Home / Current Issue / Paper 1703984
Machine Learning and Predictions on Effect of Fiber Size, Hybridization on the Mechanical Properties of Coir/Luffa Reinforced Hybrid Composite
Subject area: Science,Engineering and Technology · Area of research: Material Science
Abstract
This study is to further investigate at what percentage of coir/luffa fibers will generate hybrid composite with the best mechanical behavior while maintaining 30% fiber weight with 70% epoxy resin polymer. The composites were produced by varying the fiber weight fractions and the fiber size. Each composite comprises 30% fiber and 70% epoxy resin. At corresponding fiber sizes of 200 ?m, 400 ?m, and 600 ?m, hybridization of the fibers was carried out in the following weight ratios: 0:30, 10:20, 15:15, 20:10, and 30:0 wt/wt of Coir/Luffa. When the tensile strength, flexural strength, and impact strength of the composites were examined, it was found through controlled experiments that these properties increased as the fiber size increased. Additionally, it was found that sample R [15%wt Coir and 15%wt Luffa] provided the optimum mechanical qualities across a range of fiber sizes because the fibers were distributed evenly throughout the matrix. Machine learning tools such as artificial neural networks [ANN] and fuzzy logic designers were utilized to model and foretell the experimental outcomes of the hybrid composites. The model's output variables were tensile strength, flexural strength, and impact strength; the input variables were the weights of coir and luffa and the fiber size. Based on their coefficient of determination [R2], the two analyzed models' performances and appropriateness were compared. Tensile strength, flexural strength, and impact strength all have fuzzy logic coefficients of determination [R2] of 0.9752, 0.9773, and 0.9730, respectively. Similar to this, the ANN coefficients of determination [R2] for tensile strength, flexural strength, and impact strength are 0.9608, 0.904, and 0.9378, respectively. Based on this statistical analysis, the fuzzy logic designer produced a much more accurate prediction than the ANN in terms of the coefficient of determination [ R2] and mean square error [MSE] values.
Keywords
Fuzzy logic designers, Artificial Neural Network, Mean square error, Hybrid Composite, Coefficient of determination, contour plot
References
[1] I. G. Chibueze, C. U. Atuanya, C. C. Nwobi Okoye and C. M. Obele, Int. J. of Eng, Tech and Sci., 20, 106 (2021).
[2] Shiv Sharmaa, Santosh Tamanga, D. Devarasiddappab and M. Chandrasekrana, Pro. Mater. Sci., 6, 1805 (2014).
[3] K. R. Sumesh, K. Kanthavel, A. Ajithram and P. Nandhini, J. of Poly. Sci., 24, 014 (2019)
[4] S. Li, B. Ball and E. Donner, J. of Poly. Sci., 76, 1257 (2018).
[5] M. J. Sheykh, A. Tarmian and K. Doosthoseini, J. of Poly. Sci., 74, 4537 (2017).
[6] K. R. Sumesh, K. Kanthavel and S. Vivek, J. of Mat. Res, 6, 29 (2019).
[7] R. Rahman , S. Hamdan and E. J. Amani, J. of Materials., 76, 763 (2014).
[8] S. Vivek, and K. Kanthavel, J. of Comp. Part B., 160, 170 (2018).
[9] H. P. Abdul Khalil, M. Masri, C. K. Saurabh, M. R. Fazita, A. A. Azniwat and N. A. Sri Aprilia, J. of Mat. Res. Exp., 16, 170 ( 2017).
[10] V. Arumuga Prabu, S. Thirumalai Kumaran and M. Uthayakumar, Part. Sci. Tech., 36, 402 (2012).
[11] M. M. Kabir, H. Wang, K. T. Lau and F. Cardona, J. of Comp. Part B., 43, 2883 (2012).
[12] A. Muhammad, R. Rahman and S. Hamdan, Poly. Bull., 4, 203 (2015).
[13] J. F. Pereira, D. P. Ferreira and J. Bessa, J. of Poly. Com., 4, 60 (2015).
[14] M. Mittal and R. Chaudhary Mat. Res. Exp., 43, 2183 (2018).
[15] A. Nopparut and T. Amornsakchai, J. of Poly. Test., 52,141 (2016).
[16] H. Ishikawa, H. Takagi and A. N. Nakagaito, J. of Comp. Inter, 21, 329 (2014).
[17] J. P. Patel and P. H. Parsania, Poly. Bull., 74, 485 (2016).
[18] E. Manaila, M. Daniela, G. Craciun and L. Surdu, Poly. Bull., 71, 2001 (2014).
[19] A. K. Balan, S. M. Parambil, S. Vakyath, J. T. Velayudhan, J. of Mat. Sci.,52, 6709 (2017).
[20] N. Saba, O. Y. Alothman , Z. Almutairi, M. Jawaid, J. of Poly. Sci., 201, 138(2019).
[21] D. Sreeramulu, and N. Ramesh, Mat. Tod. Proc. 5, 6518 (2018).
[22] J. Lu, P. Askeland and L. Drzal, J. of Poly. Sci., 49, 1285 (2008).
