Select Page

Industrial plants cannot move into the Smart Factory era without first tapping into Big Data. In a study released by PwC, only 4% of businesses surveyed indicated that they extract the full value of the information they possess. Despite the hype by some industry analysts, there is a disconnect between the promise of Big Data and reality faced by facility owners.

Professor Gary King from Harvard University famously stated that Big Data is not about the Data. In fact, Big Data itself is not the end goal. Big Data is the means to operationalize a strategy that was previously unattainable: applying live insights from human and machine behavior to the manufacturing process.
As factories put in place infrastructure that will support the Smart Factory, these are three pitfalls to avoid.

Over-Reliance on Internal Resources
When was the last time you received a CV from an unemployed data scientist? When HBR cheekily calls data scientists “the sexiest job of the 21st century,” it is obvious that factories will find it hard to compete with the financial, cyber, consumer or digital health industries to recruit sufficient numbers of qualified data scientists to support Industry 4.0.

Large organizations may afford data science Centers of Excellence to support various parts of their business, but it is a stretch for factories to expect to build this competency in-house. Instead, factory owners need to recognize their limitations and to find cognitive technology solutions.

The alternative to a data scientist is the so-called “citizen” data scientist. The citizen data scientist does not have formal education in big data/statistics but can use models based on predictive analytics. Think paramedical versus trained doctors.

From a tool perspective, visualization and analytics capabilities can replace the need for data scientists to analyze and interpret data. Only through user-friendly technologies will Smart Factories to turn front-end employees into citizen data scientists.

Confusing “Big” Data with “Important” Data
There was a time that factories used human-generated business rules to decide which data to analyze. In the field of predictive maintenance (PdM), because of bandwidth and computing constraints, only high priority sensors data is typically monitored using SCADA tools.

The concept of limiting which sensors to monitor is now antiquated. Applying Artificial Intelligence, factories can identify evolving asset degradation and machine failure by detecting abnormal sensor data behavior. The most advanced algorithms are agnostic to asset class or sensor type. Even sensors that were seemingly “unimportant” and were overlooked may provide advanced warnings for asset downtime.

New tools serve to democratize data. With cloud-based solutions that replace traditional SCADA systems, the reliability or maintenance engineer does not need to prioritize which sensors to monitor.

Misunderstanding the Role of Data Analytics
To realize the untapped value of Big Data, factories will need to make changes in their production environment in real time. Big Data can be analyzed in real time and significantly affect operations

From a management perspective, analytics will continue to play an important role in strategic planning and decision making. However, some of the human analysis which filters the raw data will be replaced by real-time Business Intelligence tools that provide a global view. In the past, management did not have real-time visibility into the performance of machine assets or their influence on operational efficiency. In the Smart Factory, emerging threats to a factory production line will be immediately visible to multiple stakeholders in an organization and not just at an operational level.

By operationalizing Analytics, manufacturers will be able to adjust activity at a machine, facility or even organizational level.

Summary and Conclusion
Big Data represents a fundamental shift in how facilities will be able to operationalize insights from their production environments. As the Big Data discipline emerges, industrial plants will need to re-think how to build competencies from a people, technology and process perspective that will support their migration to the Smart Factory.

Big Data
Machine Learning Based Monitoring
Smart Factory

Like this article? Please share it using these icons....Share on FacebookShare on Google+Tweet about this on TwitterShare on LinkedIn
Nir Dromi

Nir Dromi

Data scientist and Machine Learning expert with 10 years of experience in startups and the Weizmann Institute. He holds a Bsc. and an Msc. in Computer Science.