GE Healthcare’s Beijing plant is one of the largest factories producing computed tomography (CT) systems in the world. More than 1,000 CT systems, from legacy 8 slices to latest 64 slices, are assembled and tested there every year. Recently, the plant has been looking for ways to harness the Industrial Internet to increase productivity, image quality, and manufacturing flexibility.
Using robotic system in staging process
CT Staging is the last step of production during which subsystems and accessories are collected and system-level installations and tests are performed. Each CT system’s function, performance, and safety is checked on the staging line to ensure the whole system meets System Test Specific Tolerance (TST) drawing requirements. Because the staging test includes x-ray exposure, it’s conducted in special bay rooms with lead protection. Currently the staging tests, which include scanning phantoms, take several days and the bays are occupied during the entire time. Phantoms are specially designed objects that are scanned in order to evaluate, analyze and tune the CT’s performance. Usually each CT system that is being tested requires one staging operator to be dedicated to the test for the duration. Since staging is the last step of CT manufacture, no other assembly or testing processes can run parallel with staging. So staging is the most time-consuming and resource-demanding process of CT manufacturing and is usually the bottleneck for product fulfillment.
A team from GE Global Research in Shanghai has built a specialized robotics system to help ease that bottleneck. The system picks up and places testing phantoms into the CT gantry so they can be tested. The system can hold up to seven imaging phantoms of different sizes and weights to meet staging requirements. It is comprised of three single-axis robots and a rotary table, and its accuracy is higher than 0.01 mm. Another unique feature of this system is that it is movable! It can be uninstalled, moved and reinstalled easily between different bay rooms. This helps the plant use the robot’s maximum capability and reduce investments in equipment and space.
The robotic system has the potential to perform 30% of the work in the existing staging process. The robotics system has the potential to save the business millions of dollars annually by reducing production lead time and inventory carrying cost.
Industrial internet in the factory
Bringing the Industrial Internet into a manufacturing facility opens up additional interesting possibilities. For us automation engineers, the first thing we think about is developing web-based SCADA (Supervisory Control and Data Acquisition) and Human-Machine Interface systems. The SCADA system provides control of remote equipment via coded signals that operate over communications channels. The human-machine interface, or HMI, presents processed data to a human operator, allowing the human operator to monitor and interact with the process. With the Industrial Internet, it will be possible to move the traditional central control room to the cloud — or even make it available on mobile devices.
Our team is using GE’s Predix software platform to build tools that allow us to monitor the CT production line, in particular the staging process. This is the first time we have used the Predix platform, and we’ve already seen excellent results. For example, we were able to build a user interface that meets our needs without great effort because we used the Predix platform and followed Healthcare design patterns.
We are also harnessing the power of big data and cloud computing to make the most of the test images created during the staging process. The CT factory will produce about 2 million images annually in the staging process — that’s about the same number of CT images that 50 hospitals will produce in the same amount of time. Testing operators evaluate each image, label them as pass or fail, and even get some simple diagnostic results from the images. This data is an excellent resource for machine learning. It can help the manufacturing plant self-diagnose issues with images taken during the staging process and reduce trouble-shooting time in the future.