1. Surface roughness prediction model for CNC machining of polypropylene
- Author
-
R D Allen, Stephen T. Newman, Vimal Dhokia, Parag Vichare, and Sandeep Kumar
- Subjects
Polypropylene ,Engineering ,Artificial neural network ,business.industry ,Mechanical Engineering ,Design of experiments ,Process (computing) ,Mechanical engineering ,Industrial and Manufacturing Engineering ,chemistry.chemical_compound ,Machining ,chemistry ,Surface roughness ,Numerical control ,business ,Statistical hypothesis testing - Abstract
Cutting strategy research has traditionally been focused on hard materials that are intrinsically difficult to machine. An increase in the desire for personalized products has led to the requirement of the direct machining of polymers for personalized products. Little research is evident in the literature on the analysis of optimal machining parameters for machining materials such as polypropylene. One of the vital factors that affects the quality of polypropylene products and the respective machining strategy is surface roughness. This research is aimed at extracting information on the machining of polypropylene materials. A surface roughness predictive model based on neural networks has been developed. The design of experiments approach is used to obtain an adequate predictive model for the process planning which is further utilized as an input to the predictive model. The model mainly hinges on three independent variables namely spindle speed, feed rate, and depth of cut. Extensive experimental work on different network topologies and training algorithms has been performed to predict the behaviour of the surface roughness for machined polypropylene products. The results illustrate the benefits of being able to determine surface roughness values. This allows for the determination of optimal cutting strategies and tooling for the required surface roughness. The performance predictive model has been found to be satisfactory over the dataset for polypropylene machining. Hypothesis testing has also been carried out to identify the confidence of the predictive model.
- Published
- 2008