ハイパースペクトルデータに基づいためっき工程の状態認識

抄録

Demands for condition monitoring in manufacturing environment have recently increased in a variety of industrial sectors. This study presents a newly developed monitoring method for electroless nickel plating process using hyperspectral data. The hyperspectral data is a combination of two-dimensional spatial information and spectral information with high wavelength resolution. In order to confirm the usefulness and validity of the quantitative evaluation of the condition such as deterioration of the plating solution, the multivariate analysis on hyperspectral data was applied. Partial least squares (PLS) regression of hyperspectral data was used for estimating a model equation for predicting the concentration of sodium phosphite, which causes deterioration of the plating solution. The plating solution can be monitored based on the proposed method using a hyperspectral camera capable of acquiring the hyperspectral data. The correlation coefficient between the concentration of sodium phosphite estimated by the proposed method and by titration was 0.97. The experimental results confirmed that the proposed method can evaluate the deterioration of plating solution quantitatively.