April 2016
Article
AutomatedBuildings.com

[an error occurred while processing this directive]
(Click Message to Learn More)


IoT Uprising or IoT Revolution?

Taming Unruly Data at Scale
Dr. Michael Georgescu
Dr. Michael Georgescu
Director of Research
Ecorithm


Articles
Interviews
Releases
New Products
Reviews
[an error occurred while processing this directive]
Editorial
Events
Sponsors
Site Search
Newsletters
[an error occurred while processing this directive]
Archives
Past Issues
Home
Editors
eDucation
[an error occurred while processing this directive]
Training
Links
Software
Subscribe
[an error occurred while processing this directive]

The Internet of Things (IoT) – a movement to digitize the engineered physical world – is changing how companies manage their assets and conduct business. As the IoT transformation takes place and new sources of information emerge, organizations are trying to harness these newly formed data pipelines. Just as the agricultural and industrial revolutions of centuries past transitioned society to new processes and production methods, the IoT revolution is reshaping how information is produced, distributed, and analyzed. To properly channel the transformative power of this revolution, we need ways to gather and analyze IoT data properly. However, this is much more easily said than done.

Many say the first application of the IoT was in commercial buildings – an Intranet of Things, if you will. Commercial buildings have been using Building Management Systems (BMSs) to collect sensor data from devices and operate control systems for many years. However, until recently, this data was only used for sporadic problem-solving at the device level – not for insight into overall system performance.

Fundamentally, buildings exist to provide a comfortable environment for occupants. More recently, it’s been acknowledged that this should be done as efficiently as possible. Seeing a large opportunity for operational improvements across the board, data analytics software companies are now collecting and analyzing building sensor data that was previously wasted. Through this analysis, companies are able to improve comfort and energy efficiency, predict when devices will need maintenance, and discover stability and efficiency issues in the the HVAC system.

Though commercial buildings were the ‘first to the party’ in the IoT world, they have also proven to be the toughest to analyze effectively.

Building analysis has two substantial challenges. The first is the task of automatically analyzing data from a complex building. Buildings are affected by internal and external conditions that are continually changing, such as occupancy and weather. These changes make maintaining a high level of comfort and efficiency at all times very difficult. Additionally, each building’s operation – both systemically and at the device level – has unique constraints, such as design limitations or equipment performance degradation. Even with the involvement of experienced engineers, using existing conventional analysis methods to optimize the performance of complex building systems is extremely difficult and time consuming – and definitely not fully automated.

The second critical obstacle building data analytics platforms face is repeating effective analysis at scale. Every building is its own special snowflake! No two buildings have the same system set-up, physical location, or occupancy pattern. Because buildings are so widely divergent in their operational characteristics, analytics must, inherently, be uniquely designed for each individual building in order to achieve valuable, targeted results. However, custom-tailoring software from scratch for every building is inhibitively expensive and time consuming. Therefore, what is required to enable scalable, software-driven building management? An analysis process that identifies the macro-level similarities of building system types (and utilizes them as a starting point), yet also distinguishes the micro-level differences of individual devices within a single building (in a few days, rather than many months).

Ecorithm’s platform overcomes these challenges of automation and scale. Through the use of two technologies – an analysis process known as the spectral method, alongside proprietary, fast-sampling algorithms – Ecorithm’s software automatically extracts problematic system behavior from large sensor networks, condensing the noisy information into the most important factors impacting performance. Ecorithm has built the rigorous expertise of the “building whisperer” – and their ability to identify critical system issues by seemingly holding a finger to the breeze – into the software, therefore eliminating the need to custom-tailor software from scratch. The net result is that Ecorithm’s software provides device-level specificity and system-level insights in a rapidly deployable and immensely scalable software platform.

[an error occurred while processing this directive]Let’s examine the first element of Ecorithm’s technology mentioned above, the spectral method. Building systems are an assortment of sensors, actuators, controllers, and parameters. Though their interactions are complicated, they are far from random. The spectral method is based on the principle that interacting systems create oscillations. By analyzing a system in terms of these oscillations, the relationships between individual devices – and the behavior characteristics of the whole system – are captured. To state it simply, the spectral method describes motion patterns of a system. Having the power to define the magnitude, frequency and phase* of these motion patterns unlocks a never-before-seen level of automated diagnostic accuracy.

Ecorithm’s fast-sampling algorithms – the second element of technology mentioned above – identify when basic, irregular conditions occur at the component level (e.g., sensors above/below setpoint, actuators that are under or over responsive, or control signals that reach saturation). Though individually these anomalies do not indicate an overt issue, identifying where they overlap at the device, system, and building level enables the accurate identification of faults or optimization opportunities – whether they be at the device or system level – and truly eliminates the false positives that are innate to solutions utilizing other types of analysis. In this way, Ecorithm’s fast-sampling algorithms automatically pinpoint the core, underlying cause of each fault and the source of each optimization opportunity.

As described above, quality analysis and diagnostics of buildings – or any complex system – is no easy task. The primary challenge for any IoT analytics platform is to correctly identify relevant patterns automatically, and in doing so, extract key insights about a particular system. Ecorithm’s platform is uniquely able to analyze IoT data and discover issues that were previously invisible, unlocking countless opportunities as systems migrate to the digitized machine world.

*Direction of the trend in relation to other system components and the outside world



About the Author

Dr. Georgescu is the Director of Research at Ecorithm, Georgescu completed his PhD under Prof. Mezic with his thesis on the Analysis of Systems in Building using Spectral Koopman Operator Methods. Dr. Georgescu’s background is centered on studying the complex behaviors exhibited in the building environment. Using spectral methods, these complex behaviors can be understood by measuring the oscillatory interactions that occur between various building systems. Georgescu has established the use of spectral methods in various areas of building analysis including design, fault detection, and modeling. At Ecorithm, these methods are utilized for automated analysis of the large datasets which are produced by building sensor networks to characterize and improve building operational efficiency.

footer

[an error occurred while processing this directive]
[Click Banner To Learn More]

[Home Page]  [The Automator]  [About]  [Subscribe ]  [Contact Us]

Events

Want Ads

Our Sponsors

Resources