The large diameter borehole pressure relief is one of the effective means used to prevent rock burst in coal mines. It is of great significance to study the change of coal body stress during drilling to prevent the rock burst. At present, the relational model between drilling parameters and coal stress is rarely studied, and the precision is limited. Therefore, a new PSO-NGO-SVR coal stress inversion model was proposed based on northern goshawk optimization (NGO) and support vector regression (SVR). Firstly, the Tent chaotic mapping was introduced into the model during the initialization phase of the NGO population, and the advantages of particle swarm optimization (PSO) were integrated into the northern hawk algorithm, resulting in better performance of the improved northern hawk algorithm. Next, the improved northern eagle algorithm was used to iteratively optimize hyperparameters in support vector regression. Finally, the model was established with the optimal hyperparameters after iteration. The results show that the convergence speed and accuracy of the improved northern eagle algorithm are greatly improved, and the PSO-NGO-SVR coal stress inversion model has relatively high accuracy.