研究目的
To develop a geometrical model with assembly defects consideration to reduce the calibration time of additive manufacturing machines by generating correction tables more efficiently.
研究成果
The developed geometrical models accurately simulate the laser spot position and quantify the impact of assembly defects. The identification algorithm efficiently reduces calibration time by requiring only a single measurement phase. Tests on an industrial machine show that 95% of deviations are below 20 μm, with significant time savings achieved. Future work could address additional defect sources and improve model robustness.
研究不足
The model does not account for all defect sources such as thermal defects, optical defects, or measurement uncertainties beyond assembly defects. Local errors may persist due to unmodeled factors like surface imperfections or non-linearities. The basis for defect identification is not orthogonal, leading to potential multiple solutions.
1：Experimental Design and Method Selection:
The study involves developing a nominal geometrical model and a model with 30 assembly defects for the opto-mechanical chain of an additive manufacturing machine. The models are based on geometrical optics and use transformation matrices to simulate the laser beam path. An identification algorithm (Algorithm 1) is used to fit the model to experimental data.
2：Sample Selection and Data Sources:
Data is generated from a virtual machine with known defects for initial testing, and real data is obtained from an industrial LPBF machine. The dataset includes joint configurations (Qexp) and corresponding laser spot positions (Xexp) measured with a 10 μm uncertainty.
3：List of Experimental Equipment and Materials:
An industrial LPBF machine is used, with components including a laser source, galvanometric scan head, mirrors, windows, and a work plane. A reference plate is engraved for measurements.
4：Experimental Procedures and Operational Workflow:
The machine is preheated to account for thermal effects. A grid is engraved on a reference plate, and positions are measured. The identification algorithm iteratively adjusts the model parameters to minimize deviations between simulated and measured positions.
5：Data Analysis Methods:
Deviations are analyzed using maximum displacement criteria, cumulative histograms, and statistical measures (mean, standard deviation). The Newton-Raphson method is used for numerical inversion of the kinematic models.
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
