Disaggregation Of Loads
Electricity Circuits for the Purposes of Billing (2016)
A range of sophisticated algorithms to disaggregate lighting loads on electrical circuits such that run times could be supplied to an automated M&V system.
Project OverviewThe Client
The client is an energy efficiency building retrofit company operating across the US. The client uses a scalable Measurement and Verification system that can be applied to small businesses.
The client was interested in a tool that would enable the creation of a business model to facilitate the large pool of SME customers interested in the value of its lighting retrofit program but didn't have the initial down payment to invest in the program.
The precise and cost-effective disaggregation system developed by Full Stack Energy enabled advanced audit analysis. The patent pending disaggregation system developed by Full Stack Energy in conjunction with the client allowed customers and utilities to be confident with the savings presented in every M&V project.
Business Process Development, Machine Learning, QA Testing
Swift, Mathematica, Python, AWS, Grafana, InfluxDB, MySQL, MQTT
Full Stack Energy provided a team of four specialists - a Python, AWS Developers, a Machine Learning Engineer, a Data Processing Engineer and an Embedded Hardware Engineer.
The Disaggregation System was envisaged as a way of calculating the probability of a particular fixture being on or off at a specific point during the day and subsequently using this information to make predictions through machine learning where sub-metering would not be feasible to install.
Using advanced disaggregation algorithms customer lighting usage and savings are calculated and presented in a bill that can be generated at any time.
Deep knowledge of the energy sector enabled the team and the client to quickly collaborate on innovative solutions. Using advanced disaggregation algorithms, customer lighting usage and savings is calculated and presented in a bill that can be generated at any time.
Machine learning leverages historical data from circuits and sites across the country to train a system such that it will be able to recognise system occurrences, raise alerts, auto classify conditions and identify trends.
Metered data analyzed by disaggregation system to identify runtimes and expressed as switch-level savings.
Savings are aggregated and presented in a shared savings bill to the customer each month for the project term.