Interferometric Synthetic Aperture Radar (InSAR) is extensively used for monitoring and analyzing surface deformation, as well as sudden natural disasters, playing a crucial role in disaster prevention and mitigation. The Phase Linking (PL) method enables the estimation of interference phase from distributed scatterers (DS) in InSAR samples, thereby enhancing the accuracy and quality of measurement results. However, the PL process involves intricate data processing, consumes significant computing resources and time, and poses challenges for large-scale operations. This paper proposes an estimation model based on the Temporal Convolution Network (TCN) architecture, which leverages both temporal and spatial information from interferograms to estimate the interference phase of DS pixels rapidly. The proposed model was applied to an open-pit mining area in Inner Mongolia using a dataset comprising 31 Sentinel-1A satellite images captured over the area. The results demonstrate that this model can significantly reduce mathematical operations involved in the PL process while mitigating noise in InSAR interferograms. In future work, with continuous refinement of the model, it is expected to achieve the goal of quickly and accurately estimating the phase information of DS pixels in a wide range of research areas.