The rock-breaking efficiency of disc cutters is significantly influenced by complex geological conditions and mechanical parameters in tunnel excavation using tunnel boring machines (TBM). The accurate prediction of this efficiency under varying geological conditions is imperative for TBM. A prediction method was proposed for the rock-breaking efficiency of TBM disc cutters in jointed rock masses through the integration of a support vector machine (SVM) with intelligent optimization algorithms. Specifically, particle swarm optimization (PSO), moth-flame optimization (MFO), and black-winged kite algorithm (BKA) were utilized to optimize the penalty factor and kernel function parameters of the SVM. Seven parameters, including confining pressure, joint spacing, joint dip angle, rock strength, penetration depth, cutterhead thrust, and rolling force were selected as inputs and specific energy was defined as output for the prediction model. A dataset was constructed based on full-scale linear cutting tests and discrete element numerical simulations for model training and testing. The results show that the BKA-SVM model demonstrates superior performance in terms of stability, computational efficiency, and prediction accuracy when compared to the PSO-SVM and MFO-SVM models. Furthermore, the sensitivity analysis of the predicted data reveals that rock strength and penetration depth are the most influential factors, with an optimal penetration depth for rock breaking identified as 3 mm. Additionally, a positive correlation has been observed between rock strength and specific energy, with a more pronounced increasing trend. The research results provide valuable insights for the optimization of rock-breaking efficiency during TBM excavation with potential applications for enhancing operational efficiency and reducing energy consumption.