Effelinus, Marcellino (2024) Optimasi Desain Casing Electric Control Unit (ECU) Menggunakan Metode Backpropagation Neural Network (BPNN) dan Genetic Algorithm (GA). Other thesis, Institut Teknologi Sepuluh Nopember.
![[thumbnail of 5007201080-Undergraduate_Thesis.pdf]](http://repository.its.ac.id/style/images/fileicons/text.png) |
Text
5007201080-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 October 2026. Download (5MB) | Request a copy |
Abstract
Microcontroller merupakan bagian penting yang terdapat didalam Electric Control Unit (ECU) seperti microcontroller STM32 karena merupakan otak dari ECU yang berfungsi sebagai pengatur nilai input dan output. Dengan perkembangan teknologi yang berfokus pada pengurangan emisi, ECU menjadi salah satu inovasi yang dipasangkan dengan inovasi lainnya seperti penggunaan Dual Diesel Fuel (DDF). Pada temuan ini, bahan material dari casing ECU akan digantikan menggunakan bahan pilomer filamen yang sering digunakan dalam 3D printing. Penggantian material dari casing ECU menjadi bahan polimer filamen dilakukan untuk mengurangi biaya produksi yang mahal, material polimer yang digunakan tidak memiliki nilai ketahanan panas sekuat bahan seperti logam. Pergantian material pada casing ECU pernah diuji distribusi termalnya, sehingga dilakukan simulasi pada casing ECU. Dalam penelitian ini, dilakukan simulasi dengan Metode Computational Fluid Dynamics (CFD) dan Finite Element Methods (FEM) pada struktur dimensi casing ECU untuk mendapatkan nilai distribusi termal berupa temperatur maksimum dan regangan pada casing ECU menggunakan software ANSYS Workbench 2024. Penelitian dimulai dengan menentukan data dimensi pada casing ECU, kemudian melakukan desain model 3D casing ECU menggunakan software Solidworks 2022. Model 3D yang telah didesain akan disimulasikan menggunakan software ANSYS Workbench 2024 untuk menentukan nilai temperatur maksimum dan nilai strain. Simulasi dilakukan pada beberapa variasi dimensi yaitu panjang 120 dan 130 mm; lebar 100, 110, dan 120 mm; dan tebal 1.0, 1.2, 1.4, 1.6, 1.8, dan 2.0 mm. Dari hasil yang didapatkan, dilakukan analisa pada data dengan optimasi desain casing ECU menggunakan metode Backpropagation Neural Network (BPNN) kemudian dilanjutkan dengan metode Genetic Algorithm (GA). Desain optimum dari casing pada Electric Control Unit (ECU) akan disimulasikan kembali dengan ANSYS sebagai validasi hasil akhir. Hasil penelitian dari simulasi didapatkan 36 data yang akan dilakukan uji training pada BPNN untuk mendapatkan output net terbaik. Pada output temperatur maksimum net terbaik berupa 2 hidden layer, 6 neuron tiap hidden layer, dan activation function tansig dengan MSE sebesar 2,4 x 10-14 serta pada output regangan net terbaik berupa 5 hidden layer, 3 neuron tiap hidden layer, dan activation function tansig dengan MSE sebesar 1,4 x 10-10. Dengan menggunakan metode GA, diperoleh parameter terbaik pada variasi dimensi panjang 130 mm, lebar 120 mm, dan tebal 1,5 mm. Hasil prediksi BPNN-GA untuk output temperatur maksimum adalah 326,9708 K dan output regangan adalah 0,025039 mm, hasil ini kemudian akan disimulasikan dengan ANSYS dengan nilai temperatur maksimum 328,526 K dan regangan 0,024488. Dimana selisih error temperatur maksimum antara prediksi optimasi BPNN-GA dengan hasil validasi berupa simulasi adalah 0,4756% dan selisih error regangan adalah 2,2092%.
==============================================================================================================================
The microcontroller is an important part of the Electric Control Unit (ECU) such as the STM32 microcontroller because it is the brain of the ECU which functions as a regulator of input and output values. With technological developments that focus on reducing emissions, the ECU has become one of the innovations that is paired with other innovations such as the use of Dual Diesel Fuel (DDF). In this finding, the material of the ECU casing will be replaced using pilomer filament material which is often used in 3D printing. Changing the material of the ECU casing to filament polymer material was carried out to reduce expensive production costs, the polymer material used does not have as strong a heat resistance value as materials such as metal. The thermal distribution of the material in the ECU casing was tested, so a simulation was carried out on the ECU casing. In this research, simulations were carried out using Computational Fluid Dynamics (CFD) and Finite Element Methods (FEM) on the dimensional structure of the ECU casing to obtain thermal distribution values in the form of maximum temperature and strain in the ECU casing using ANSYS Workbench 2024 software. The research began by determining the data dimensions of the ECU casing, then design a 3D model of the ECU casing using Solidworks 2022 software. The 3D model that has been designed will be simulated using ANSYS Workbench 2024 software to determine the maximum temperature and strain values. Simulations were carried out on several variations of dimensions, namely lengths of 120 and 130 mm; widths 100, 110, and 120 mm; and thicknesses of 1.0, 1.2, 1.4, 1.6, 1.8, and 2.0 mm. From the results obtained, analysis was carried out on the data by optimizing the ECU casing design using the Backpropagation Neural Network (BPNN) method then continued with the Genetic Algorithm (GA) method. The optimum design of the casing on the Electric Control Unit (ECU) will be simulated again with ANSYS as validation of the final results. The research results from the simulation obtained 36 data which will be tested for training on BPNN to get the best net output. The best net maximum temperature output is in the form of 2 hidden layers, 6 neurons per hidden layer, and activation function tansig with MSE of 2.4 x 10-14 and the best net strain output is 5 hidden layers, 3 neurons per hidden layer, and activation Tansig function with MSE of 1.4 x 10-10. Using the GA method, the best parameters were obtained for varying dimensions of length 130 mm, width 120 mm and thickness 1.5 mm. The BPNN-GA prediction results for the maximum temperature output are 326,9708 K and the strain output is 0.025039 mm. These results will then be simulated with ANSYS with a maximum temperature value of 328,526 K and a strain of 0.024488. Where the maximum temperature error difference between BPNN-GA optimization predictions and validation results in the form of simulation is 0.4756% and the strain error difference is 2,2092%.
| Item Type: | Thesis (Other) |
|---|---|
| Uncontrolled Keywords: | Electric Control Unit, Computational Fluid Dynamics, Finite Element Methods, BPNN, GA |
| Subjects: | T Technology > T Technology (General) > T57.62 Simulation T Technology > T Technology (General) > T57.84 Heuristic algorithms. T Technology > TA Engineering (General). Civil engineering (General) > TA347 Finite Element Method T Technology > TA Engineering (General). Civil engineering (General) > TA357 Computational fluid dynamics. Fluid Mechanics |
| Divisions: | Faculty of Industrial Technology and Systems Engineering (INDSYS) > Mechanical Engineering > 21201-(S1) Undergraduate Thesis |
| Depositing User: | Marcellino Effelinus |
| Date Deposited: | 22 Aug 2024 03:23 |
| Last Modified: | 22 Aug 2024 03:23 |
| URI: | http://repository.its.ac.id/id/eprint/114372 |
Actions (login required)
 |
View Item |