Technical Program

Thanks to the community support, Haystack Connect 2017 has a full schedule of Keynote Speakers, Industry Presentations, Panel Discussions and Technical Sessions to bring everyone up-to-date on the latest trends and technologies.

Leveraging Haystack Tagging to Enable Root Cause Analysis at Scale

Leveraging Haystack Tagging to Enable Root Cause Analysis at Scale

Tuesday, May 9, 2017, 16:00 - 16:25
Breakout Sessions Track 3

Speaker Bio

BUENO connects to more than 250 buildings with a total floor area of 80M sq. ft. With the help of tagging, a centralised analytics library is applied across these buildings which have a wide range of functionality including shopping centres, schools and universities, hospitals, hotels and office buildings. Tagging must be robust in order to cope with these differences. Implementation of the Haystack tagging convention assists in standardising each building into a common model to which analytics can be applied.

However, the tagging process can present challenges. One of the big challenges BUENO faces is the large number of points that require tagging, a very time consuming and labour intensive process. To solve this we have developed an automatic tagging tool where we use machine learning techniques based on an ever-growing library of tagged points to determine which points need to be tagged and which tags to apply. In addition to the initial tagging of a site, the auto-tagging tool ensures continual improvement in the quality of tagging by calculating the accuracy of previously tagged sites.

The application of tags does not necessarily lead to producing useful analytics and so we must continue learning from the data, to extend on the information provided by the tags. As a part of this process we have developed an automated asset register which stores meta-data on individual pieces of equipment. This meta-data can include information such as previously calculated coefficients used for data models or historical averages and peak values, which provides great benefits to our analytics, as well as calculated performance indicators allowing us to dynamically track asset performance.

These learning processes are essential to providing root cause data analytics, which require sufficient information to identify the underlying cause of an issue. This approach results in more efficient analytics, faster resolution time and better outcomes and has resulted in the resolution of more than 2000 root cause issues in Australian buildings over the past 12 months.