RANCANG BANGUN SMART ENERGY METER BERBASIS IOT UNTUK PERAMALAN KONSUMSI KWH LISTRIK PRABAYAR DENGAN ALGORITMA XGBOOST

    Muhamad Ajis, - and Muhammad Rizalul Wahid, - and Dewi Indriati Hadi Putri, - (2025) RANCANG BANGUN SMART ENERGY METER BERBASIS IOT UNTUK PERAMALAN KONSUMSI KWH LISTRIK PRABAYAR DENGAN ALGORITMA XGBOOST. S1 thesis, Universitas Pendidikan Indonesia.

    Abstract

    Pertumbuhan konsumsi listrik rumah tangga yang pesat akibat kemajuan teknologi dan perubahan gaya hidup menekankan pentingnya perencanaan energi yang efisien. Smart Energy Meter (SEM) yang dikombinasikan dengan machine learning menawarkan solusi yang menjanjikan untuk mengoptimalkan penggunaan listrik melalui analisis prediktif. Penelitian ini mengusulkan penerapan algoritma Extreme Gradient Boosting (XGBoost) untuk memprediksi konsumsi listrik per jam berdasarkan data historis SEM yang dikumpulkan antara bulan April hingga Juni 2025. Alur pemodelan mencakup tahap prapemrosesan data, rekayasa fitur, seleksi fitur berdasarkan ambang korelasi, serta optimasi hyperparameter menggunakan Optuna dengan pendekatan pencarian Bayesian. Model akhir menunjukkan akurasi tinggi dengan nilai MAE sebesar 0,0115, RMSE sebesar 0,0150, dan R² sebesar 0,9803 menggunakan cross validation TimeSeriesSplit dengan 5 fold. Eksperimen prediksi selama satu minggu menunjukkan model memiliki performa yang kuat, namun tetap sensitif terhadap kejadian eksternal yang tidak berulang seperti hari libur nasional. Hasil ini menegaskan potensi XGBoost sebagai alat yang efektif untuk perencanaan konsumsi listrik rumah tangga, serta pentingnya mempertimbangkan faktor kontekstual eksternal dalam pengembangan selanjutnya. ----- The rapid growth in household electricity consumption, driven by technological advancements and changing lifestyles, underscores the need for efficient energy planning. Smart Energy Meters (SEMs) combined with machine learning offer a promising solution to optimize electricity usage through predictive analytics. This study proposes the implementation of Extreme Gradient Boosting (XGBoost) to forecast hourly electricity consumption using historical SEM data collected between April and June 2025. The modeling workflow involved data preprocessing, feature engineering, feature selection based on correlation thresholds, and hyperparameter optimization using Optuna with Bayesian search. The final model achieved high accuracy with an MAE of 0.0115, RMSE of 0.0150, and R² of 0.9803 using five-fold TimeSeriesSplit cross-validation. A one-week forecasting experiment demonstrated the model's robustness but also revealed sensitivity to non-recurring external events such as national holidays. These results highlight the potential of XGBoost as a powerful tool for residential electricity planning and emphasize the importance of incorporating external contextual factors in future enhancements.

    [thumbnail of S_MKB_2101942_Title.pdf] Text
    S_MKB_2101942_Title.pdf

    Download (331kB)
    [thumbnail of S_MKB_2101942_Chapter1.pdf] Text
    S_MKB_2101942_Chapter1.pdf

    Download (336kB)
    [thumbnail of S_MKB_2101942_Chapter2.pdf] Text
    S_MKB_2101942_Chapter2.pdf
    Restricted to Staf Perpustakaan

    Download (826kB)
    [thumbnail of S_MKB_2101942_Chapter3.pdf] Text
    S_MKB_2101942_Chapter3.pdf

    Download (999kB)
    [thumbnail of S_MKB_2101942_Chapter4.pdf] Text
    S_MKB_2101942_Chapter4.pdf
    Restricted to Staf Perpustakaan

    Download (2MB)
    [thumbnail of S_MKB_2101942_Chapter5.pdf] Text
    S_MKB_2101942_Chapter5.pdf

    Download (263kB)
    [thumbnail of S_MKB_2101942_Appendix.pdf] Text
    S_MKB_2101942_Appendix.pdf
    Restricted to Staf Perpustakaan

    Download (1MB)
    Official URL: https://repository.upi.edu/
    Item Type: Thesis (S1)
    Additional Information: https://scholar.google.com/citations?user=Yb_xiagAAAAJ&hl=en&authuser=1 ID SINTA Dosen Pembimbing: Muhammad Rizalul Wahid: 6780434 Dewi Indriati Hadi Putri: 6720737
    Uncontrolled Keywords: XGBoost, Peramalan, Smart Energy Meter, Time Series, IoT XGBoost, Forecasting, Smart Energy Meter, Time Series, IoT
    Subjects: T Technology > TA Engineering (General). Civil engineering (General)
    T Technology > TK Electrical engineering. Electronics Nuclear engineering
    Divisions: UPI Kampus Purwakarta > S1 Mekatronika dan Kecerdasan Buatan
    Depositing User: Muhamad Ajis
    Date Deposited: 21 Aug 2025 03:13
    Last Modified: 21 Aug 2025 03:21
    URI: http://repository.upi.edu/id/eprint/135449

    Actions (login required)

    View Item View Item