Predictive modelling of acoustic emission signal data for corrosion assessment: A modified dimensional analysis based approach

Acoustic emission (AE) technique has been proved as a powerful technique for structural health monitoring. AE technique can also be efficiently used for evaluation of corrosion activity in concrete. Despite its advantages, an effective analysis of data recorded by AE technique still founds to be a challenging task demanding an appropriate damage assessment methodology. For quantification of damages, various methods for analysis of AE data have been proposed but not a single method has found to be standardized for specific application. In this paper, a procedure for analysis of AE signals is proposed using modified dimensional analysis method. Many times, it becomes difficult to choose the appropriate AE parameter which can be effectively co-related to the physical feature for development of accurate prediction model. Hence, in the present research work an attempt has been made to develop a model by incorporating primary characteristic AE waveform parameters. A corrosion rate prediction model using modified dimensional analysis of AE signals is developed and compared with the model developed using non-linear regression analysis. The performance of two models is further assessed using different statistical parameters. The study demonstrated that the methodology of modified dimensional analysis indicated improvement in the corrosion rate predictions. Thus, modified dimensional analysis can be implemented as a promising method for analysis of complex AE signal data as well as for development of statistical modelling of corrosion phenomenon in reinforced concrete based on recorded AE parameters.