6
A. Kumar et al.
options in the stock market very precisely, this reduces the chances of human error
involved. Nowadays, it is also used to calculate the stock prices with the help of the
history of the stock prices and various company data as the training data sets [21].
Machine Learning in Manufacturing. In the present years, smart manufacturing
ideas are taking a step ahead, and according to the research carried by Wang [22],
various machine learning models are made to improve the system performance in
the manufacturing systems [22].
Machine Learning in Plastic Moulding. It is evident from the research done by
Tellaeche and Arana [23] that machine learning algorithms are very useful in the
quality control in the plastic moulding industry as they can perform fault detection
in fabrication of plastic products better that the traditional quality control methods
[23].
1.4 Research Gap
Application of machine learning in the field of rotational moulding has not yet been
fully explored. Like in the other fields, machine learning can also be applied in
rotational moulding for the prediction of mechanical properties of roto moulded
product for the given oven residence time by using machine learning models.
2 Methodology
For the prediction of mechanical properties for the given oven residence time, a
training data was taken which had the value of tensile strength, impact strength and
flexure strength for different oven residence timings which were obtained by doing
experiments manually by Ramkumar [24]. From these training data set, machine
learning models were created (here polynomial regression model and linear regression model supervised learning models, as there were both input parameters and
output variables in the training data sets), in Python. For building the required
machine learning model, Python programming was used.
From these models, various prediction curves (mechanical property vs. oven residence time) were obtained. Further, curves from the two models (linear regression
model and polynomial regression model) were compared on the basis of precision
of the obtained value of mechanical property, and then, best suited model for the
prediction was selected.
A. Kumar et al.
options in the stock market very precisely, this reduces the chances of human error
involved. Nowadays, it is also used to calculate the stock prices with the help of the
history of the stock prices and various company data as the training data sets [21].
Machine Learning in Manufacturing. In the present years, smart manufacturing
ideas are taking a step ahead, and according to the research carried by Wang [22],
various machine learning models are made to improve the system performance in
the manufacturing systems [22].
Machine Learning in Plastic Moulding. It is evident from the research done by
Tellaeche and Arana [23] that machine learning algorithms are very useful in the
quality control in the plastic moulding industry as they can perform fault detection
in fabrication of plastic products better that the traditional quality control methods
[23].
1.4 Research Gap
Application of machine learning in the field of rotational moulding has not yet been
fully explored. Like in the other fields, machine learning can also be applied in
rotational moulding for the prediction of mechanical properties of roto moulded
product for the given oven residence time by using machine learning models.
2 Methodology
For the prediction of mechanical properties for the given oven residence time, a
training data was taken which had the value of tensile strength, impact strength and
flexure strength for different oven residence timings which were obtained by doing
experiments manually by Ramkumar [24]. From these training data set, machine
learning models were created (here polynomial regression model and linear regression model supervised learning models, as there were both input parameters and
output variables in the training data sets), in Python. For building the required
machine learning model, Python programming was used.
From these models, various prediction curves (mechanical property vs. oven residence time) were obtained. Further, curves from the two models (linear regression
model and polynomial regression model) were compared on the basis of precision
of the obtained value of mechanical property, and then, best suited model for the
prediction was selected.