Abstract: In order to effectively control the quality of laser milling layer, BP neural network model was established between the laser milling quality of layer(the milling layer width, milling depth) and the milling layer parameters(laser power, scanning speed and defocus amount). Using particle swarm optimization (PSO)BP neural network weights and thresholds, quality prediction model based on particle swarm neural network was built. The proposed PSO-BP algorithm solve the problem that the general BP algorithm iteration speed was slow, and prone to local optimum. Al2O3 ceramics laser milling quality prediction model was taben to realize the algorithm.The simulation results show that the number of iterations of proposed PSO-BP algorithm, and the prediction error are greatly reduced. The built quality prediction model has high prediction accuracy and practical value.