Knowledge Management for Grinding Technology

This thesis describes an investigation concerned with development of a grinding knowledge warehouse system (GKWS). Based on a study of previous work on knowledge management and technique for a selection of grinding conditions, the thesis proposes a novel methodology to deal with missing data in surface and cylindrical grinding using Genetic Programming.

The GKWS provides a guided tool for users to support the decision-making process to provide suggestions for selecting grinding conditions using rule-based reasoning (RBR) and case-based reasoning (CBR) and it can learn from new and previous grinding cases to improve and expand the CBR cases.

The GKWS developed a new methodology to deal with missing data in grinding operations. The new methodology is built on If-Then rules, mathematical equations and modelling using genetic programming (GP). Dealing with missing data improves the performance of knowledge discovery in the GKWS and the results of the CBR.

The GP is developed for modelling surface roughness in cylindrical and surface grinding. The developed GP model for surface grinding shows the ability to predict the surface roughness parameter especially when the GP terminals vary and the same material and wheel are used.

The discussion forum facilitates and supports transferring tacit knowledge into explicit knowledge where the users can exchange their ideas, send questions and answers, and pass on important links. The tacit knowledge is acquired directly from the knowledge engineers. The debate and discussion in GKWS will create new knowledge that is accessible and available when needed.