An in-process adaptive control of surface roughness in end milling operations

Abstract

In recent years, the computer numerical control (CNC) in machining manufacturing processes plays a positive and important role. A CNC machine enhances both quality and productivity. However, the variance of machining processes still occurred and the quality was still an issue for engineers to solve. To solve the issue, a quality inspection was conducted to ensure the satisfaction of customers. The quality inspection takes time and it should be eliminated from the aspect of lean production. To eliminate the wasting time of inspection, the in-process quality inspection was proposed to measure the quality characters during the machining processes without wasting time. Surface roughness (Ra), a key index to evaluate and determine the quality of a product, is influenced by the machine parameters and other factors resulting from the cutting tool. Surface roughness has a direct effect on the functional characteristics of the workpiece such as fatigue, fracture resistance, and surface friction. The traditional measurement of surface roughness always takes time and if the quality were not within the specification, the machine should stop and make adjustment. Therefore, to maintain the quality assurance and eliminate the inspection and adjusting time, an in-process neural-based adaptive control of surface roughness in end milling operations was proposed in this study. A neural network was applied as a decision-making algorithm to immediately predict the surface roughness. Following by the prediction, the predicted Ra was then compared to customer's specification. If the quality is not within the specification, the adaptive control system will be launched to adjust the feed rate and spindle speed of the CNC machine.