In order to prevent the occurrence of stuck drilling accidents during drilling, an improved sparrow search algorithm (ISSA) predictive model method (ISSA-SVM) to optimize support vector machines (SVM) was proposed, and an improved adaptive nonlinear declining inertia weight was introduced into the finder position update formula. Levy flight strategy was introduced into the alert position update formula. The principal component analysis method (PCA) was used to reduce the dimensionality of the measured drilling data of a large foreign oilfield, and the penalty parameters and nuclear parameters were used to predict the stuck drilling accident. The experimental results show that the prediction accuracy of ISSA-SVM is as high as 85.185 2%, and the convergence speed is faster, which shows that ISSA-SVM can effectively predict stuck drilling accidents.