Sistem Monitoring dan Peramalan Intensitas Cahaya Matahari Berbasis Internet of Things Menggunakan Metode SARIMA

Sari

Abstrak: Intensitas cahaya matahari merupakan parameter lingkungan yang dinamis dan dipengaruhi oleh perubahan waktu serta kondisi atmosfer. Penelitian ini bertujuan membangun sistem monitoring intensitas cahaya berbasis Internet of Things (IoT) serta menerapkan analisis time series menggunakan model Seasonal Autoregressive Integrated Moving Average (SARIMA) untuk peramalan intensitas cahaya. Sistem mengintegrasikan sensor BH1750, mikrokontroler ESP32, Firebase Realtime Database, dan web dashboard. Data dikumpulkan selama 30 hari (9 Februari-10 Maret 2026) dengan frekuensi lima kali per hari sehingga diperoleh 150 observasi, dan diuji menggunakan metode Black Box Testing. Sensor BH1750 memperoleh tingkat akurasi 93,70%. Model SARIMA(0,0,0)(1,1,1)[5] terpilih sebagai model terbaik dengan nilai Akaike Information Criterion (AIC) sebesar 2.568,98. Uji Ljung-Box menghasilkan p-value 0,4726 dan 0,7616, menunjukkan residual model memenuhi asumsi white noise. Evaluasi model menghasilkan Mean Absolute Percentage Error (MAPE) sebesar 10,08% dan Root Mean Square Error (RMSE) sebesar 1.613,54 lux, dengan peramalan tujuh hari ke depan menunjukkan pola musiman yang konsisten. Disimpulkan bahwa sistem mampu melakukan monitoring secara real-time serta menghasilkan peramalan dengan akurasi yang baik. Berbeda dengan penelitian terdahulu yang umumnya hanya berfokus pada monitoring, penelitian ini menawarkan kebaruan berupa integrasi monitoring IoT dan peramalan SARIMA.
Kata kunci: Intensitas Cahaya. Internet of Things, Monitoring, SARIMA, Time Series

Abstract: Sunlight intensity is a dynamic environmental parameter influenced by temporal changes and atmospheric conditions. This study aims to develop an Internet of Things (IoT)-based sunlight intensity monitoring system and apply time-series analysis using the Seasonal Autoregressive Integrated Moving Average (SARIMA) model to forecast sunlight intensity. The system integrates a BH1750 sensor, an ESP32 microcontroller, a Firebase Realtime Database, and a web dashboard. Data were collected over 30 days (February 9 – March 10, 2026) at a frequency of five readings per day, yielding 150 observations, and the system was validated using Black Box Testing. The BH1750 sensor achieved an accuracy rate of 93.70%. The SARIMA(0,0,0)(1,1,1)[5] model was selected as the optimal model, with an Akaike Information Criterion (AIC) value of 2,568.98. Ljung-Box tests yielded p-values of 0.4726 and 0.7616, indicating that the model residuals satisfied the white noise assumption. Model evaluation resulted in a Mean Absolute Percentage Error (MAPE) of 10.08% and a Root Mean Square Error (RMSE) of 1,613.54 lux, with a seven-day forecast demonstrating a consistent seasonal pattern. It is concluded that the system is capable of real-time monitoring and generating forecasts with good accuracy. Unlike previous studies that generally focused solely on monitoring, this research offers novelty through the integration of IoT monitoring and SARIMA forecasting.
Keywords: Light Intensity, Internet of Things, Monitoring, SARIMA, Time Series