In the field of smart metering, guaranteeing the authenticity and dependability of data is important to efficient energy management, consumption forecasting, and invoicing. In this research, our work is focused upon detecting the irregularities in smart meter data. For example, we applied three independent algorithms, namely Isolation Forest, One-Class SVM, and Density-Based Spatial Clustering of Applications with Noise (DBSCAN), for anomaly detection over time based on Received Signal Strength Indicator (RSSI) and Signal-to-Noise Ratio (SNR) readings received from smart meters. Our research started with extensive data preprocessing, which included normalization of the data and imputation of missing values. We then used each of the anomaly detection models. The Isolation Forest algorithm successfully detected outliers through the distribution of the data. One-Class SVM algorithm classified the anomalies as those data that were on the edge of the normal data and did not touch the normal data and while with the DBSCAN algorithm we identified as anomalies those points that were not assembled in a densely clustered point of normal data. Models were evaluated based on precision, recall, F1-score, and ROC AUC. The analysis revealed that each of the models had distinctive capabilities in detecting different kinds of anomalies, such as irregular RSSI and SNR values and deviations from expected signal patterns. The Isolation Forest and One-Class SVM models were resilient in identifying prominent errors whereas the ability of DBSCAN in detecting dispersed and vulnerable associated defects were revealed. The results shown the importance to apply multiple models for catch with the greater diversity of anomalies that could appear in smart meter data. Results provide technical guidance on how the surveillance and maintenance of smart meter network can be improved, and thus how to have more accurate and reliable energy consumption data.