PREDICTIVE MAINTENANCE OF WIRE BREAKAGE AND QUALITY YIELD OF WIRE SAW TOOLS WITH MULTIVARIATE LIMIT MODEL FOR THE 27th EU PVSEC

摘要

Data from a large solar-panel manufacturer show that wire breakage occurred over 25% from allfailures and quality yield lower than 90% occurred 32% of the time. With this motivation, predictive maintenancemodels are developed to predict wire breakage and low quality yield. Specifically, easy-to-deploy multivariate limitmodels are developed, involving the following key steps: calculation of sensor statistics, data preparation toaccurately group data, determination of most important sensors, development of univariate models of the effect ofthese key sensors, analysis of how well model predictions overlap, and building multivariate models using overlapanalysis. The result shows that 57% of all wire breakage events and 71% of all low quality yield resulting in waferloss are predicted.