Rancang Bangun Monitoring Photovoltaic Berbasis Internet Of Things Untuk Deteksi Dan Klasifikasi Fault Menggunakan Algoritma Random Forest

Ranggana, Chesta Adabi (2026) Rancang Bangun Monitoring Photovoltaic Berbasis Internet Of Things Untuk Deteksi Dan Klasifikasi Fault Menggunakan Algoritma Random Forest. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5009221018-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (5MB) | Request a copy

Abstract

Sistem photovoltaic (PV) rentan terhadap berbagai gangguan (fault) yang dapat menurunkan performa dan efisiensi produksi energi listrik, sehingga diperlukan deteksi dan klasifikasi fault secara otomatis untuk menjaga keandalan instalasi PV. Penelitian ini merancang dan membangun sistem monitoring PV berbasis Internet of Things (IoT) yang terintegrasi dengan algoritma Random Forest untuk mendeteksi dan mengklasifikasikan delapan kelas kondisi, yaitu normal, open circuit, tiga varian partial shading, dan tiga varian Loose Connection Fault. Mikrokontroler ESP32 digunakan sebagai unit akuisisi data dengan sensor PZEM-017, pyranometer SN-300AL, DS18B20, dan NTC 10K B3950, yang dikirimkan melalui protokol MQTT ke broker HiveMQ Cloud, diproses oleh Node-RED pada Google Cloud Platform (GCP), disimpan di InfluxDB Cloud, dan divisualisasikan melalui dashboard Grafana Cloud. Model Random Forest dilatih menggunakan 8.675 data dengan sembilan fitur input, kemudian di-deploy melalui Flask API sehingga hasil prediksi kondisi fault dapat ditampilkan secara real-time. Seluruh sensor terkalibrasi dengan akurasi di atas 95%, dan performa jaringan memenuhi standar TIPHON dengan packet loss 0%, delay 272,24 ms, dan jitter 85,94 ms. Hasil pengujian menunjukkan akurasi model sebesar 99,31% dengan macro average F1-score 0,986, di mana seluruh kelas mencapai F1-score di atas 97%. Analisis SHAP mengonfirmasi bahwa model membedakan tiap kondisi berdasarkan penanda fisik yang sesuai. Penelitian ini berkontribusi pada pencapaian Sustainable Development Goals (SDGs), khususnya SDG 7 (Affordable and Clean Energy) dan SDG 9 (Industry, Innovation and Infrastructure).
=====================================================================================================================================
Photovoltaic (PV) systems are susceptible to various fault conditions that can degrade performance and reduce electrical energy production efficiency, making automatic fault detection and classification essential to maintain PV installation reliability. This study designs and builds a PV monitoring system based on the Internet of Things (IoT) integrated with the Random Forest algorithm to detect and classify eight condition classes, namely normal, open circuit, three variants of partial shading, and three variants of Loose Connection Fault. An ESP32 microcontroller is used as the data acquisition unit with PZEM-017, SN-300AL pyranometer, DS18B20, and NTC 10K B3950 sensors, transmitting data via the MQTT protocol to the HiveMQ Cloud broker, processed by Node-RED on Google Cloud Platform (GCP), stored in InfluxDB Cloud, and visualized through a Grafana Cloud dashboard. The Random Forest model was trained using 8,675 data points with nine input features, then deployed via a Flask API so that fault condition predictions can be displayed in real time. All sensors were calibrated with an accuracy above 95%, and the network performance met the TIPHON standard with 0% packet loss, 272.24 ms delay, and 85.94 ms jitter. Testing results show a model accuracy of 99.31% with a macro average F1-score of 0.986, where all classes achieved an F1-score above 97%. SHAP analysis confirmed that the model distinguishes each condition based on its corresponding physical indicators. This research contributes to the achievement of the Sustainable Development Goals (SDGs), particularly SDG 7 (Affordable and Clean Energy) and SDG 9 (Industry, Innovation and Infrastructure).

Item Type:	Thesis (Other)
Uncontrolled Keywords:	photovoltaic, deteksi fault, Internet of Things, Random Forest, monitoring real time, SDG 7, SDG 9, photovoltaic, fault detection, Internet of Things, Random Forest, real-time monitoring, SDG 7, SDG 9
Subjects:	T Technology > T Technology (General) > T58.8 Productivity. Efficiency
Divisions:	Faculty of Industrial Technology and Systems Engineering (INDSYS) > Physics Engineering > 30201-(S1) Undergraduate Thesis
Depositing User:	Chesta Adabi Ranggana
Date Deposited:	03 Aug 2026 04:05
Last Modified:	05 Aug 2026 02:01
URI:	http://repository.its.ac.id/id/eprint/142024

Actions (login required)

View Item