Collaborative Big Data Analytics for Smart Buildings
Big Data combined with A.I for Smart Buildings is already making its way into the marketplace and it will soon become a necessity for buildings that truly wish to maintain comfort & control (health and wellness) for their occupants whilst also maintaining their sustainability targets. To achieve these diametrically opposed goals high-quality models are needed which must produce high-quality results to facilitate high-quality decisions and control.
The ease of collecting relevant data, along with the possibility to store it remotely on the cloud together with the processing applications, makes this hybrid data model almost inevitable. It’s ability to discover properties of the building otherwise hidden and to provide prediction (predictive control) mean you no longer have to react to events, you can see them coming!
That's called predictive control and predictive maintenance, an absolute must for a really smart building.
With predictive control you can leverage events in the future to maximize opportunities for your building or prepare your building well in advance of any negative effects.
Data Requirements for the Next Generation of Smart Buildings. Today's BMS systems now serve a multitude of functions from their primary design of control to data storage and dashboarding. Traditional BMS systems are now limited by their hardware capability and justly are not designed to perform advanced analytics, machine learning, and artificial intelligence. To perform these higher-level learnings data integrity is required where dropouts and artifacts common with BMS systems must be filtered out. In order to future proof your building, the data processed by your building must be managed and filtered such that it is reliable, trustworthy and of sufficient frequency to meet the dynamics of the building. In short without robust data integrity the value you can extract from your data for the benefit of your building is limited. To that end, BMS systems are not equipped for such learning or exploitation of the data and this is where platforms such as the Digital Building Platform from Future Decisions offer a new capability to bring the power of A.I. to your building in a BMS/Hardware independent way. The approach brings the power of enterprise-grade infrastructure for data management and security to BMS controls.
Establishing ground truth data is obtaining data that represents the correct operation of the building management system and labeling it as such. Such data integrity is an incredibly important step for any next-level data analytics, machine learning and the application of predictive controls. Ground truth data is very hard to ensure given the complexity of modern BMS systems which have strayed from their primary role. While various performance tests and commissioning tests aid in such integrity they are momentary snapshots and do not represent the rest of day, week, month or year.
However, with technology such as the Future Decisions Digital Building Platform continual monitoring is possible no matter the day or season as each piece of data is scrutinised. Augmenting such scrutiny is also the ability to compare and contrast if a set of buildings that collaborate have been classified as similar, which would increase the confidence of stating that a building operates correctly and provides benchmarks within the organization for performance indexing. Another important data process is data validation, which is ensuring that the data collected from data sources are clean and error-free. It is a process that is very tightly intertwined with fault detection and diagnosis, as faults can also result in incorrect data. The process of learning the building physics and operations takes a week up to a month however once the initial models are created their accuracy increases over time as they learn. Such capability significantly enhances the capability of your building operations from fault finding through to sustainable operations.
Prediction of energy consumption is important for planning and operation of electric utilities for energy suppliers, but also for building owners to increase buildings’ energy efficiency. The more data there is on similar buildings at different locations (i.e. operating under different circumstances, e. g. weather conditions), the more accurate models can be built, which would translate into more accurate predictions and higher quality decisions.
The remaining smart buildings data analytics processes can share the conclusions on the benefits of collaborative data processing, especially since most of them are data-based. In the following two subsections we detail the opportunities and models of collaboration.
To get all the possible benefits in a real Smart Building you would have A.I. and M.L. (artificial Intelligence and Machine learning). For that you need very good quality data to start with, you need it at sufficient frequency to meet the building physics and over large time scales so you can learn the building in different seasons to give some examples.
For a new scientific approach to meet the above-stated challenges a state of the art software platform has been created. Even though other systems (BMS etc.) will emphasize that secure, real-time (live), correct Big data from the BMS will show the system is working correctly, a BMS system is not designed to learn and act on the analytics or the correctness of the data.
For that, you need to check out the new innovative physics and scientific approach of Future Decisions
There is no point in having data you must extract the value from it !!