Data Analytics: The Key to Unlocking the True Power of the IoT
- Machines connected pre-IoT too, but data not utilized to fullest potential
- Data Analytics: With power of analytics, descriptive, diagnostic and even predictive decisions possible
- Both historical analytics & real time analytics aid in enterprise digital transformation
- How a global Pump manufacturer used the power of analytics to enhance business models and customer experience
Why the real benefits of the Internet of Things come from Data driven decision making rather than connectivity.
When the machines are connected to the internet, you start getting the data. But one may ask, my machines were connected to systems before as well and I was getting the data before IoT also. So, what has changed with IoT? Why does the data matter now more than ever? True. Earlier the machines have been connected to some extent and much of this data has existed in some shape or form, but the data was available in siloed manner or was locked in incompatible systems. This data was only available at a machine level and accessible to a few individuals only making it difficult to make sense at the plant level. As a result, the data has never been fully utilized as part of a broader analytics effort to foster predictive maintenance, or to optimize energy usage of factory assets, or to respond to critical events like an overheating or pump failure to minimize production loss. With IoT, the whole equation has changed. Since all your machines are connected to the internet and the data is available in real-time, now everyone (depending on the individual role) from shop floor to boardroom has access to same “Real” time data to help make an informed decision. The data is not only helping you in Descriptive (What happened) or Diagnostic (Why it happened) Analysis, but also to understand behavioral patterns to get Predictive (What will happen) and Prescriptive (What actions to take). So far, factories were relying on a small set of siloed data but with IoT and Big Data, you have access to the entire dataset, real-time with technology that helps you deep dive into analytics.
Hence, the big data analytics play an important role, in putting machine data to work and transforming this data from arcane strings of numbers into actionable information for the user. Analytics are equally important from a business point of view. As, while reports say the IoT market is estimated to have a compound annual growth rate (CAGR) of 26.9% the Analytics market is estimated to have a CAGR of 31%. If decision makers want to ensure that their current and future IoT projects deliver timely returns, their focus should be on applying better analytics.
Analytics can be broadly divided into traditional or historic data analytics and Real Time data analytics. Traditional analytics are used to find patterns in long term data. It can identify inefficiencies and help design better systems. It can compare the performance of systems across locations aiding standardization, reduce the risks on investments and help businesses find the areas best suited to the application of resources. Forbes Insights has released an informative brief with 5 reasons why industrial companies should use Advanced Analytics to reduce risks. According to Forbes:
- Analytics leaders report 15% gains in operating margins and revenues
- Analytics can significantly improve risk profiles
- Decision makers can model future requirements to take proactive actions
- Routine decisions can be delegated to computers so executives can focus on high-value projects
- Managers can more quickly and easily optimize resources to increase production and improve efficiency
To understand, how data analytics are changing business models and customer experience, let us take an example of the world’s largest pump and compressor manufacturer. They have thousands of water pumps installed in farmlands, many of them in rural and remote parts of India. These pumps are connected to IoT platform sending data realtime. Critical parameters like vibrations, the speed of water flow, electricity, and pump condition (Active/Inactive) are captured realtime. Alerts are sent based on pump conditions and a field engineer is informed beforehand. Now, having access to pump condition and its behavioral patterns, the field service engineer can do preliminary failure analysis before reaching the location and go with right tools to fix the problems. Cases which earlier used to take more than one visits are now reduced to Preventive and prescriptive alerts and to more actionable single visit.
So, how is such historical data put to use? The volume of data from IoT systems can be vast and relate to diverse parameters. Presented as-is, there is nothing to differentiate it from noise. To make sense of such data, it needs to be made available in a form that users can easily understand. Cloud computing and Big Data analytics can be used to condition this data and make it more accessible. For Instance, when a certain automotive manufacturer analyzed the data from their paint shop, they found more particle inclusions in the paint of certain arbitrary units. Using correlation algorithms to cluster the parameters measured with parameters from other machines in the plant, it was found that these specific units had passed by the sand blasting machine while it was in operation. While no particles from the machine directly came in contact with the affected units, the vibrations from the machine were causing a greater amount of inclusions. Data analysis revealed a causal link that would have been nearly impossible to identify manually.
“It’s easy to connect machines and collect data. What matters most is do you use the collected data to infer intelligence and derive actionable insights”, says Vinay Nathan, CEO Altizon systems. “Many enterprises have connected machines but they struggle with the right analytics tools to parse through the data and get insights. A lot of companies are just at causal analysis, they need to move to correlation using big data analytics.” With Real Time data, the challenge is not volume, but complexity. Once an IoT architecture has been established, it starts to generate huge amounts of data. As the number of devices and scale increases so does the complexity of the data generated. The interconnectivity of these systems also exponentially increases the kinds of insights that can be drawn from this data. Here the benefits of analytics come through Machine Learning and Complex Event Processing. You need to ensure the IoT Platform is not only capable of processing your big data with in-stream analytics but also has machine learning algorithms to add temporal analytics and make sense of processed data and bring patterns.
Real Time data is used to create an automated stimulus-response system, where changes in parameters trigger appropriate responses using technologies such as Edge and Fog computing. Machine Learning allows analytics to automatically infer when system parameters are out of bounds by comparing values to previously recorded ones thus improving response times and preventing cascading failures. Systems can be made context-aware so that they understand the relevance of the parameters they measure, thus automating process adjustments in real time, such that maximum efficiency can be maintained throughout the production cycle. In process intensive industries such as the Steel and Chemical Industries, it ensures production quality is always optimal.
Complex event processing is a method by which available information from various sources is analyzed in real time to correlate and infer vital intelligence which otherwise is not obvious to predict. For example, while it can be simple to note that HVAC units consume more power on hotter days, on these same days productivity in more humid work environments is lower due to reduced contact times can be a harder thing to infer. This analytics reduce a lot of the inferencing required by the user and directly provide actionable insights, reducing user effort and eliminating the possibility of human error due to misinterpretation of the data.
In addition to the derived direct value from data, analytics also allow systems to become aware, raise alerts for anomalous behavior, notify users in cases where the system cannot respond autonomously and isolate sections of a network in the event of a security breach. With newer applications of emerging technologies such as artificial intelligence, neural networks, and deep learning, data analytics will enable the IoT to perform functions far beyond its current scope, creating an intelligent system that works in conjunction with its human counterparts to achieve new standards of productivity, efficiency, and effectiveness. With the power of big data analytics, we are capable of finding the unknown and use it to build sustainable competitive advantage and newer business models.
Are you tapping into the power of big data analytics within your connected factory or field assets? We would love to hear your experience as to ánalytics is influencing your business decisions.