This demonstration aims to leverage existing machine learning technology to enhance Department of Defense energy meter data analysis and its usage for portfolio planning. This project will showcase practical use case examples through an accessible online data visualization platform with the aim of integrating this functionality into the Army’s Meter Data Management System (MDMS). The technical focus of this project lies in rectifying data quality issues, such as data gaps, outliers, and faulty data, through the application of multiple data imputation and machine learning techniques. Then, with the curated data the project will leverage machine learning algorithms to enrich the integrity and reliability of the dataset by developing predictive models for improved estimated energy usage for unmetered buildings. The overarching goal of the project aligns with future roadmap efforts to harness machine learning in MDMS data applications, with an emphasis on prioritizing data cleansing for critical functions such as predictive forecasting and fault detection.

Machine learning, a subset of artificial intelligence learns from data to enhance performance without explicit programming. It operates on pattern recognition, adapting and refining predictive models for increased efficiency. Utilizing this technology for data quality enhancement involves a dynamic process of refining and correcting inaccurate or incomplete data points. Leveraging algorithms and various models involving pattern recognition within the dataset enables the technology to distinguish valid data from anomalies caused by errors, malfunctions, or inconsistencies. Through techniques like imputation and regression analysis, this project aims to have more complete datasets. This approach optimizes data integrity and empowers organizations to make informed decisions and draw meaningful insights from their information assets.

Higher quality meter data with fewer gaps offers several significant benefits to the Army. It enables more accurate analysis and decision-making by providing a complete and reliable picture of energy consumption patterns to aid with higher precision building energy use intensity for the Army portfolio. This, in turn, enhances the effectiveness of energy management strategies, leading to optimized resource allocation and reduced operational costs. Additionally, consistent, high-fidelity meter data quality is a requirement to be able to leverage commercial off-the-shelf software to provide higher-level analysis such as fault detection diagnostics and predictive forecasting. (Anticipated Project Completion - 2027)