The sophistication and intelligence of industrial machinery are continually evolving, making prognostics and health management (PHM) increasingly essential. Among these, the prediction of the remaining useful life (RUL) of equipment has been the subject of extensive research. For existing data-driven methods, there are numerous model parameters, insufficient accuracy in predictions, and a lack of interpretability in RUL prediction. In this article, we propose a lightweight method that integrates the BiMamba framework with a Physics-Informed Neural Network (PINN) framework. BiMamba enhances the performance advantages of large-scale data processing by leveraging its linear memory complexity and efficient long-sequence processing capabilities. At the same time, the introduction of the PINN physical framework incorporates physical laws, thereby enhancing both the accuracy and interpretability of the predictive model. The proposed method is validated on the C-MAPSS dataset, with results demonstrating significant improvement in RUL prediction accuracy while requiring few trainable parameters, offering significant support for intelligent maintenance and reliability management of industrial equipment.