International Journal of Recent Technology and Engineering (IJRTE)
ISSN: 2277-3878, Volume-8 Issue-3, September 2019
235
Published By:
Blue Eyes Intelligence Engineering
& Sciences Publication
Retrieval Number: C3952098319/19©BEIESP
DOI:10.35940/ijrte.C3952.098319
Abstract: Stainless steel is most extensively utilized material in
all engineering applications, house hold products, constructions,
because it is environment friendly and can be recycled. The
principal purpose of this paper is to implement different data
science algorithms for predicting stainless steel mechanical
properties. Integrating Data science techniques in material
science and engineering helps manufacturers, designers,
researchers and students in understanding the selection,
discovery and development of materials used for various
engineering applications. Data science algorithms help to find
out the properties of the material without performing any
experiments. The Data Science techniques such as Random
Forest, Neural Network, Linear regression, K- Nearest Neighbor,
Support vector Machine, Decision Tree, and Ensemble methods
are used for predicting Tensile Strength by specifying processing
parameters of stainless steel like carbon content, sectional size,
temperature, manufacturing process. The research here is
developed as part of AICTE grant sanctioned under RPS scheme
[19] and it aims to implement different data science algorithms
for predicting Tensile strength of steel and identifying the
algorithm with decent prediction accuracy.
Index Terms: Data Science algorithms, Material Science,
Mechanical properties of steel, process parameters of steel,
Statistical measures of accuracy.
I. INTRODUCTION
The designers, manufacturers, researchers of material
sciences and engineering have habitually contingent on
results obtained from the experiments that are conducted in
testing laboratory to identify the mechanical properties of
any material. So, to obtain the desired properties for a
material they need to customize composition of material and
process parameters of the material prior of conducting the
experiment. But these procedures demand massive
expenditure and time to figure out the properties of
materials. Different types of materials in material science
and engineering includes ceramics, Biomaterials,
Composites, Concretes, Electronic and Optical materials,
Glasses and Metal alloys. This paper on the whole focuses
on Steel which is a metal alloy and prominent material used
in wide variety of applications like construction,
infrastructure, automobiles, machine appliances, ships, cars,
weapons, vessels, household purposes, surgical instruments,
etc. Steel is predominantly accustomed in many applications
due to its elevated tensile strength and recyclability without
Revised Manuscript Received on September 15, 2019
Dr. N. Sandhya, Department of computer science and engineering,
VNR Vignana Jyothi Institute of Engineering & Technology, Hyderabad,
India.
V. Sowmya, Department of computer science and engineering, VNR
Vignana Jyothi Institute of Engineering & Technology, Hyderabad, India.
Dr. Chennakesava Rao Bandaru, Department of Mechanical
Engineering, Vignana Jyothi Institute of Engineering & Technology,
Hyderabad, India.
Dr. G. Raghu Babu, Department of Mechanical Engineering, Vignana
Jyothi Institute of Engineering & Technology, Hyderabad, India.
dropping down the mechanical property values. To meet the
modern market requirements and to withstand the
competitiveness many industries emphasized to produce
high precision and good quality steel products. But using the
traditional approach of conducting tensile tests using UTM
(universal tensile testing machine) and some other tests are
not effective in terms of cost and time consumed for tests for
finding out mechanical properties of steel. There is a lot of
progress in the recent years by using Data Science
techniques in material sciences and engineering for
discovering the design, structure, physical and mechanical
properties of any material. The main goal of data science is
to diminish cost and reduce time mechanical properties
prediction to help the manufacturers and designers of
advanced material science and engineering. Accurate
prediction of material mechanical properties and its behavior
based on existing data has been the persistent effort of many
material researchers. Data Science has succeeded in adding
value to business models with the help of statistics,
machine learning and deep learning. The main aim of the
Data Science is to develop novel approaches, algorithms,
tools, methods and the associated infrastructure to extract
the high value information based on the available data and
resources. The Data Science techniques are widely classified
into machine learning, regression, logistic regression,
pattern recognition, Feature selection, Text mining,
Attribute modelling, k-means clustering, Association
analysis, Anomaly detection, Social network analysis,
collaborative filtering, Time series forecasting, Model
fitting, cross validation, LTV and RFM Analysis, etc. This
paper aims to implement data science techniques to predict
tensile strength of Stainless steels.
II. RELATED WORK
Titus Thankachan.et.al [1] proposed method predicts tensile
strength, yield strength of hydrogen charged aluminum
alloys using artificial neural network by considering
composition of alloy and various processing parameters as
an input to the model. GH Senussi.et.al [2] presented an
approach for predicting stainless steel micro hardness using
artificial neural network, where micro hardness of six
different types of stainless steel is predicted based on the
distance from the base surface. Raghuram Karthik Desu.et.al
[3] predicted Tensile strength, yield strength of Austenitic
stainless steel 304L and 316L using feed forward back
prorogation neural network. The composition of steels is
kept constant while predicting the properties at varying
strain rates and at elevated temperatures. Yang Weng.et.al
[4] implemented Single Index
model that predicts Tensile strength, Yield strength of hot
rolled strips of C-Mn steels
concerning its chemical
composition and processing
parameters. Vandana
Prediction of Mechanical Properties of Steel
using Data Science Techniques
N. Sandhya, Valluripally Sowmya, Chennakesava Rao Bandaru, G. Raghu Babu