How do you convert incompatible, disorganized, or too much data into actionable insights?

Excerpt from the Dubai Solar Show 2020

4 min read

The data your asset creates is a very valuable resource; unless it can help create meaningful reports and a clear forecast of the environment, then the data stands meaningless. Unorganized huge sets of data make it practically impossible for organizations to make accurate conclusions regarding their customers’ needs, thereby creating a thread of poor decisions that have little basis on facts. Like many other industries that heavily depend on consumer data, the solar sector is also bombarded with mountains of data now and then, that they don’t know what to do with, but which they are afraid of letting go lest it proves useful in the future. As this unorganized data is fast becoming an unsustainable burden to the solar sector, it can definitely be foreseen as a serious hurdle to progress in the near future. Owing to the negative effect of unorganized data, digitalization has come to the forefront. Data-led O&M was the topic of discussion at the Dubai Solar Show 2020. This exhibition aimed at achieving integration between all sectors of the energy industry and bringing together trade and technical sectors whilst facilitating the creation of new trade opportunities, enhancing an exchange of ideas and experiences, and showcasing the latest innovations.


Watch the excerpt

“In my opinion, the approach to data organization is what is critical,” begins Rahul Sankhe, President and Co-Founder of SenseHawk, and one of the speakers at the Dubai Solar Show 2020. He emphasized that, as a basic organizing framework for data, digital twins are pretty much the talk of the solar industry today. “We know that the data is collected from different sources, might be a SCADA system, a drone, some other sensor, or even from on-field human activities. As long as you can automatically organize this data in a way that it’s organized or indexed to a digital twin of the asset, it becomes easier to mine for better insights, and correlating various variables that could be impacting performance of the plant,” adds Rahul. Data organization is not just about deploying new technologies, but also about deploying new ways of organizing, managing, and recording all information.

Quality is the Key

“According to me, automatically correcting bad-quality data is the key, and we have worked on it very strongly,” cites Daniel Ramirez Watson, head of Business Development at Isotrol, and one of the speakers at the Dubai Solar Show 2020. “In the end, data gathering, correcting, and ensuring that you have the correct data is always the first step. However, to make any KPI calculation, if you’re starting with bad-quality data, or incomplete data, your KPIs and everything else that is going to follow behind is not going to be accurate. It’s not going to be representing the reality of the plant,” Daniel exclaims. Talking of some of the ways to ensure correct quality data, he adds, “count with the correct censoring of the plant, backup checks like drone inspections, and having a robust SCADA system that is able to identify by comparing similar wind turbines, similar string, similar inverter behavior, and is able to fill in these gaps either automatically with algorithms that are set up with per customer requirements for industry good practices, or even be able to do it manually.”

Daniel further says, “During the system, the data is yours. The customer owns the data, why will they not be able to go in and change it? If they see a specific current reading, a temperature reading, or anything in their plant that is not correct, they have to be able to go in there, modify this data, and recalculate the KPIs and all the indicators that they have for data analytics.” While the data organization challenges can take a backseat with all these viable solutions, obtaining huge amounts of data from the systems with a lot of readings, and identifying spurious data keeps standing at the long-standing queue of challenges. 

Machine Learning: A Way Ahead

Out of all of the thousands and millions of data points that the solar sector is getting every day and every hour, having systems in place to identify incorrect readings or things amiss, is the next key. Rahul says, “We must use machine learning based approaches and models to model the right kind of behavior, and flag off any exceptions that are not genuine. Spurious data streams are an important element of the approach.” Comparing to the usages of advanced analytics, artificial intelligence, machine learning techniques, and more in the wind industry, Daniel mentions that the PV industry is still struggling with the data acquisition and data quality. “Prior to performing advanced analytics, a lot of simpler things like corrected identification of alarms and having a very strong knowledge of the different OEMs, need to be done. Because if you don’t have a strong knowledge on how the inverter or wind turbine has to behave, or what their correct power curve is, you don’t have a benchmark. So having a really strong knowledge of how the plant can perform, how the operation is managed, and the kind of issues (clipping, soiling, tracker blockage, tracker deviation, string disconnection) that you can have on operations and everything else will eventually help you to model the data and run these algorithms.” This will help humans, who otherwise would not be able to manage this humongous amount of data and a huge amount of signals, alarms. 

SenseHawk has excelled AI-based data management. You can take advantage of it by contacting us for a free consultation.

SenseHawk

SenseHawk

SenseHawk, Inc. is a provider of cloud based drone data analytics and productivity tools for the solar industry. With customers in 20 countries, SenseHawk solutions are used by solar & EPC companies for terrain analytics, construction monitoring, thermography and task management on solar sites
[addtoany]

Sign up for insights and opinions from the best in the solar.

Leave a Comment