Using support vector regression and non-dominated sorting genetic algorithm in multi-objective optimization of milling of S50C steel under MQL condition

Istrazivanja i projektovanja za privredu

Ba Nghien Nguyen¹  van Canh Nguyen²  Xuan Truong Nguyen²  van Que Nguyen²  Tien Dung Hoang²  Thuy Duong Nguyen³ 

[1] Hanoi University of Industry, Faculty of Information Technology, Hanoi, Vietnam;Hanoi University of Industry, Faculty of Mechanical Engineering, Hanoi, Vietnam;Hanoi University of Science and Technology, School of Mechanical Engineering, Hanoi, Vietnam;

关键词: support vector regression;    non-dominated sorting genetic algorithm;    minimum quantity lubricant;    multi-objective optimization;   

DOI  :  10.5937/jaes0-31366

来源: DOAJ

【 摘 要 】

The modern machining industry faces reducing manufacturing cost pressure and improve product quality expectations. Due to this competition a manufacturer must continually identify cost cutting opportunities in manufacturing process. The keys technology represents cost-saving opportunities associated with reducing cutting fluid consumption, cutting energy consumption, and improves the overall performance of machining operations at the same time. Hence, multiple response optimization techniques have recently become the focus of research to improve product quality by increasing surface quality and reduce costs by reducing cutting force, cutting fluids, and energy consumption of the cutting process recently. Among various optimization methods, few Taguchi-based likes as TOPSIS, COPRAS, MOORA, VIKOR... were chosen to solve complex multiple criteria problems. However, the limitation of these techniques is that it was helped to rank and select the best parameters set for the implemented experiments only. In this work, an attempt was made to streamline the milling and coolant condition parameters of S50C carbon steel under MQL condition. Experiments were performed based on Taguchi's L27 orthogonal array. Five input factors includes machining parameters including cutting speed (Vc), feed (fz), and depth of the cut (ap) combined with two coolant parameters such as coolant air pressure (P), flow rate of lubricant (Q) were considered the variants, while specific cutting energy (Ec), material removed rate (MRR) and surface roughness were response variables. Particle swarm optimization Support Vector Machines (SVM) was used to generate the regression model, then Non-dominated Sorting Genetic Algorithm (NSGA) was used to optimize surface roughness (Ra), specific cutting energy (Ec), and production rate (MRR).

【 授权许可】

Unknown