Intelligent decision support system for predicting and managing Building Energy Performance over whole life cycle of buildings

I. Scientific excellence

This research covers two major areas of current engineering research, Building operations and management and Energy Efficiency.
Buildings represent a large portions of the world’s consumption and associated CO2 emissions, hence, prediction of building energy consumption is vital for decision making in order to reduce energy consumption and lower CO2 emissions.

Currently, throughout the Built Environment there are huge discrepancies in predicted and actual energy utilisation making it incredibly difficult for facility managers to accurately plan building designs, operational costs and any refurbishment or maintenance costs. This is largely due to inaccurate regulatory models, inconsistencies between digital design and as-built data, client energy systems not understood or effectively embedded and poor operational commissioning. In addition, Demand Side Response (DSR) energy practices are an emerging trend. Organisations who can be flexible with their energy (reducing or shifting their consumption) can engage with their suppliers on reward based contracts, subsequently reducing their energy costs.

Novel Method based on machine learning data analytics to provide accurate forecasts and enable energy performance management through scenario planning simulation and smart alerting notifications. The research will develop a novel predictive algorithm to predict energy loads and consumption using historical data from BMS (Building Management System) and smart metres. The predicted profiles will be compared with real time energy load and any diversion will be picked up by intelligent systems and suggestions for corrective action to reduce energy consumption (i.e. real time energy load exceeds predicted one) or investigate any performance issues by building assets.

II. Clear aim and hypothesis

Aims
To develop and implement an intelligent energy performance management platform to
• simulate energy profiles,
• predict consumptions and
• compare them with real-time set-points to enable appropriate energy performance decisions to be made over life cycle of buildings.

Hypothesis
• data- driven building energy predictions models provide more accurate energy prediction than physical simulation models
• Energy prediction errors are reduced using short term energy prediction models as opposed to long term energy prediction.
• Better prediction models are obtained by including occupant behaviour
• reduced energy costs are obtained by shifting demand response using decision support system.


III. Methodology and innovations

The approach to collect historical energy load data from BMS and smart meters and test different prediction models using Matlab software as development platform. Different profiles will be evaluated using 'absolute mean error' and it is anticipated that a mix for different algorithms will be used to predict the loads. Further, decision support systems based on comparing predicted modes against real time to advice on approaches to bring real time under control and within 2-3% of the predicted profile.

IV. Strategic relevance

The work is of a strategic relevance to the UK and the EU in their drive to bring energy consumption under control and ultimately reduce energy consumption and produce accurate prediction tools to be used for demand response programmes.

V. Interdisciplinary and fit with relevant DTA programme

This is an interdisciplinary project as it brings control engineering, big data analytics and facility management to develop a more reliable algorithms and decision support system for energy management.

Applications

Applicants must apply using the online form on the University Alliance website at https://unialliance.ac.uk/dta/cofund/how-to-apply/. Full details of the programme, eligibility details and a list of available research projects can be seen at https://unialliance.ac.uk/dta/cofund/

The final deadline for application is Monday 8 October 2018. There will be another opportunity to apply for DTA3 projects in the spring of 2019. The list of available projects is likely to change for the second intake.