A Dynamic Monitoring Method of Temperature Distribution for Cable Joints Based on Thermal Knowledge and Conditional Generative Adversarial Network

The dynamic monitoring of the temperature distribution of power equipment is crucial for safe operation. The monitoring method based on digital twin technology has received extensive attention due to its ability to provide more timely and more comprehensive analysis. However, the existing methods only use physics-driven or data-driven methods separately, which cannot simultaneously meet application requirements such as low cost, high efficiency, effective results, and sufficient training data. Therefore, a new dynamic monitoring method of temperature distribution based on thermal knowledge and conditional generative adversarial network (CGAN) is proposed in this article, with a high-voltage cable joint taken as an example. First, the steady-state numerical model of the temperature field is established. The thermal image set under different operation conditions is obtained through simulation. The data-driven method based on the CGAN model is applied to learn the data laws between operation conditions and thermal images and constructs a mapping relationship between them. Then, the physics-driven method based on thermal knowledge is used to reduce feature dimensions and extract input features from real-time measurement data. These input features drive the generator of the CGAN model to generate the thermal images dynamically. By comparing the generation results with numerical and experimental results, the proposed method can achieve the dynamic monitoring of temperature for cable joints, which has comprehensive advantages with regard to the time efficiency, computational cost, monitoring accuracy, and generalization ability. It provides new insights into the digital twin technology of power equipment.