A multi-objective genetic algorithm (GA) approach for optimization of surface grinding operations

doi:10.1016/S0890-6955(02)00074-3

International Journal of Machine Tools and Manufacture

Volume 42, Issue 12, September 2002, Pages 1327-1334

https://doi.org/10.1016/S0890-6955(02)00074-3 Get rights and content

Abstract

A genetic algorithm (GA) based optimization procedure has been developed to optimize grinding conditions, viz. wheel speed, workpiece speed, depth of dressing and lead of dressing, using multi-objective function model with a weighted approach for surface grinding process. The procedure evaluates the production cost and production rate for the optimum grinding condition, subjected to constraints such as thermal damage, wheel wear parameters, machine tool stiffness and surface finish. New GA procedure is illustrated with an example and optimum results such as production cost, surface finish, metal removal rate are compared with quadratic programming techniques.

Introduction

Optimization analysis of machining process are usually based on either minimizing production cost, maximizing production rate, or obtaining the fitness possible surface quality by using empirical relationships between the tool life and the operating parameters. Optimization analysis are applicable also to grinding process provided the suitable tool life equations are available. Fortunately, many researchers have published a number of such equations for the practical grinding process in which numerous process variables are involved. The development of comprehensive grinding process models and computer-aided manufacturing provides a basis for realizing grinding parameter optimization.

Previous work on the optimization of grinding parameters has concentrated on the possible approaches for optimizing constraints during grinding [1]. The technique of optimizing both grinding and dressing condition for the maximum work piece removal rate subjected to constraints on workpiece burn and surface finish in an adaptive control grinding system can be found in ref. [2]. The use of quadratic programming for the optimization of grinding parameters subject to multi-objective function has been reported in ref. [3]. In the previous work of the authors, genetic algorithm (GA) based optimization procedure has been successfully implemented for solving surface grinding process problem by considering single objective function [4].

This paper describes a GA based optimization procedure to optimize grinding conditions, viz. wheel speed, workpiece speed, depth of dressing, and lead of dressing, using a multi-objective function model with a weighted approach for surface grinding. The procedure evaluates the optimal grinding conditions subjected to constraints such as thermal damage, wheel-wear parameters and machine-tool stiffness. Initially a detailed description of the mathematical model of the grinding process is given. Then the optimization procedure is described. Finally an example taken from the literature [3] is used to illustrate new approach.

Section snippets

Mathematical model of the surface grinding process

The aim of carrying out a grinding operation is to obtain the finished product with minimum production cost, maximum production rate, and the finest possible surface grinding finish. The mathematical model proposed by Wen [3] is used in this work.

Solution methodology

Most of the researchers have used traditional optimization techniques for solving machining problems [10]. The traditional methods of optimization and search do not fare well over a broad spectrum of problem domains.

Traditional techniques are not efficient when practical search space is too large. These algorithms are not robust. Numerous constraints and number of passes make the machining optimization problem more complicated. Traditional techniques such as geometric programming, dynamic

Genetic algorithm

GA form a class of adaptive heuristics based on principles derived from the dynamics of natural population genetics. The searching process simulates the natural evaluation of biological creatures and turns out to be an intelligent exploitation of a random search. A candidate solution (chromosome) is represented by an appropriate sequence of numbers. In many applications the chromosome is simply a binary string of 0 and 1. The quality of its fitness function, which evaluates a chromosome with

Numerical illustration

The following data [3] is used as the input data for this optimization problem.

Description	Symbol	Value
No. of work piece on table (pc)	P	1
Length of the work piece (mm)	L_W	300
Empty length of grinding (mm)	L_e	150
Width of work piece (mm)	b_W	60
Empty width of grinding (mm)	b_e	25
Cross feed rate (mm/pass)	f_b	2
Total thickness of cut (mm)	a_W	0.1
Down feed of grinding (mm/pass)	a_p	0.0505
Number of spark out grinding (pass)	S_p	2
Diameter of wheel (mm)	D_e	355
Width of wheel (mm)	b_S	25
Grinding ratio	G	60
Distance of wheel idling (mm)	S_d	100

Implementation of GA

The implementation of the GA technique for the multi-objective optimization problem is discussed below.

