Automating the Future of Manufacturing Plants
Redefining Supply Chain Management
The Industrial IoT Attack Surface
Be Change Ready in an Evolving Manufacturing Vertical
Machine Learning in Manufacturing: Moving to Network- Wide Approach
Paul Boris, CIO - Advanced Manufacturing, GE
Thyssenkrupp Elevator's Internet of Things
By Rory Smith, Director of Strategic Development Americas, Thyssenkrupp Elevator
To date, we have connected around 150,000 elevators and escalators to the Microsoft Azure Cloud using cellular modems. These units continuously report their activities, error codes, and their operational status to the cloud. We use this data to accomplish the following goals:
1. Reduce the number of breakdowns.
2. Detect when an elevator or escalator has broken down.
3. When a breakdown occurs, return that elevator or escalator to service more properly.
Each day the data in the cloud for each elevator is analyzed using Machine Learning, a form of Artificial Intelligence (AI). The probability of a breakdown in the next week is calculated. When the breakdown probability reaches a specific level, a technician is dispatched to the site. The goal here is to “fix it before it breaks”. In some cases, the probability of a breakdown has reached a significant level, but the breakdown is not imminent. In this case, the technician is advised to check certain things on his next scheduled service visit.
When a breakdown can’t be prevented, the breakdown is detected in the cloud, and a technician is then dispatched.
Changing the way one does business can bring big improvements in operational efficiency and customer satisfaction
Often the building manager is surprised when the technician arrives because he wasn’t aware the elevator was out of service.
Before the technician arrives at an installation with a breakdown, he receives advice on what is wrong and how to fix the problem. An expert system, a form of AI, known as “The Virtual Coach” gives the technician the four most probable causes for the breakdown. The most probable cause is correct 80% of the time. Collectively, the second, third, and forth probable causes are correct 10% of the time. It is expected that the Virtual Coach will get better, over time, due to learning algorithms.
The three goals have been met. In fact, everything is working better than expected. The big winner in this story is the customer. These three goals improve customer satisfaction.
What’s next? What’s next is more difficult to execute but brings huge benefits to both the customer and the company. The two things that are most significant are as follows:
1. Product improvement
2. Changes in operations.
The monitoring of so many units lets us see what components work better than others. If many breakdowns are associated with a component, the component can be improved. If a component is substituted for a different component, one can quickly determine if the new component is equal to or better than the part it replaced.
Changing the way one does business can bring big improvements in operational efficiency and customer satisfaction. However, these changes are not easily accomplished.
When we perform maintenance and what we do when we are on site is based on experience. While experience is good, statistical data, that we now have thanks to IoT, is better. We are finding that some of the service tasks we have traditionally performed have no real value. We are also identifying tasks that we did not perform in the past, that we need to perform.
Changing the way one does business is never easy. The total change in operations that one must execute to take full advantage of IoT will be painful. Not embracing the benefits of IoT can put your company at risk. However, these changes are a win-win situation—your customers have a higher satisfaction level, and you have a more efficient operation.