Microsoft has a lot of technologies which have no consumer facing counterparts. Some of these technologies are Microsoft Azure, Azure Machine Learning, and Power BI. These technologies are rooted in big data and the cloud. Companies all over the world deal with massive volumes of data and need a way to make sense of it all. Most consumers never see these technologies implemented or how much they affect their daily lives.
One such example of Microsoft technologies being used to improve systems, is at Carnegie Mellon University (CMU). In a blog post, Microsoft went through how CMU used a variety of Microsoft technologies to save money by making their buildings more efficient. CMU gathered data from sensors in the heating ventilation and air conditioning (HVAC) and in the plumbing combined with weather data to predict the ambient temperature of their buildings. Once they have modeled how the temperature will change they can adjust the heating or cooling systems to achieve the perfect temperature.
This is a very complicated problem. The solution is not simply gathering more data. At some point a decision needs to be made on how to adjust the systems to control the temperature. This is where Microsoft comes in. First CMU gathered all of the relevant data concerning building temperature such as: outside temperature, sunlight, internal temperature, how long it took to affect the internal temperature, how heat moves about the building, etc. From there CMU used Microsoft Azure to predict the data they needed but didn’t have directly. For example the team didn’t have a record or table of how much sun hit the building, so they used Azure machine learning to predict it.
Then the CMU team built a model in Azure to target a specific temperature for the start of the work day. Controlling the building’s systems this way could lead to a 20% reduction in energy costs. For a large institution, reduced energy costs means massive savings. The team at CMU mention that working with Microsoft Azure and the easy to use machine learning tools enabled them to get off the ground quickly. Creating models for predicting sunlight on buildings could be done in days instead of weeks or months. This enabled the team to test the models faster and reap the benefits sooner.
