Fuzzy Knowledge-based Modeling and Statistical Regression in Abrasive Wood Machining

Document Type


Publication Date



Abrasive machining processes in wood largely determine the surface quality of a product. These processes are highly unpredictable due to variation in tooling and process conditions as well as in wood species considered. Two methods of predicting the performance of abrasive wood machining processes are considered here. A comparison between non-linear regression modeling and fuzzy knowledge-based modeling is presented. Complex experimental models are developed and empirical studies performed. With data from these studies, several regression models are fitted and a fuzzy knowledge-based model using fuzzy rules is also developed. The abilities of these two models to explain the experimental results are compared and then used to predict experimental results based on process conditions. These predictions are subsequently examined in the context of additional experimental results. The two models are similar in their ability to explain the data and to predict outcomes based upon new inputs. However, the regression model appears to offer improved performance as the size of the data set increases, whereas the fuzzy model appears to offer improved performance when the size of the data set is small. It is concluded that fuzzy models appear most appropriate when subjective and qualitative data are utilized and the number of empirical observations available is small.


Copyright 2004 Forest Products Society

A link to full text has been provided for authorized subscribers.

Publication Title

Forest Products Journal

Published Citation

Carrano, Andres L., James B. Taylor, Robert E. Young, Richard L. Lemaster, and Daniel E. Saloni. "Fuzzy knowledge-based modeling and statistical regression in abrasive wood machining." Forest products journal 54, no. 5 (2004): 66-72.

Peer Reviewed