Prediction of Shrinkage Behavior of Stretch Fabrics Using Machine-Learning Based Artificial Neural Network

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The manuscript is suitable for publication in your prestigious journal. However, there are some comments that should be considered such as:

1.      The authors should revise the abstract, in its current form abstract show’s introduction

2.      The authors should not repeat the title words in the Keywords section such as Machine Learning, based Artificial Neural Network

3.      The authors should remove Figure. 1 & 2 from the introduction

4.      Show the novelty in the introduction

5.      English typo mistakes must revise in the entire manuscript

6.      Authors should revise the figures, in current form figures are not crystal clear

7.      Materials section should revise and write the materials name and manufacturing company name with the country name

8.      Reference must be in advance

Author Response

Author Response File: Author Response.docx

Reviewer 2 Report

In my opinion the topic is interesting and important. However, the manuscript is not properly prepared. Majority of text concerns the general information about the ANN. The ANN is a statistical tool and it is commonly known and widely applied. An assessment of prediction errors is also very well known. It is nothing new. An application of the ANN in a proper way is the most important matter. The experimental data should be appropriately prepared and introduces. In the manuscript it is difficult to assess it because the data are presented in insufficient way. Authors do not present the numer of the investigated fabrics. The information about the fabrics being investigated is insufficient too. We do not know the raw material applied in the fabrics. In my opinion it is also important factor influencing the fabric shrinkage. 

In the Table 1 it is alack of units of particular parameters. Authors do not present the method of assessment of fabric shrinkage and also other structural parameters. 

Authors do not explain sufficiently why they applied such architecture of ANN - why 4 layers?

EPI and PPI are were introduced as one input parameter. Why? They are two separate parameters.

Weave structure should be also introduced as two parameters because in majority of cases repeat of weft and repeat of warp are different. 

The R and P values are not explained. Symbils in the equations (1), (2) (3) and (4) should be explained too. 

In table 3 the units should be presented. 

In Conclusions authors stated that the research presented confirmed that the ANN can be successfully used to predict the fabric shrinkage. However, they do not show any graph presenting the comparison of predicted and measured values of shrinkage. 

Author Response

Response to reviewers’ comments:

Reviewer 2:

In my opinion the topic is interesting and important. However, the manuscript is not properly prepared. Majority of text concerns the general information about the ANN. The ANN is a statistical tool and it is commonly known and widely applied. An assessment of prediction errors is also very well known. It is nothing new. An application of the ANN in a proper way is the most important matter. The experimental data should be appropriately prepared and introduces. In the manuscript it is difficult to assess it because the data are presented in insufficient way. Authors do not present the numer of the investigated fabrics. The information about the fabrics being investigated is insufficient too. We do not know the raw material applied in the fabrics. In my opinion it is also important factor influencing the fabric shrinkage. 

The fibre blends used was 96-98% cotton and 2-4% elastane. ANN literature is improved.

In the Table 1 it is alack of units of particular parameters. Authors do not present the method of assessment of fabric shrinkage and also other structural parameters. 

Added the units of all parameters.

Authors do not explain sufficiently why they applied such architecture of ANN - why 4 layers?

4 layers were recognised as minimum requirement (doing various trails previously) to study 5 input parameters effect on one output parameter.

EPI and PPI were introduced as one input parameter. Why? They are two separate parameters.

EPI and PPI are different but they are taken together to obtain the effect of construction parameters as a whole.

Weave structure should be also introduced as two parameters because in majority of cases repeat of weft and repeat of warp are different. 

Here we are trying to study the weave structure as a whole. In further other study we can try to evaluate them separately and determine its effect.

The R and P values are not explained. Symbils in the equations (1), (2) (3) and (4) should be explained too. 

The definitions and units are now provided.

In table 3 the units should be presented.

Units are added now. 

In Conclusions authors stated that the research presented confirmed that the ANN can be successfully used to predict the fabric shrinkage. However, they do not show any graph presenting the comparison of predicted and measured values of shrinkage. 

The metrics were obtained from the error analysis of the developed machine learning model between the predicted and measured values which shows error within the acceptable bounds. “r” value also represents the correlation coefficient.

Author Response File: Author Response.docx

Reviewer 3 Report

The introductory part is written clear and interesting. It is waiting that a serious discussion on the stretch fabrics will follow at the modern mechanics and/or mathematics level.

The authors prefer to use the artificial intellect approach. Why not. In this case, the reader is waiting for, first, the clear formal mechanical and mathematical statement of the problem - the artificial intellect methods are mathematical methods and the problem of control/improving the stretch properties of the fabrics is the problem in mechanics.

After that, the reader is waiting for the clear formal mechanical and mathematical solution to the problem.

It seems to me, the discussion of the problem, after the good introductory part, is off the mathematics and mechanics standard. I recommend presenting the solution to the problem in the standard mathematics/mechanics manner.

I strongly recommend adding an example which demonstrates the control/improving stretch properties of the fabric with the explanation from the mechanics point of view, taking into account that the authors investigate (with artificial intellect or other methods) the mechanical properties of fabric

Author Response

Reviewer 3:

The introductory part is written clear and interesting. It is waiting that a serious discussion on the stretch fabrics will follow at the modern mechanics and/or mathematics level.The authors prefer to use the artificial intellect approach. Why not. In this case, the reader is waiting for, first, the clear formal mechanical and mathematical statement of the problem - the artificial intellect methods are mathematical methods and the problem of control/improving the stretch properties of the fabrics is the problem in mechanics. After that, the reader is waiting for the clear formal mechanical and mathematical solution to the problem. It seems to me, the discussion of the problem, after the good introductory part, is off the mathematics and mechanics standard. I recommend presenting the solution to the problem in the standard mathematics/mechanics manner. I strongly recommend adding an example which demonstrates the control/improving stretch properties of the fabric with the explanation from the mechanics point of view, taking into account that the authors investigate (with artificial intellect or other methods) the mechanical properties of fabric

This study involves developing a neural network algorithm to predict fabric shrinkage before actual manufacturing of the fabric. So, as to achieve final fabric close to what is demanded. In further studies mathematical model can be employed.

Please check that all references are relevant to the contents of the
manuscript.

Yes references are checked now.

Author Response File: Author Response.docx

Round 2

Reviewer 2 Report

The manuscript was improved significantly. Some of my remarks have been taken into account. 

Reviewer 3 Report

Since the corrections were done, I agree with the publication of this paper