Data fusion at the source: Standards and technologies for seamless sensor integration — Alan Weber

Process engineers are continually looking for ways to understand equipment behavior and improve process performance as feature sizes shrink and the related process windows narrow. In additional to collecting more and more data from the equipment directly, leading-edge factories often use custom, external sensors to improve process visibility and control. Some even extract information from internal equipment log files to look deeply into the mechanisms that may affect process performance and/or equipment productivity. As these efforts yield results, the number and variety of data sources required to support new analysis algorithms and applications will continue to grow with each manufacturing process node (see Figure 1). The problem with this explosion of data sources is that most of them are connected to the factory systems with custom, in-house software solutions, and the collected data is usually stored in whatever form/location is most convenient for the initial consuming application. Only rarely will any of this data be stored in a comprehensive, engineering database suitable for general use across the factory. This results in a “trail mix” style of system architecture that becomes increasingly difficult to understand and support over time. Clearly, a better approach is required.