As the real-time digital counterpart of a physical system or process, digital twins are utilized for system simulation and optimization. Neural networks are one way to build a digital twins model by using data especially when a physics-based model is not accurate or even not available. However, for a newly designed system, it takes time to accumulate enough data for neural network moded and only an approximate physics-based model is available. To take advantage of both models, this paper proposed a model that combines the physics-based model and the neural network model to improve the prediction accuracy for the whole life cycle of a system. The proposed model was able to automatically combine the models and boost their prediction performance. Experiments showed that the proposed hybrid model outperformed both the physics-based model and the neural network model.