 |
International Society of Science and Applied Technologies |
| Robust Parameter Design of Multiple Functional Responses Based on Transfer Learning and Its Application in Additive Manufacturing | ||||
| Author | Mingyue Chu
|
|||
| Co-Author(s) | Jianjun Wang
|
|||
| Abstract | Robust parameter design (RPD) plays an important role in the quality control of additive manufacturing (AM) processes. However, with the advancement of sensor technology, the quality characteristics of AM parts have evolved from traditional independent scalar values to high-dimensional curves, also referred to as functional responses. Additionally, only limited data is available for each product due to the relatively high cost of AM. These factors pose significant challenges in building accurate predictive models for the AM process. To address these issues, this article proposes a transfer learning method based on the tensor regression algorithm, which can build predictive models and optimize parameters for multiple functional quality responses of AM products under data-scarce conditions. The performance of the proposed method is evaluated using real data from Jiangsu Province High-end Equipment Quality Improvement Engineering Center. The experimental results demonstrate that the proposed method can more accurately model and optimize high-dimensional quality characteristics, even with limited training data.
|
|||
| Keywords | multiple functional responses, robust parameter design, transfer learning, additive manufacturing | |||
| Article #: RQD2025-359 | ||||
Proceedings of 30th ISSAT International Conference on Reliability & Quality in Design |