Identifikasi Beban pada Sistem Tegangan Rendah Berbasis Harmonisa dan IoT dengan Artificial Neural Network secara Realtime

Nugroho, Irfan Purwito (2020) Identifikasi Beban pada Sistem Tegangan Rendah Berbasis Harmonisa dan IoT dengan Artificial Neural Network secara Realtime. Undergraduate thesis, Institut Teknologi Sepuluh Nopember.

[img] Text
07111340000113-Undergraduate_Thesis.pdf - Accepted Version
Restricted to Repository staff only

Download (1MB) | Request a copy

Abstract

Semakin banyaknya penggunaan beban non linier pada sistem tenaga listrik dapat menimbulkan berbagai masalah. Masalah yang sering terjadi antara lain meningkatnya rugi-rugi daya, meningkatnya temperatur konduktor kabel, adanya arus pada fasa netral, salah ukur pada KWH-meter, dan pendeknya usia peralatan listrik. Di sisi lain penggunaan peralatan listrik yang berlebihan sering dilakukan oleh pelanggan PLN. Satu diantara penyebab pemborosan penggunaan listrik ini adalah tidak termonitornya kondisi beban terpasang. Oleh karena itu, perlu dikembangkan peralatan monitoring beban yang bisa diakses secara realtime. Dalam tugas akhir ini dikembangkan simulasi model identifikasi beban yang terdiri dari microcontroller, sensor arus, dan model neural network yang kedepannya dapat digunakan untuk membantu dalam pembuatan alat monitoring beban secara realtime. Metode yang digunakan dalam pengembangan simulasi identifikasi dan monitoring ini adalah dengan mendeteksi profil harmonisa arus dari setiap jenis beban dan dilakukan identifikasi beban melalui model artificial neural network. Dilakukan beberapa percobaan konfigurasi parameter model ANN sehingga didapatkan model dengan keakuratan dan keefisienan tertinggi. Hasil dari tugas akhir ini diharapkan ditemukan konfigurasi ANN dengan keakuratan paling tinggi. Konfigurasi tersebut kedepannya bisa dimanfaatkan dalam pembuatan alat monitoring dan identifikasi beban berbasis harmonisa serta neural network secara realtime. ================================================================= The increasing amount of non-linear loads on the electric power system can cause various problems. Problems that often occur include power loss, temperature rise of the conductor cable, the presence of current in the neutral phase, a measurement error on the KWH-meter, and the short lifetime of the electrical equipment. On the other hand, the over limit usage on electrical equipment is more likely done by PLN customers. One of the causes of this wasteful use of electricity is that the installed load conditions are not monitored. Therefore, it is necessary to develop a load monitoring equipment that can be accessed in real time. In this final project a simulation model is developed that consists of a microcontroller, current sensor, and a neural network model which can be used in the future to assist in the production of load monitoring tools in real time. The method that being used in the development of this identification and monitoring simulation is to detect the current’s harmonisa profile of each load type and load identification is carried out using an artificial neural network model. Several configuration experiments on parameter of ANN model are conducted in order to acquire the highest accuracy and efficiency. The results of this final project are expected to find the highest accuracy for ANN configuration. In the future it can be used in production of monitoring tools and harmonisa based load identification as well as neural network based load in real time.

Item Type: Thesis (Undergraduate)
Uncontrolled Keywords: Harmonisa, Artificial Neural Network, Multi Layer Perceptron, Python, Internet of Things, Harmonics
Subjects: T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK3226 Transients (Electricity). Electric power systems. Harmonics (Electric waves).
Divisions: Faculty of Electrical Technology > Electrical Engineering > 20201-(S1) Undergraduate Thesis
Depositing User: Irfan Purwito Nugroho
Date Deposited: 20 Sep 2020 03:38
Last Modified: 20 Sep 2020 03:38
URI: https://repository.its.ac.id/id/eprint/81988

Actions (login required)

View Item View Item