4
A. Kumar et al.
moulding processes as the molten plastic is not forced to take any shape. And due
to this, products with good mechanical properties can be manufactured. Products
manufactured from this technique have wide range of applications in various fields
like in agriculture, automobiles, medical equipment’s, furniture tanks, etc. [1–3].
This whole process is completed mainly in four stages as which starts with
charging the mould with a thermoplastic material (usually linear low density
polyethene) which is in the powdered form with an average of 500-micron particle
size. Secondly, heating and melting of the powder in the mould which is rotated biaxially, so that the melted powder sticks to the surface of mould followed by cooling
of the mould and lastly de-moulding of the part at a temperature which is slightly
higher than the room temperature.
There is no doubt that there have been great technological advancements in rotational moulding process in the recent decade. It has been evolved from a manufacturing process overlooked for many as ‘black art’ to a refined moulding process
for making high quality products, which cannot be made by any other moulding
processes [4, 5]. Rotational moulding process is capable of producing products that
have different mechanical properties like tensile strength, impact strength and flexure
strength. And this can be done only by changing oven residence time. In present
scenario of industries, there has been a need of products of different mechanical
properties, and for this, we need to know the appropriate oven residence time for
manufacturing such products [6, 7]. The term ‘oven residence time’ is generally
understood to mean the total time a polymer resides in the oven from room temperature till the oven in turned off. At present, these mechanical properties for different
oven residence timings are obtained by manually done experiments. And due to this,
the whole process becomes costly and time consuming.
A less costly and time-consuming approach can be using machine learning models
to predict the mechanical properties of the roto moulded product for different oven
residence timings.
1.2 Machine Learning
Machine learning is a subdivision of artificial intelligence (AI) that gives a model
which can learn and improve with time without being explicitly instructed or
programmed, so instead of writing codes, some input data is provided in the generic
algorithm, and the algorithm itself generates a logic from the given data. These
algorithms are created in such a way that they learn and improve themselves over
time when are provided with more data [8, 9]. In the past recent years, it can be
observed that machine can be used as to automatize different tasks that were thought
of as only humans can do like playing games, face recognition, text generation, etc.
[10–12] Applying machine learning in various industries like in medicine, transportation, banking can be very useful as it can be used as to perform tasks like video
surveillance, spam and malware, fraud detection, prediction and many more [13–15].
Machine learning has mainly divided into four parts as follows [16–18]:
A. Kumar et al.
moulding processes as the molten plastic is not forced to take any shape. And due
to this, products with good mechanical properties can be manufactured. Products
manufactured from this technique have wide range of applications in various fields
like in agriculture, automobiles, medical equipment’s, furniture tanks, etc. [1–3].
This whole process is completed mainly in four stages as which starts with
charging the mould with a thermoplastic material (usually linear low density
polyethene) which is in the powdered form with an average of 500-micron particle
size. Secondly, heating and melting of the powder in the mould which is rotated biaxially, so that the melted powder sticks to the surface of mould followed by cooling
of the mould and lastly de-moulding of the part at a temperature which is slightly
higher than the room temperature.
There is no doubt that there have been great technological advancements in rotational moulding process in the recent decade. It has been evolved from a manufacturing process overlooked for many as ‘black art’ to a refined moulding process
for making high quality products, which cannot be made by any other moulding
processes [4, 5]. Rotational moulding process is capable of producing products that
have different mechanical properties like tensile strength, impact strength and flexure
strength. And this can be done only by changing oven residence time. In present
scenario of industries, there has been a need of products of different mechanical
properties, and for this, we need to know the appropriate oven residence time for
manufacturing such products [6, 7]. The term ‘oven residence time’ is generally
understood to mean the total time a polymer resides in the oven from room temperature till the oven in turned off. At present, these mechanical properties for different
oven residence timings are obtained by manually done experiments. And due to this,
the whole process becomes costly and time consuming.
A less costly and time-consuming approach can be using machine learning models
to predict the mechanical properties of the roto moulded product for different oven
residence timings.
1.2 Machine Learning
Machine learning is a subdivision of artificial intelligence (AI) that gives a model
which can learn and improve with time without being explicitly instructed or
programmed, so instead of writing codes, some input data is provided in the generic
algorithm, and the algorithm itself generates a logic from the given data. These
algorithms are created in such a way that they learn and improve themselves over
time when are provided with more data [8, 9]. In the past recent years, it can be
observed that machine can be used as to automatize different tasks that were thought
of as only humans can do like playing games, face recognition, text generation, etc.
[10–12] Applying machine learning in various industries like in medicine, transportation, banking can be very useful as it can be used as to perform tasks like video
surveillance, spam and malware, fraud detection, prediction and many more [13–15].
Machine learning has mainly divided into four parts as follows [16–18]: