Sistem Kontrol Neural Network PID (NNPID) pada Inverter Baterai Terintegrasi IoT untuk Impor–Ekspor Energi

Widigda, Hosea Derius (2026) Sistem Kontrol Neural Network PID (NNPID) pada Inverter Baterai Terintegrasi IoT untuk Impor–Ekspor Energi. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5022221033-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (11MB) | Request a copy

Abstract

Pengelolaan energi pada sistem smart grid memerlukan sistem kontrol yang adaptif dan andal, khususnya pada inverter baterai yang berperan penting dalam proses impor dan ekspor energi. Inverter baterai konvensional umumnya masih memiliki keterbatasan dalam kemampuan adaptasi terhadap perubahan kondisi operasi, pemantauan parameter secara real-time, serta pengaturan arus pengisian dan pengosongan baterai yang optimal. Keterbatasan ini mendorong perlunya pendekatan kontrol yang lebih cerdas dan adaptif untuk menjamin kinerja sistem secara menyeluruh. Penelitian ini mengusulkan penerapan sistem kontrol Neural Network PID (NNPID) pada inverter baterai yang terintegrasi dengan Internet of Things (IoT), khususnya untuk mengoptimalkan proses ekspor energi ke jaringan, yang dilengkapi dengan manajemen kontrol pengisian pada mode impor guna mendukung operasi dual-mode secara menyeluruh. Sistem kontrol NNPID dirancang untuk mengendalikan daya ekspor baterai LiFePO4 secara real-time berdasarkan sinyal error daya, sementara parameter State of Charge (SoC), arus, tegangan, dan suhu baterai dipantau melalui Battery Management System (BMS) sebagai basis pengambilan keputusan mode operasi dan proteksi sistem, serta berinteraksi dengan programmable regulator dan grid-tied inverter. Mikrokontroler ESP32 berfungsi sebagai pusat pemrosesan data yang menjalankan algoritma NNPID sekaligus mengelola komunikasi IoT melalui antarmuka berbasis web. Sensor daya PZEM DC dan AC digunakan untuk pemantauan parameter kelistrikan secara real-time pada sisi baterai maupun sisi jaringan. Hasil pengujian menunjukkan bahwa pada mode ekspor, sistem NNPID mampu mengikuti setpoint daya pada rentang 100 hingga 600 W dengan steady state error yang konsisten di bawah 1,6% dan rise time stabil (19–42 detik), tanpa indikasi osilasi yang signifikan. Performa ini mengungguli PID konvensional yang mengalami perlambatan rise time hingga 110 detik serta kontrol Fuzzy yang mengalami kegagalan konvergensi dan osilasi liar pada setpoint tinggi. Pada mode impor, sistem pengisian baterai dengan kontrol proporsional mampu mempertahankan akurasi dengan steady state error di bawah 3,8% pada rentang setpoint 50–250 W. Integrasi IoT memungkinkan proses pemantauan dan pengendalian sistem secara real-time melalui protokol komunikasi terbuka, sehingga mendukung pengelolaan energi yang lebih efektif pada lingkungan smart grid.
==================================================================================================================================
Energy management in smart grid systems requires an adaptive and reliable control system, particularly for battery inverters that play a crucial role in energy import and export operations. Conventional battery inverters still exhibit limitations in adapting to dynamic operating conditions, real-time parameter monitoring, and regulation of battery charging and discharging currents. These limitations necessitate a smarter and more adaptive control approach to ensure overall system performance. This research proposes the implementation of a Neural Network PID (NNPID) control system for an IoT-integrated battery inverter, specifically to optimize the energy export process to the grid, complemented by a dedicated charging control scheme for the import mode to support overall dual-mode operation. The NNPID controller is designed to regulate export power from a LiFePO4-based battery system in real time based on the power error signal, while State of Charge (SoC), battery current, voltage, and temperature are monitored through the Battery Management System (BMS) as the basis for operating-mode decisions and system protection, and the controller interacts with a programmable regulator and a grid-tied inverter. An ESP32 microcontroller serves as the central processing unit, executing the NNPID algorithm and managing IoT communication through a web-based interface. PZEM DC and AC power sensors are employed for real-time monitoring of electrical parameters on both the battery and grid sides. Test results demonstrate that in export mode, the NNPID system tracks power setpoints from 100 to 600 W with a consistent steady-state error below 1.6% and a stable rise time ranging from 19 to 42 seconds, without significant oscillation. This performance outperforms conventional PID, which suffers from a degraded rise time up to 110 seconds, and Fuzzy control, which exhibits convergence failure and wild oscillations at higher setpoints. In import mode, the battery charging system with proportional control maintains accuracy with a steady-state error below 3.8% across the tested setpoint range of 50–250 W. IoT integration enables real-time monitoring and control through open communication protocols, supporting more effective energy management in smart grid environments.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Neural Network PID, Inverter Baterai, IoT, Smart Grid, SCADA, Neural Network PID, Battery Inverter, IoT, Smart Grid, SCADA
Subjects:	A General Works > AI Indexes (General) A General Works > AI Indexes (General) Q Science > Q Science (General) > Q325.5 Machine learning. Support vector machines. T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK1007 Electric power systems control T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK2692 Inverters T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK3070 Automatic control
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Electrical Engineering > 20201-(S1) Undergraduate Thesis
Depositing User:	Hosea Derius Widigda
Date Deposited:	21 Jul 2026 06:09
Last Modified:	21 Jul 2026 06:09
URI:	http://repository.its.ac.id/id/eprint/135897

Actions (login required)

View Item