Prediksi Kuat Tekan Beton berdasarkan Variasi Komposisi menggunakan Jaringan Syaraf Tiruan

Hutapea, Josua Margandatua (2023) Prediksi Kuat Tekan Beton berdasarkan Variasi Komposisi menggunakan Jaringan Syaraf Tiruan. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 02311740000093-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 September 2025. Download (3MB) | Request a copy

Abstract

Telah dilakukan penelitian mengenai jaringan syaraf tiruan dengan metode backpropagation yang digunakan untuk memprediksi kuat tekan beton berdasarkan komposisi beton. Penelitian diawali dengan mengagregat informasi komposisi semen dalam rentang 100- 475 kg/m3 dan usia beton antara 7-50 hari. Selanjutnya, data training dibagi menjadi tiga level kuat tekan, yaitu level rendah dengan rentang 0-15 MPa, level menengah dengan rentang 16- 35 MPa, dan level tinggi dengan rentang 36-50 MPa. Kemudian, dibuat model jaringan syaraf tiruan dengan 10 variasi hidden layer dan 3 variasi nodes. Selanjutnya dilakukan perbandingan nilai MAE (Mean Absolute Error) dari model-model jaringan syaraf tiruan tersebut. Setelah itu, diambil beberapa model dengan 4 variasi hidden layer, 3 variasi jumlah node, dengan pembagian 3 level kuat tekan beton. Model-model ini digunakan untuk memprediksi kuat tekan beton, dan hasilnya kemudian dibandingkan dengan nilai aktual dari pengukuran kuat tekan beton. Lalu, diambil satu model jaringan syaraf tiruan dengan nilai MAE terkecil dan hasil prediksinya mendekati hasil aktual pengukuran kuat tekan beton. Hasil penelitian menunjukkan bahwa jaringan syaraf tiruan yang memiliki nilai MAE terendah adalah jaringan syaraf tiruan pada level kuat tekan rendah (0-15 MPa), dengan hidden layer sejumlah 8 buah, dengan jumlah node pada masing-masing hidden layer sejumlah 64. Nilai MAE yang didapat adalah 0.3414.
==================================================================================================================================
Research has been conducted on artificial neural networks using the backpropagation method to predict the compressive strength of concretes based on concrete composition. The study began by aggregating data on cement composition ranging between 100-475 kg/m3 and the age of concrete ranging between 7-50 days. Subsequently, the data is then categorized into three levels of compressive strength, low level ranging between 0-15 MPa, medium level ranging between 16-35 MPa, and high level ranging between 36-50 MPa). Following data preparation, a series of artificial neural network models are constructed, featuring ten variations of hidden layers and three variations of nodes. These models are then compared based on their respective Mean Absolute Error (MAE) values. Furthermore, a subset of models is selected, encompassing four variations of hidden layers and three variations of nodes, with each model categorized according to the three levels of concrete compressive strength. These chosen models are deployed to predict concrete compressive strength, and their predictions are then compared against actual measured values. After that, one model exhibiting lowest MAE value and whose predicted value are close to the actual results of compressive strength measurements was chosen. The results obtained from this study are that the artificial neural network model that has the lowest MAE value is the model trained using the low compressive strength level data, with 8 hidden layers, and 64 nodes in each hidden layer. The MAE value obtained is 0.3414.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Beton, Jaringan Syaraf Tiruan, Kuat Tekan, TensorFlow; Concrete, Artificial Neural Network, Compressive Strength
Subjects:	Q Science > QA Mathematics > QA76.87 Neural networks (Computer Science) T Technology > TA Engineering (General). Civil engineering (General) > TA439 Lightweight concrete. High strength concrete.
Divisions:	Faculty of Industrial Technology and Systems Engineering (INDSYS) > Physics Engineering > 30201-(S1) Undergraduate Thesis
Depositing User:	Josua Margandatua Hutapea
Date Deposited:	14 Sep 2023 04:23
Last Modified:	14 Sep 2023 04:23
URI:	http://repository.its.ac.id/id/eprint/103281

Actions (login required)

View Item