Signal processing techniques for filtering acoustic emission data in prestressed concrete

The current state of infrastructure in the United States and worldwide has raised the need for reliable structural health monitoring techniques. Piezoelectric sensing, such as acoustic emission, has recently gained attention due to its high sensitivity and associated capability for early detection of damage. The high sensitivity of this method, however, results in the collection of data not directly related to damage growth. Current filtering procedures focus primarily on parametric analysis of the collected signals. This study focuses on developing more robust filtering techniques for acoustic emission data collected from a prestressed concrete specimen. Simulated data was generated to enable proper identification of the source of the collected signals. Filtering criteria were developed through characterization of the energy content using a wavelet transform. The developed filters were capable of separating the induced target signals from other signals with reasonable accuracy, and the results were verified through source location. The developed filters were validated using acoustic emission data collected during a load test.