Image enhancement and stress separation of thermoelastically measured data under random loading

Traditional thermoelastic stress analysis (TSA) presupposes that the structure being analyzed is cyclically loaded at a constant amplitude and frequency. This approach typically has been used to satisfy the adiabatic reversible assumptions. We employ an alternative signal analysis technique that enables one to evaluate the magnitude of the individual components of stress in a component subjected to a loading that is random in both frequency and magnitude. However, the nature of the measured information does not change; i.e., data are inherently noisy, and edge information is unreliable. The latter two aspects have caused many thermoelastic stress analyses to be more qualitative than quantitative. We emphasize the development of the TSA technique into a practical, noncontacting quantitative method for stress analyzing actual engineering structures that are randomly loaded. The ability to determine the individual stresses thermoelastically under random loading is demonstrated. (Author)