In recent years, with the rapid development of augmented reality technology, its technical application fields have become more and more extensive. This paper studies the application of augmented reality technology in engineering projects, mainly including target detection technology, feature matching technology, and 3D modeling technology. The article first improves YOLOv5 based on the original YOLOv5 algorithm by integrating the attention mechanism. The SURF algorithm is improved, the FAST-SURF algorithm is adopted for feature extraction and matching of images, and an engineering project simulation system is designed using augmented reality technology. Then, the improved YOLOv5 algorithm was tested for performance, and the experiment concluded that the improved YOLOv5 algorithm has high precision and recall. For experimental analysis of image stabilization based on feature matching, the ORB model and FAST-SURF were used. The experimental results are in line with expectations, and FAST-SURF has a good virtual fusion effect for images. This paper takes the electromechanical engineering project as a case for simulation experiments and uses Monte Carlo method simulation to verify the reliability of the method, for which traditional technology and augmented reality technology were used to predict the project duration of the project engineering. The results show that the project duration prediction cycle based on augmented reality technology is shorter, the project plan is more feasible, and the probability of achieving the project duration with augmented reality technology is stable at about 40.