Thin plate structures are one of the most fundamental components in the field of engineering, with applications spanning across building engineering, aerospace engineering, and marine engineering, among others. The structural safety of thin plates is closely related to internal damage and boundary service conditions. Currently, extensive theoretical and experimental research has been conducted on the structural damage detection of thin plates, yielding a series of results. However, the majority of methods rely on fixed, discrete damage parameters as identification indicators. Therefore, traditional methods lose their applicability for continuous damage in structures (such as spatially distributed degradation of material parameters). In recent years, with the rapid development of computer technology, physics-informed neural networks have been proven to solve many complex partial differential equation problems, especially excelling in solving inverse problems.