[23] D. M. Krishnudu, D. Sreeramulu and P. V. Reddy Inter. J. of Eng., 11, 567 (2019).
[24] D. Mohana Krishnudu, D. Sreeramulu and P. V. Reddy, J. of Nat. Fibers., 1, 10 (2019).
[25] B. V. Ramnath, B. J. Kokan and R. Niranjan Raja, J. Of Mat. Des., 51, 357 (2013).
[26] M. Ramesh, K. Palanikumar and K. H. Reddy, J. of Comp. Eng., 4, 9 (2013).
[27] M. Boopalan, M. Niranjanaa and M. J. Umapathy, J. of Comp. Eng., 51, 54 (2013).
[28] Sahas Bansal, M. Ramachandran and Pramod Raichurkar, J. Poly. Bull., 23, 406 (2017).
[29] C. C. Nwobi-Okoye, I. E. Umeonyiagu and C. G. Nwankwo, Nig. J. of Tech., 32, 13 (2015).
[30] C. C. Nwobi-Okoye and I. E. Umeonyiagu, Afri. J. of Sci. Tech. Inno and Deve. 8, 264 (2016).
[31] C. C. Nwobi-Okoye and I. E. Umeonyiagu, Niger. J. Tech. Res., 8, 22 (2013).
[32] C. C. Nwobi-Okoye, I. E. Umeonyiagu and C. G. Nwankwo, Niger. J. Tech. Res., 32, 13 (2015).
[33] C. C. Nwobi-Okoye and I. E. Umeonyiagu, Niger. J. Tech., 29, 55 (2015).
[34] S. J. Russell and P. Norvig, “Pearson Educational Inc”, 10, 45 ( 2003).
[35] C. U. Atuanya, C. C. Nwobi-Okoye and O. D. Onukwuli, Inter. J. of Mech. and Mats. Eng., 7, 20(2014).
[36] B. S. Keerthi Gowda, G. L. Easwara and R. Velmurugan, J. of Soft. Comp. in Civil. Eng., 4, 79 ( 2020).
[37] L. Zadeh,” Information and control”, 8, 353 (1965).
[38] M. C. Anukwonke, I. G. Chibueze and N. N. Nnuka, Int. J. of inn. Eng. Tech. and Sci., 4, 25 (2021).
[39] A. Mirsoleimani, A. Amooey, S. Ghasemi and S. Salkhordeh, Ind. and Eng. Chem. Res., 54, 9844 ( 2015).
[40] A. MOSLEMI, International Conference: Papers, Forest Pro and Res Society., 3, 42 9 (1989).
[41] T. A. Negawo, Y. Polat and F. N. Buyuknalcaci, J. of Comp. and Struct., 207, 589 (2019).
How to cite this paper
@article{1703984,
author = {Chibueze Ikechukwu Godswill, Anukwonke Maxwell .C., Imah Adindu, Umebamalu Chinedu, Otamiri Humble. C},
title = {Machine Learning and Predictions on Effect of Fiber Size, Hybridization on the Mechanical Properties of Coir/Luffa Reinforced Hybrid Composite},
journal = {Iconic Research And Engineering Journals},
year = {2023},
volume = {6},
number = {7},
pages = {372-389},
issn = {2456-8880},
url = {https://www.irejournals.com/formatedpaper/1703984.pdf},
abstract = {This study is to further investigate at what percentage of coir/luffa fibers will generate hybrid composite with the best mechanical behavior while maintaining 30% fiber weight with 70% epoxy resin polymer. The composites were produced by varying the fiber weight fractions and the fiber size. Each composite comprises 30% fiber and 70% epoxy resin. At corresponding fiber sizes of 200 ?m, 400 ?m, and 600 ?m, hybridization of the fibers was carried out in the following weight ratios: 0:30, 10:20, 15:15, 20:10, and 30:0 wt/wt of Coir/Luffa. When the tensile strength, flexural strength, and impact strength of the composites were examined, it was found through controlled experiments that these properties increased as the fiber size increased. Additionally, it was found that sample R [15%wt Coir and 15%wt Luffa] provided the optimum mechanical qualities across a range of fiber sizes because the fibers were distributed evenly throughout the matrix. Machine learning tools such as artificial neural networks [ANN] and fuzzy logic designers were utilized to model and foretell the experimental outcomes of the hybrid composites. The model's output variables were tensile strength, flexural strength, and impact strength; the input variables were the weights of coir and luffa and the fiber size. Based on their coefficient of determination [R2], the two analyzed models' performances and appropriateness were compared. Tensile strength, flexural strength, and impact strength all have fuzzy logic coefficients of determination [R2] of 0.9752, 0.9773, and 0.9730, respectively. Similar to this, the ANN coefficients of determination [R2] for tensile strength, flexural strength, and impact strength are 0.9608, 0.904, and 0.9378, respectively. Based on this statistical analysis, the fuzzy logic designer produced a much more accurate prediction than the ANN in terms of the coefficient of determination [ R2] and mean square error [MSE] values.},
keywords = {Fuzzy logic designers, Artificial Neural Network, Mean square error, Hybrid Composite, Coefficient of determination, contour plot},
month = {January},
}