• Internet of Things
  • Harpreet Singh
  • APR 06, 2017

IoT and Prognostic Analytics for Predictive Maintenance

Need training for Internet of Things? Browse courses developed by industry thought leaders and Experfy in Harvard Innovation Lab.

Given the high costs of industrial assets, extending their life cycle by harnessing the power of prognostic analytics for predictive maintenance is essential to maximizing your return on this substantial investment. Leverage your industrial data to lower maintenance costs, increase safety, raise productivity, and improve profits.

Challenges and Opportunities

Perhaps no aspect of the Big Data revolution encapsulates its scope better than the rise of the Internet of Things (IoT). Tiny sensors, dramatic increases in storage capacity and processing power, real-time analytics of unprecedented sophistication, and the ability to immediately translate that data into meaningful action—they are all reflected in the emergence of the 50 billion mini-machines that we call the Internet of Things. The almost unlimited data that we are able to gather from IoT devices, particularly the digital control systems that manage them, can be used to predictively maintain system operations so as to optimize asset performance. Specifically, Experfy has been working on a number of use-cases to reduce the strain on industrial assets, extend their lifecycle, improve their productivity and generate enormous cost savings so as to optimize their performance in real-time.

Those cost savings are non-trivial as the use of predictive maintenance to improve the efficiency of control systems by just 1% has been estimated to generate $2 to $3 billion in savings annually for airlines; $4 to $5 billion annually for utilities; $5 to $7 billion annually for oil and gas companies; $4 to $5 billion annually in health care; and $1 to $2 billion in the transportation sector.

The challenge has always been in how to mine the data and analyze it for effective deployment. Experfy’s data scientists and proprietary technology have achieved a new milestone by overcoming this challenge and reaping the full benefits of harnessing industrial data to enable predictive maintenance operations on control systems. We have done this by partnering with engineers, managers, and operations staff to develop a solution that is tailored to specific needs and objectives as they relate to the requisites of competitiveness in a particular industry. The goal is almost always to maximize profits, while increasing the safety, efficiency, and productivity of your operations.


Experfy data scientists have deep expertise in artificial intelligence, machine learning, prognostic analytics and operations research—the pillars of predictive maintenance in IoT control systems. Leveraging this AI expertise, Experfy has created an advanced machine learning platform for prognostic analytics that can mine data from digital control systems and provide real-time insights for predictive maintenance. To be precise, prognostic analytics give a perspective or foresight on what is going to happen when and with which probability by assessing the extent of deviation or degradation of a system from its expected normal operating conditions. In calculating the future performance of an industrial system, prognostic analytics are superior to predictive analytics because they are more effective with the closed, constrained systems, self-contained systems that are typical in the industrial sector. Most importantly, prognostic analytics are essential to predictive maintenance because they answer the fundamental questions: When should I expect a malfunction? When should I react? You can then intervene with a pre-devised solution instead of responding to an emergency whose deleterious impact is compounded by inadequate preparation. The opportunity benefit of preventative maintenance is profound in that your personnel will no longer have to spend their time collecting and aggregating data from myriad industrial systems. The time dimension that is intrinsic to our prognostic analytics solutions makes them uniquely effective in the industrial context where time is money.

The basis of our approach is to analyze the industrial data from your control systems and to identify the bottlenecks that need to be addressed to enhance performance, increase safety, lower costs and maximize profits. Which aspects of your system are most likely to fail and when? What kind of solution should you have at hand and when should you be ready to deploy it? Which staff should be ready to engage in preventative maintenance and when? How do you iteratively refine your control system so as to minimize failure rates and prolong the life of your system? How do you quantify the benefits of preventative maintenance to make a business case to senior management?

 Data scientists can leverage pattern recognition and machine learning techniques in order to:

  • Identify instrument failures before they occur
  • Calculate remaining useful life/failures to 75% likelihood
  • Quantify mechanical issues with control system
  • Develop control strategies to optimize the performance of systems
  • Lower costs of operations while increasing safety

The utility these approaches can be measured in terms of reduced staff time spent maintaining systems; reduced expenditures on maintenance; increased productivity; reduced down time of industrial systems; and reductions in days lost by your workforce to industrial accidents. Your customers will also have greater systems. Collectively, these metrics are the underpinnings of greater profits.

Boston city bkg

Made in Boston @

The Harvard Innovation Lab


Matching Providers

Matching providers 2
comments powered by Disqus.