Fatah, Daffa Kurnia (2025) Integrasi Sistem Prediksi Daya Listrik untuk Manajemen Energi Rumah Cerdas Berbasis Internet Of Things dan Machine Learning. Other thesis, Insitut Teknologi Sepuluh Nopember.
![]() |
Text
02311940000068-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 April 2027. Download (19MB) | Request a copy |
Abstract
Efisiensi dalam manajemen energi listrik rumah tangga melibatkan pengelolaan energi yang lebih efisien dan akurat agar tidak adanya pemborosan daya. Dalam penelitian ini, teknologi Internet of Things (IoT) dan algoritma Long Short-Term Memory (LSTM) dipadukan untuk menciptakan suatu sistem yang dapat memprediksikan dengan akurasi optimal, yang mampu memperoleh data secara langsung. Dalam sistem ini, digunakan hardware berbasiskan microcontroller ESP32 dan sensor PZEM-004T, untuk mendapatkan dan menganalisis daya yang digunakan menggunakan antarmuka website. Adapun arsitektur dari model LSTM yang digunakan terdiri dari 3 lapisan yaitu 128 unit Bidirectional LSTM, 64 unit, 32 unit LSTM dan terdapat Dropout serta Batch Normalization disetiap lapisan untuk mencegah terjadinya overfitting. Model ini dilatih menggunakan dataset yang terdiri dari barang elektronik yang berbeda di dalam rumah, seperti AC, kipas angin, dan alat elektronik lainnya selama dua bulan. Pengujian model ini menggunakan R2 Score menunjukkan akurasi masing-masing model adalah 0.989 (AC), 0.999 (kipas angin), dan 0.991 (perangkat lainnya). Selain itu sistem ini juga dilengkapi dengan website monitoring tools yang memungkinkan pengguna untuk melihat konsumsi energi secara real time, memeriksa data sebelumnya, serta mendapatkan peramalan konsumsi energi untuk 24 jam kedepan.
==============================================================================================================================
Efficiency in household electrical energy management involves more efficient and accurate energy management to avoid power wastage. In this research, Internet of Things (IoT) technology and the Long Short-Term Memory (LSTM) algorithm are combined to create a system that can predict with optimal accuracy, capable of acquiring live data. In this system, hardware based on the ESP32 microcontroller and PZEM-004T sensor is used, to obtain and analyze the power used using a website interface. The architecture of the LSTM model used consists of 3 layers, namely 128 Bidirectional LSTM units, 64 units, 32 LSTM units and there is Dropout and Batch Normalization in each layer to prevent overfitting. The model was trained using a dataset consisting of different electronic items in the house, such as air conditioners, fans, and other electronic devices for two months. Testing this model using R2 Score shows the accuracy of each model is 0.989 (AC), 0.999 (fan), and 0.991 (other devices). In addition, the system is also equipped with website monitoring tools that allow users to view energy consumption in real time, check previous data, and get energy consumption forecasting for 24 hours.
Item Type: | Thesis (Other) |
---|---|
Uncontrolled Keywords: | Daya Listrik, Internet of Things (IoT), Long Short-Term Memory (LSTM), Pemantauan Real-Time, Prediksi Konsumsi Energi, Sensor PZEM-004T, Electric Power, Energy Consumption Prediction, PZEM-004T Sensor, Real-Time Monitoring |
Subjects: | T Technology > T Technology (General) T Technology > T Technology (General) > T57.5 Data Processing |
Divisions: | Faculty of Industrial Technology and Systems Engineering (INDSYS) > Physics Engineering > 30201-(S1) Undergraduate Thesis |
Depositing User: | Fatah Daffa Kurnia |
Date Deposited: | 04 Feb 2025 04:48 |
Last Modified: | 04 Feb 2025 04:48 |
URI: | http://repository.its.ac.id/id/eprint/118101 |
Actions (login required)
![]() |
View Item |