Claim Missing Document
Check
Articles

Found 2 Documents
Search

Rancang Bangun Pengendali dan Pemantau Suhu pada Dua Inkubator Telur Ayam Otomatis Rudi Darmawan; Sri Ratna Sulistiyanti; Syaiful Alam
Jurnal Ilmiah Teknik Pertanian - TekTan Vol 5 No 3 (2013)
Publisher : Jurusan Teknologi Pertanian

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.25181/tektan.v5i3.848

Abstract

In this research, it was designed a temperature controller and monitoring system of  two automatic chicken eggs incubator. Based on references, hatched chicken eggs require temperature 38oC-40oC and relative humidity about 55-65%. The temperature sensor used is LM 35 and humidity sensor used is DHT 11. Temperature controller uses microcontroller ATmega 328 P. The temperature is controlled by switching on/off method on incandescent bulbs used as heating tool and DC fan as cooling tool. Monitoring system uses a computer with LabVIEW software and LCD 16x2 for incubator. Wheel shelf of eggs will automatically rotate egg on 06.00 A.M. at 0o and 18.00 P.M. at 60o using servo motor. Communication used by two incubator to transmit data to computer is uses I2C method. LM35 temperature sensor is calibrated by using thermometer and the obtained error is 0.285714%, while DHT 11 humidity sensor is calibrated by Hygrometer and the obtained error is 1.5%. The result of system test shows that the average stability of temperature on incubator 1 and 2 is 39.01oC and the average humidity is 60%. Incubator monitoring system of temperature and humadity is displayed with real time uses  LabVIEW software. Keywords: eggs incubator, controlling, monitoring, temperature, ATmega 328P
Studi Komparasi Fungsi Aktivasi Sigmoid Biner, Sigmoid Bipolar dan Linear pada Jaringan Saraf Tiruan dalam Menentukan Warna RGB Menggunakan Matlab Ikhwan Pamungkas; Sumadi Sumadi; Syaiful Alam
Jurnal Serambi Engineering Vol 7, No 4 (2022): Oktober 2022
Publisher : Fakultas Teknik

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.32672/jse.v7i4.4776

Abstract

Abstract Neural network Backpropagation is a good method to use to determine RGB color (Red, Green, Blue) because it can give high accuracy values. Neural network backpropagation there are several activation functions that can be used. This study aims to determine the activation function and to form the optimal network architecture in the backpropagation in determining RGB colors. Neural network model backpropagation was developed using 3 types of activation functions, namely Binary Sigmoid, Bipolar Sigmoid, and Linear. Based on the training results, the Bipolar Sigmoid activation function produces the highest accuracy value compared to the Binary Sigmoid activation function and the Linear activation function. The optimal network architecture is modeled using 3 nodes in the input layer, 2 hidden layers consisting of 2 hidden layer nodes, and 1 node in the output layer. In the model that has been made, the percentage of network training and testing accuracy is 100% resulting in the smallest MSE with a value of 6,1E-03 in the 97th iteration in 485 milliseconds..Keywords: activation function, artificial neural network, backpropagation, RGB. AbstrakJaringan saraf tiruan backpropagation merupakan salah satu metode yang baik digunakan untuk menentukan warna RGB (Red, Green, Blue) karena dapat memberikan nilai akurasi yang tinggi. Pada jaringan saraf tiruan backpropagation terdapat beberapa fungsi aktivasi yang dapat digunakan. Penelitian ini bertujuan untuk mengetahui fungsi aktivasi serta membentuk arsitektur jaringan yang optimal pada jaringan saraf tiruan backpropagation dalam menentukan warna RGB. Pada penelitian ini, model jaringan saraf tiruan backpropagation dikembangkan menggunakan 3 jenis fungsi aktivasi yaitu Sigmoid Biner, Sigmoid Bipolar, dan Linear. Berdasarkan hasil pelatihan, fungsi aktivasi Sigmoid Bipolar menghasilkan nilai akurasi tertinggi dibandingkan fungsi aktivasi Sigmoid Biner dan fungsi aktivasi Linear. Arsitektur jaringan yang optimal dimodelkan dengan menggunakan 3 node pada input layer, 2 hidden layer yang terdiri dari masing-masing 2 node hidden layer, dan 1 node pada output layer. Pada model yang telah dibuat, persentase akurasi pelatihan dan pengujian jaringan adalah sebesar 100% menghasilkan MSE terkecil dengan nilai 6,1E-03 pada iterasi ke-97 dalam waktu 485 milidetik.Kata Kunci: fungsi aktivasi, jaringan saraf tiruan, backpropagation, RGB.