Small sample area selection laser melting process parameter multi-objective optimization method and related equipment
By constructing relative density and crack density prediction models using Gaussian process regression and Bayesian optimization algorithms, the high cost and low efficiency problems in the optimization of selective laser melting process parameters are solved, achieving high-precision and rapid multi-objective optimization, which is applicable to selective laser melting processes with different equipment and materials.
Patent Information
- Authority / Receiving Office
- CN Β· China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- SHANGHAI ZHONGTIAN SCIENCE & TECHNOLOGY AEROSPACE TECHNOLOGY CO LTD
- Filing Date
- 2026-03-06
- Publication Date
- 2026-07-10
AI Technical Summary
Existing methods for optimizing selective laser melting process parameters suffer from high experimental costs, long cycles, and low efficiency. They also struggle to simultaneously achieve the goals of high relative density and low crack density, and the models lack accuracy and reliability, especially with small sample data where they are prone to overfitting and insufficient validation.
A Gaussian process regression method is used to construct a prediction model for relative density and crack density. The Matrn kernel function is used to describe the nonlinear relationship. White noise is introduced to kernelize the noise. The optimal parameters are searched in the process parameter space by combining the Bayesian optimization algorithm.
It achieves high-precision prediction and multi-objective optimization with small sample data, quickly finds the optimal process parameters, reduces experimental costs and time, and is applicable to selective laser melting processes with different equipment and materials.
Smart Images

Figure CN122366077A_ABST
Abstract
Citation Information
Patent Citations
SLM forming performance prediction and process parameter optimization method and system
CN116204998A