Generative adaptable design based on hidden Markov model

Kaifeng Wang

doi:10.1016/j.aei.2024.103034

2025-03-01

PRE

Kaifeng Wang

Jianye Li

Zhilin Sun *

DOI：10.1016/j.aei.2024.103034

原文链接

原文求助

摘要

En 中文

With the advent of next-generation information technologies and artificial intelligence, the paradigm of product design is shifting towards a more data-driven approach. To facilitate the reuse of historical product data, designers require systematic and structured methods for data analysis and processing to enhance design efficiency. The Hidden Markov Model (HMM), a sophisticated statistical model adept at analyzing and forecasting sequential data patterns, is a powerful tool enabling designers to process and utilize historical product data effectively. This paper proposes a generative adaptable design method based on HMM to assist designers in rapidly generating design solutions. The proposed method utilizes HMM to model the structural modules and their connections within historical products, enabling designers to efficiently reuse existing product modules to generate diverse design configurations optimized to user requirements. Concurrently, a three-dimensional sparse matrix is used to mathematically represent the connection relations among module interfaces, facilitating the transformation of HMM-derived into design solutions with rational connections. The entropy method of product adaptability is utilized to evaluate the design solutions and identify the optimal solution. A case study of the design process for the Hybrid Additive Manufacturing device is conducted to verify the effectiveness of this method. The results indicate that the proposed method effectively assists designers in generating multiple design solutions and determining the optimal design outcome through the adaptability evaluation, thereby enhancing the overall efficiency of the design process.

Keyword:

Data-driven

Adaptable design

Hidden Markov model

Product design