Research on optimization of additive manufacturing tool path based on improved neural network
To address the internal quality problem of selective laser sintering(SLS),this study proposes a Bayesian optimization back-propagation(BO-BP)neural network path optimization model for improving the SLS molding quality and reducing the energy consumption in the printing process.First,a path-planning process is established for printed copies,and dynamic programming technology is used to obtain the final path in a given laser grid and remove redundant paths.Subsequently,an SLS model is established for the laser sintering process based on the thermal conductivity differential equation,and the path for this process is numerically simulated.The final temperature values for each grid node are then used to calculate the temperature gradient values of internal nodes in the material grid and rank them.This value is used as the dataset for training and testing the neural network algorithm.Furthermore,the corresponding average thermal gradient distribution map is obtained through the numerical image transformation of the laser path,and the laser path is subsequently converted into a grayscale path image,which is trained using the improved BO-BP neural network algorithm.Finally,the Softmax function is used to convert the neural network output into the probability of each image.Linear optimization,unsupervised learning neural network,genetic BP neural network,and BP neural network are selected to compare the prediction results.The results show that the proposed BO-BP neural network algorithm has the advantages of high prediction accuracy and high search speed compared with other algorithms.
万方数据
维普
