Electron Cyclotron Emission (ECE) diagnostic serve as a fundamental tool for measuring the plasma electron temperature distribution in Tokamak. Under complex plasma scenarios, Synthetic ECE simulations are employed to assist in the interpretation of diagnostic signals; however, these simulations are computationally intensive. A machine learning-based surrogate model for Synthetic ECE is proposed to enable rapid prediction of electron temperature distribution. Validation results demonstrate that the surrogate model achieves approximately an order-of-magnitude speedup in predicting ECE signals under challenging operating conditions compared to conventional Synthetic ECE simulations, thereby effectively meeting the requirements for real-time signal analysis in Tokamak control systems.

CNN, Machine Learning, Synthetic-ECE