Result and comparison

The operating parameters, optimum production cost, optimum production rate, surface finish and combined objective function obtained by GA and quadratic programming are given in Table 1, Table 2.

From these tables, it is observed that the combined objective function obtained by GA is 47.27% better than those obtained by quadratic programming for rough grinding and 6.3% better for finish grinding. For rough grinding, cost is 14.62% higher but the work piece removal parameter is higher by 24%. For

Conclusion

For solving the machining optimization problems, various conventional techniques have been used so far. It is observed that, the conventional methods are not robust, because of the following reasons:

•
The convergence to an optimal solution depends on the chosen initial solution.
•
Most algorithms tend to get stuck to a sub-optimal solution.
•
An algorithm efficient in solving one machining optimization problem may not be efficient in solving a different machining optimization problem.
•
Algorithms are not

References (15)

X.M. Wen et al.
Microcomputer based optimization of the surface grinding process
J. Mater. Process. Technol.
(1992)
S. Malkin, Practical approaches to grinding optimization, Milton.C. Shaw Grinding Symposium, ASME Winter Annual Meeting...
G. Amitay, Adaptive control optimization of grinding. J. Eng. Ind. ASME (1981)...
R. Saravanan, S. Vengadesan, M. Sachithanandam, Selection of operating parameters in surface grinding process using...
S. Malkin
Selection of operating parameters in surface grinding of steels
J. Eng. Ind. ASME
(1976)
S. Malkin
Thermal aspects of grinding, part 2, surface temperature and workpiece burn
J. Eng. Ind. ASME
(1974)
R.I. King et al.
Handbook of Modern Grinding Technology
(1986)

There are more references available in the full text version of this article.

Cited by (119)

