Abstract: A tool vibration acceleration prediction model based on PSO-BP network has been proposed to address the issue of cutting vibration affecting the quality and tool life of high-speed cutting of titanium alloys. High-speed cutting experiments were conducted on titanium alloy Ti-6Al-4V, with tool vibration acceleration as the research target. A Python-based back propagation (BP) neural network was constructed to predict vibration patterns, with its accuracy further improved by incorporating the particle swarm optimization (PSO) algorithm. The results demonstrate that the PSO-BP hybrid model reduces the prediction error from 10% (BP alone) to below 6%, while increasing the coefficient of determination (R²) by 30.4%. Additionally, the mean squared error (MSE) and mean absolute percentage error (MAPE) are reduced by 3% and 3.64%, respectively. The optimized model exhibits superior predictive performance, providing robust data-driven support for parameter optimization in titanium alloy machining processes.