Abstract:Abstract: During the 3D printing process, the laser spot temperature will have a direct impact on the accuracy and quality of the molded product. Aiming at the industrial demand that traditional temperature detection devices cannot achieve high detection rate and low error rate in laser spot temperature detection, this paper proposes a method for detecting and estimating laser spot temperature based on convolutional neural networks. According to the deep learning technology, the collected laser temperature training samples are used for model training using convolutional neural networks, the training results are used to predict the test set, and the laser temperature of the test laser image is estimated. In addition, improvements have been made on the basis of the traditional convolutional neural network, and the verification results show that the improved convolutional neural network can detect and estimate the laser point isotherm more accurately.