by using AE signal strength, historic index could be computed and utilized to develop a modified hazard rate function through integration of bathtub curve and Weibull function. Furthermore, to provide a practical scheme for real condition monitoring, support vector regression was utilized to produce a robust tools for failure prediction considering uncertainties exist in real structures.