Data is the oil of the 21st Century, and information from high quality data is powerful. In manufacturing, high-tech sensors and connected equipment form an industrial Internet of Things (IoT), which generates avalanches of data 24/7, often 365 days of the year. That’s what’s known as data exhaust.
The Importance of Capturing Digital Data Streams for Analytics, Smart Manufacturing and Data Exhaust
With recent developments in data analytics, machine learning and artificial intelligence (AI) have become an integral part of diverse industries. Constant streams of data are used to train algorithms that can recommend or even implement adjustments to business operations on the fly.
From energy, including oil and gas, and other utilities, to transportation and healthcare, and from entertainment and social media to weather forecasting and financial services (fintech), jobs have been shifting towards smart digital systems. Why should manufacturing not keep up?
In fact, the manufacturing industry of 2021/2022 demands tech savvy workers and smart, connected machinery and data flows. Many small towns and communities in the American heartland depend upon their local manufacturing plants for job opportunities and commerce. And making it in America is cool!
Unfortunately, gigabytes of data are wasted every moment throughout most of our manufacturing industry. This is data that contains information on various manufacturing processes, lead times, capacity parameters, machine statuses, etc. This is data exhaust, but too often it languishes in siloed, isolated, or fragmented datasets.
Yet more systematic acquisition or capture of such data, and aggregation of it, could enable predictive analytics across the manufacturing sector, if suitably analyzed and modeled. From our survey of manufacturing SMBs, we observed that most manufacturing-related data is not being collected. So we are greatly limited in our ability to use any statistical and machine learning algorithms to make manufacturing and data exhaust analytics more efficient.
Experienced machine operators can judge the machinability or feasibility of making a part on any given machine based on their own skills and experience developed over the years. This task is comparable to price prediction for a real estate property based on various parameters. The machine learning model identifies the pattern of predictor variables and their relationship with the predicting variable’s value.
Sustainment provides the insights needed so machine learning can be applied in a similar fashion to manufacturing capacity parameters, part design features and geometrical information. Each machine has parameters that describe its capacity for machining, including power and space requirements, and operational efficiency.
By developing a better understanding of industrial data flows in their factories and supply chains, manufacturers can leverage increasing opportunities to harness their own data exhaust as a valuable business asset that can improve productivity.
