Abstract: The modular and intelligent design of gun family is the state of the art. Based on the design characteristics of gun family,a knowledge-based computer-aided design process is established, and a modularization-based intelligent design method is proposed. An unified object-oriented knowledge representation is adopted for different types of knowledge in the gun family design, and the strategies based on knowledge template, machine learning, and mixed reasoning are used in the solution of the gun family definition stage and the module design stage. Specifically, the knowledge template of module division is used to guide the modular design of the gun family, and the Random-Forest-based regression algorithm is used to estimate the main parameters of the gun family. A question-answering system based on the bidirectional encoder representation from transformers (BERT) model and a mixed reasoning mode based on case-based reasoning and cross reasoning of design rules and calculation are used to assist in completing the detailed design of each module. Accordingly, a software prototype system is constructed, and 5.8 mm/7.62 mm modular gun family is designed. The results have proved the effectiveness of the proposed methods.

Key words: modulargunfamily, intelligentcomputer-aideddesign, randomforest, case-basedreasoning, bidirectionalencoderrepresentationfromtransformers, question-answeringsystem