Manufacturing to Services

Date:   Monday , February 13, 2017

Headquartered in Bengaluru, Neewee is an analytic services company. The company provides Platform as a Service (PaaS) across domains with specialization in predictive maintenance/condition based monitoring solutions for IoT.

We have frequently seen that technology has changed the way we do business. On the consumer side, we\'ve seen travel and e-Commerce change as the Internet of People grew. Similarly, the Internet of Things (IOT) is one of those enablers which will alter the way we do business. Software as a Service (SaaS), Infrastructure as a Service (IaaS), Platform as a Service (PaaS) has revolutionized IT, and similar business models will revolutionize operations and Operations Technology (OT). What is new now is the application of a SaaS-like model in manufacturing.

Manufacturing has traditionally been about delivering a product or related parts. Customers don\'t want to own the actual products; instead they just want the end result that the product delivers. For a manufacturer, this implies, shifting their view point from the product to the service outcome. This changes the relationship from a sales transaction to a mutually beneficial long-term partnership.

Before the dawn of Internet of Things (IoT), it was difficult to implement Analytics based Service solutions. It was still possible but took a lot of investment in hardware, software and communications technologies. Technology innovations are now making smart and connected assets more prevalent. Cloud computing solutions like Amazon AWS, Microsoft Azure have reduced the cost of processing power and data storage. Advancement in telecommunication with 4G/5G and introduction of LoRa are helping IoT implementation to address challenges in communication. Energy efficient sensors with cheaper price are driving the IoT implementations. All of these together are enabling usage and consumption based subscription models in manufacturing.

IoT is all about coming together of sensor data and advanced analytics to reduce operational risk and increase the value of actionable decisions. Sensors or \'things\' generate a ton data. If you look at all sensor data, it looks very similar. But, each set of data can tell its own story. At 50000 feet, the sensor data is just a time series data. What can we make out of it other than some kind of trend? Is that enough to Monetize? Probably not. Unless we dive deep and derive a meaning out of it, it is just a trend chart. This is where the analytics of things plays a key role. IoT Analytics covers cleansing or restructuring of raw sensor data, profiling of failures, detecting events and predicting possible failures based on laws of statistics and physics.

Data Harness which is the logical first step in IoT Analytics caters to the issues like availability of sensors and associated data, quality of available data, handling of false positives created during data capture & transmission and data security. The key step here is to clean these false positives by using the appropriate tools and methods. By looking at historical patterns, we can profile failures and events leading to those failures. Sensor data along with additional organizational data like asset history, failure and service requests, and maintenance and work orders can be used for analyzing the events.

Physics based models tied along with the problem patterns help in predicting failures before they occur hence reducing the unplanned downtime. Understanding of behavior of assets coupled with application of right machine learning models will lead to effective Industrial Analytics implementation. It is important to have a robust Industrial Analytics Platform which will enable quicker decision making. In our view, the following is what is required in a platform.

Scalability in Industrial Analytics Platform covers three dimensions. The first and obvious dimension is volume where the platform should be able to harness large volumes of different types of data. The other dimension is about capability to add on the new devices, new sensors to the platform seamlessly. The last dimension is the capability to define new business use cases, as and when they evolve.

Simplicity plays an important role in adaptability and popularity of any new concept. The Industrial Analytics Platform should mask the complexity of Analytics, Statistics and all the other jargons from the end user. The user facing interface should be Intuitive, Self-Serve, Configurable and enhancements should be done easily. A good solution will fit into the work processes already in place in the industry without demanding new people/skills. The platform must be able to perform the analysis on behalf of the team already at work; with only a little guidance and without them learning new skills.

Self-Learning, a mechanism that the data must be collected automatically in real-time from all relevant systems, by a productized Machine Learning application that does not need human help to verify, validate, cleanse and prepare data in the appropriate forms for analysis. It must be able to detect data patterns in real-time and swiftly alert about identified problems giving all the required information. This information should help understand the precise issue, the detected change in behavior, the potential outcome, when it will likely happen, and if possible, what to do about it. Using data analytics tools, manufacturing organizations can understand their assets in more erudite ways reduce process inefficiencies, optimize spares and inventory, cut down preventive maintenance cost thereby reducing asset lifecycle cost.