Research on Product Structure Identification and Design Optimization of Smart Furniture Based on ResNet Network

Abstract

Aiming at the problems of insufficient accuracy in structure identification and low efficiency in design optimization for smart furniture design, this paper proposes an integrated method for structure identification and design optimization based on an improved residual network. First, multi-source heterogeneous furniture datasets are constructed, and an improved model integrating multi-scale structure perception and a spatial topology joint attention mechanism is designed to achieve efficient representation and relationship modeling of complex structural features. On this basis, the parametric structure representation and multi-objective optimization framework are introduced, and the automatic generation and optimization of design schemes are realized by combining the generative model. The experimental results show that the accuracy of the proposed method in the structure recognition task is 0.956, and the IoU is 0.903, which is about 3.1% higher than that of the benchmark model on average; In terms of design optimization, the maximum structural displacement is reduced by 39.7%, the material utilization rate is increased by 19.1%, and the design efficiency is improved by about 45.2%. In addition, the number of model parameters is reduced by 26.2%, and the reasoning time is reduced by 31.5%, which verifies the good balance between accuracy and efficiency. The results show that this method can effectively improve the recognition ability and design optimization performance of complex furniture structure, and provide a feasible and efficient technical path for intelligent furniture design automation.