Towards enhancement of machinability data by multiple regression

Abstract

An investigation of various multiple regression model-building techniques has been carried out on machinability data in order to study the suitability of the empirical equations employed. A comparative analysis of the full-form first-order regression mode and quadratic regression model has been performed also, to aim for a parsimonious and easily interpretable model to be used for the computation of various aspiration levels, which latter include the minimum unit production cost and the maximum unit production rate. In order to address the problem of the update of machinability data found in current systems, the raw data of the machining responses and machining variables are retained for future update, thus enhancing the models, which relate closely to the actual production process. This module can thus be incorporated into an expert system that has been discussed elsewhere, achieving progress towards an integrated machining system. © 1989.