Statistical considerations for real-time size control systems in wood products manufacturing

摘要

Currently, sawmill machinery companies are developing real-time size lumber size control systems using non-contact laser measuring systems. These systems rely on the application of industrial statistics to large quantities of lumber thickness and width data. Because of the sampling intensity and frequent decision making in real-time systems, there is an increased chance of committing Type I or Type 11 errors when drawing conclusions if statistical methods are incorrectly applied. There is confusion in the industry concerning the appropriate statistical model to use for lumber size control. This survey of the current literature discusses three distinct methods for calculating and partitioning sawing variation, and thereby calculating control limits for control charts. This paper reviews the statistical foundation and current understanding of industrial statistics for implementing real-time SPC systems and makes recommendations for improvement.