Yazar "Aldaoud, Ahmed Ghanim Daoud" seçeneğine göre listele

Listeleniyor 1 - 1 / 1
  • Küçük Resim
    Öğe
    Selecting the ideal approach for rationalization of electrical energy consumption in smart building management using machine learning
    (İstanbul Gedik Üniversitesi, 2023) Aldaoud, Ahmed Ghanim Daoud; Ulutagay, Gözde; Gümüş, Tuğbay Burçin
    The infrastructure development of smart cities has gained a lot of attention in recent years, where energy efficiency was the major problem that researchers aim to improve. Anomaly detection for energy consumption was one of these issues, and it was a critical element to take into account while operating effective energy-saving systems, lowering the overall energy use and carbon emissions. Consequently, suggesting a potent method based on the Internet of Things (IoT) can have more relevance for recognizing aberrant usage in buildings and offering this information to customers and governments so that it can be handled in a correct way to lower the payments. As a result, this thesis proposes an optimization method to detect anomalies by using the LSTM algorithm and investigates three optimization algorithms of ADAM, AadMax, and Nadam. Two statistical modeling's are used for time series forecasting of ARIMA and SARIMAX. Anomaly detection results indicate that using LSTM with Nadar gives the best results, where MSE and RMSE achieved 0.15348 and 0.02356 respectively. Also, using ARIMA model gives the best overall results with AIC and MSE values of 0.13859 and 300.94365 respectively. This confirms the reliability and adaptability of the proposed model in optimizing anomaly detection.