The household power consumption forecasting and anomaly detection model presented in this work focused mostly on advanced machine learning techniques, such as Long Short-Term Memory (LSTM) for forecasting and Isolation Forest for anomaly identification. The primary goal is to optimize energy management by providing accurate insights into power consumption and proactively identifying irregularities in power consumption patterns. This work unfolds in several key phases, encompassing preprocessing, data exploration, model construction, testing, and analysis. In the initial phase, the dataset is thoroughly examined, revealing time-stamped power consumption data that serves as the basis for subsequent time series analysis. The existing landscape is identified as lacking robust forecasting and anomaly detection systems, prompting the adoption of advanced techniques to overcome conventional limitations. The proposed system introduces the LSTM model and the isolation forest to enhance forecasting accuracy and anomaly detection, respectively. This choice is driven by LSTM’s ability to capture temporal dependencies with an RMSE value of 0.0799, making it well-suited for analyzing power consumption patterns. The advantages of the proposed system include improved forecasting accuracy, with 4.99% of identified anomalies, and adaptability to dynamic changes in power consumption behavior. It is stated that the candidate model gave successful results. The model can also respond to the power consumption variations of a dynamic nature, which makes it applicable in actual environment.