As the manufacturing process of a transformer needs to go through many kinds of processes, the good or bad manufacturing process directly affects the performance level of the transformer. In this paper, the green manufacturing process chain of transformers is established by comprehensively considering resource and energy consumption, as well as environmental impact. Based on the existing technical architecture of the manufacturing workshop, a digital twin technology architecture system is proposed for the manufacturing process. Starting from the three dimensions of the key elements of the twin model, entity modelling, digital twin virtual modelling, and virtual-reality mapping association modelling, a unified logical data model is established, and the design of the smart factory of a transformer manufacturing enterprise is completed. Experiments show that in the new working conditions based on digital twin technology, the accuracy error in transformer production is less than 0.02mm, and the production order delay rate has significantly improved. The manufacturing process optimisation module has better processing parameter monitoring and processing process simulation performance and also realises the constant power adaptive control of the processing process, which helps to improve the processing efficiency and processing accuracy of the manufacturing process.