Reliability Evaluation of Advanced Metering Infrastructure Based on Isolation Forest Anomalies Detection

The advanced metering infrastructure (AMI) have been widely installed in building sector especially in university. The data of AMI can represent information in real time not only about electricity consumption data but also as an indicator of communal behaviour, cultural population and economic dynam-ics to derive many insights. The purpose of the evaluation is to analyses performed by the smart meter and to ensure that measurement services are available to views that operational technology infrastructure is reliable. Raw data from database will be transformation into information and knowledge which represented to transform into wisdom. This paper presents a big data mining cross industry standard process for data mining method for reliability evaluation analysis. The data modelling is used machine learning by isolation forest anomalies outlier to represented anomaly detection which using many features con-textual and variations. The reliability index parameters for the research results show that the AMI of SBM-ITB is trustworthy with a failure rate of 0.14 per day and the service availability of the smart metering system is 99.4 percent in a test period of 5199 hours.