A Deep Learning Method for Material Performance Recognition in Laser Additive Manufacturing

With the development of intelligent manufacturing and "Industry 4.0", the traditional methods of material mechanical properties evaluation cannot meet the needs of industrial production due to the shortcomings of wasted materials, tedious processes, and poor accuracy. This paper combines artificial intelligence technology to propose a new material performance evaluation method. The laser additive manufacturing is taken as the research background, three kinds of Ti6-Al-4V material microstructure images with different properties are used as data sets, based on DenseNet model, a deep convolution neural network NDenseNet is trained to optimize the network model memory and improve the recognition accuracy. The experimental results show that the accuracy of the model reaches 90.4%, loss value remains at 25%. Params and FLOPs are significantly reduced compared with DenseNet model. It only takes 0.1 seconds to process a microstructural image on a GPU processor. This method can greatly reduce the work of researchers, improve product development efficiency in industrial environment, reduce human errors, save production materials, and has guiding significance for the development of high-performance materials.