Publication

Abstract : More and more organisations are trying to install tool health analytics dashboards for CNC machines to avoid unexpected failures, maintain machining accuracy, and optimise tool change. This paper aims at developing a cyber physical production system framework for a smart tool health management system to prescribe the optimum cutting parameters to managers/operators for optimising the remaining useful life and/or material removal rate at a predefined surface finish (individually or simultaneously). This is achieved by developing (i) a machine learning algorithm to predict the remaining useful life of a cutting tool, (ii) regression models to prescribe optimum cutting parameters (iii) a machine learning algorithm for anomaly detection, and (iv) a knowledge-based system for chip conditions and tool life curves. Experiments are designed and conducted based on Taguchi L-27 orthogonal array with varying combinations of cutting parameters during the milling of a difficult to machine material (AISI H13 tool steel). The effect of cutting parameters is analysed statistically; using analysis of variance (ANOVA), response tables, and main effect plots; to prescribe optimum cutting parameters based on managerial requirements. A novel knowledge-based system is also presented that updates knowledge and information about the chip colour at different health conditions of a tool. The present work will be a significant step towards improving productivity, product quality, and reducing maintenance costs by providing practitioners with an active decision support tool that will assist them to confidently adopt optimum management and control strategies within an Industry 4.0 environment.