Multi-objective optimization for improving machining benefit based on WOA-BBPN and a Deep Double Q-Network
2023, Applied Soft Computing
To scientifically and reasonably set and adjust the machining parameters, and improve the machining benefit of the batch machining process including machining quality, performance and efficiency, this paper proposes a low cost and high accurate multi-objective optimization framework by integrating intelligent technologies from three perspectives of physical experiment design, objective function construction and multi-objective optimization. In this framework, multilayer design (MLD) which is a computer experiment design method with low redundancy, good uniformity and filling is applied to the machining physics experiment design to reduce the experiment cost. Whale optimization algorithm (WOA) optimized backpropagation neural network (BBPN) is utilized to construct the optimization objective functions to improve the accuracy of functions. And Deep Double Q-Network (DDQN) is employed to achieve multi-objective optimization of machining quality, efficiency, and performance. The effectiveness of MLD, WOA-BBPN and DDQN in multi-objective optimization of machining process are verified by comparisons of common modeling and multi-objective optimization algorithms based on the milling data. These techniques provide a good guidance for the selection of machining parameters to improve the machining benefit and demonstrates some advantages. The research of this paper also offers a good example of applying intelligent technology in machining.
A new flower pollination algorithm with improved convergence and its application to engineering optimization
2022, Decision Analytics Journal
The flower pollination algorithm (FPA) is a nature-inspired optimization algorithm that mimics the pollination behaviour of flowering plants. Despite the promising performance of FPA in solving single objective optimization problems, its convergence still poses challenges in practice. This study proposes a modified FPA with additional features from chaos theory and frog leaping algorithm augmented by inertia weights. The modified FPA proposed in this study is tested against benchmark mathematical functions, mechanical engineering design optimization problems, and machining process optimization problems. Performance comparison with other state-of-the-art optimization algorithms has demonstrated its ability in terms of convergence. The modified FPA significantly reduced the number of function evaluations by 84.14%, as compared to FPA in optimizing the benchmark mathematical functions. Besides, the proposed modified FPA outperformed others in 12 out of 15 mechanical engineering optimization problems.
A hybrid multi-objective optimization of aerosol jet printing process via response surface methodology
2020, Additive Manufacturing
Aerosol jet printing (AJP) is an emerging 3-dimensional (3D) printing technology to fabricate customized and conformal microelectronic components on various flexible substrates. Although the AJP technology has the capability of depositing fine features, the inherent contradiction between the printed line thickness and line edge roughness has a great impact on the printed line quality. In this paper, a novel hybrid multi-objective optimization method is proposed to optimize the overall printing quality of AJP based on a determined operating window. The proposed approach consists of a central composite design (CCD), a response surface methodology, a desirability function approach and a non-dominated sorting genetic algorithm III (NSGA-III). In the proposed approach, the response surface methodology is combined with the CCD to investigate and quantify the correlations between the printed line features and the key process parameters. And the conflicting relationship between the printed line edge roughness and line thickness is identified by the CCD derived response surface models (RSMs). Then the 2D optimal operating windows are determined to minimize the inherent contradiction between the printed line features at a certain print speed by the desirability function approach. Following that, the derived RSMs and the corresponding statistical uncertainty are jointly driven with a NSGA-III to systematically optimize the overall printing quality in 3D design space by more robust manners. The experimental results demonstrate that the proposed hybrid multi-objective optimization approach is beneficial to minimize the conflict between the printed line features, hence the lines can be produced with low line edge roughness and sufficient line thickness. Different from a traditional trial-and-error method in AJP, the proposed printing quality optimization approach is developed based on the principles of statistical modeling, analysis of variance and global optimization. Therefore, the proposed printing quality optimization approach is more efficient and systematic. Moreover, the data-driven based characteristic makes the proposed approach applicable to other multi-objective optimization researches in additive manufacturing technologies.
Application and performance of machine learning techniques in manufacturing sector from the past two decades: A review
2020, Materials Today: Proceedings
Advancement in technology has created wide opportunities for the researchers to utilize artificial intelligence in various fields. Numerous attempts have been made in the use of machine learning tools in the manufacturing and production sector. However, variation in the performance of techniques is creating a major quagmire for the researchers. In many cases, some methods have shown similar results while in some cases one outperformed another. Choosing the best and suitable technique for process modelling and optimization is still a challenging task for the researchers. Hence, to present a direction for the prospect investigators, in this study, the performance of different machine learning techniques applied in the manufacturing sector is reviewed by assessing many articles from the past two decades. Among several machine learning techniques reviewed in this study, application of artificial neural networks (ANN) in process modelling and optimization has become quite noticeable because of its ability to predict the output quickly and accurately. The effectiveness and practicality of ANN models in manufacturing applications are reviewed for demonstrating its pivotal role in process modeling. Observations are reported in the study.
Modeling and optimization of machining parameters during grinding of flat glass using response surface methodology and probabilistic uncertainty analysis based on Monte Carlo simulation
2019, Measurement: Journal of the International Measurement Confederation
In this paper, the performance of diamond grinding wheels was investigated. The industrial diamond crystals with a size of 140/170 mesh were utilized. The microstructure of the grinding tool was observed using a Scanning Electron Microscope (SEM) and Energy Dispersive X-ray Analysis Device (EDX). The experiments were designed using Box–Behnken method and optimum grinding parameters for glass were analytically determined. Experimental studies were carried out on a surface grinding machine in a flat glass factory. Grinding characteristics were examined with respect to surface roughness. The effects of grinding parameter on output responses were studied using analysis of variance (ANOVA). Probabilistic uncertainty analysis depends on Monte Carlo simulation was applied. Moreover, after the experiments using the optimized cutting parameters, the microstructure of the grinding wheels was analyzed. From results, the established model and optimization method could be employed for predicting surface roughness and this work is reliable and suitable for solving the problems encountered in machining operations. The lifetime of Cu-based grinding discs can be increased by adding Zn and Fe to the matrix material.
Forestry Canopy Image Segmentation Based on Improved Tuna Swarm Optimization
2022, Forests

View all citing articles on Scopus

View full text