Abstract: This paper proposed an intelligent detection model for foreign objects in railway catenary based on neural network fusion model, aimed at addressing the issue of foreign objects affecting the normal operation of railway catenary. Based on the Faster R-CNN framework, the paper added a feature pyramid structure to learn features of images at different scales, divided different types of foreign objects into bird nests and lightweight floating objects, and used ResNet50 and ResNet101 as skeleton networks to identify bird nests with a single feature and lightweight floating objects with complex and variable features, respectively, and integrated the recognition boxes of two networks to obtain accurate recognition results. Comparative experiments show that the detection results of this model are superior to conventional object detection methods, which can effectively reduce the labor cost of foreign object detection in railway catenary and provide a feasible solution for the stable operation of railway catenary.