Buildings are generators of vast amounts of data: from Building Energy Management Systems (BEMS), smart metering and sub-metering information (demand), IoT device information (sensing/control), distributed generation (RES), storage, and electric vehicle data, altogether characterized by continuously increasing growth rate, multi-diverse spatiotemporal resolutions and huge volume.
As discussed above, data come both from the building domain and from other diverse sources (energy systems, energy market, IoT devices, historical data, APIs, etc.). In the case of the BEYOND platform, the idea is that it will address numerous business and optimization needs for a variety of stakeholders involved in the value chain. The most promising value of big data derived from buildings environments is sequestrated in sharing such targeted information with energy market stakeholders and actors that we saw in our previous blog post.
BEYOND: A platform with a dual purpose
Here, it would be useful to remind what BEYOND project is doing. Our objective is to deliver a Big Data Management Platform, on top of which an advanced AI analytics toolkit will be offered allowing for the delivery of derivative data and intelligence out of a blend of real-life building data and relevant data coming from external sources.
The BEYOND platform will allow energy stakeholders to gain access to building data as well as advanced building data analytics and build their own applications and solutions, towards:
- Providing energy services to the building sector
- Improving their business processes and operations.
On the other hand, buildings will enjoy a a bundle of energy services and benefits in the form of:
- Optimized energy performance and reduction of energy costs
- New data-driven business models for financial gains stemming from data monetization or trading of building flexibility
What do we want from data to have?
To work properly, such a platform must adhere to high privacy and security requirements as well as to requirements related to performance, data management and access methods. It’s obvious that there is a need for having a clearly defined model, which can “host” and “integrate” existing data elements. The answer lies at a model known as CIM (Common Information Model: a good read about energy CIMs can be found here) which is a synthesis of other models/standards/etc. This sets some needs which are covered by the different data models to be considered, to the extent this is possible. Which are they? Lets’ look!
Interoperability: There is a high need for the overall system to work with data that is interoperable and that could be used to support different analyses and be able to turn data into insights making them digestible by the different systems. As such, the different models to be used should be in a position to allow for the easy and fast data transformation and/or linking in case this is needed.
Performance: Data must be easily created and consumed in order to serve the requirements set by a big data infrastructure, which supports both on-premises and cloud-based operations. Data transfer performance between different components which may not be placed in the same places is also important. So, it’s apparent that lightweight models would be in a better position to facilitate these purposes due to the reduced header payload they are carrying.
Security: Data security is an aspect that is of high importance to data owners who want to have strong guarantees on how their data are protected. Therefore, data models must support security methods that can guarantee end-to-end security, by utilizing methods that could verify the integrity of the data. Moreover, as encryption might be utilized to secure data, there is a need to choose data structures which can be efficient during crypto-operations in terms of performance, and that would not produce a big resource-utilization footprint.
Privacy: Data models and standards should support the operations of privacy relevant operations (such as anonymization, or pseudo-anonymization, data-obfuscation) without however losing the quality information of the payload is another important need