FEBRUARY 20238MANUFACTURING TECHNOLOGY INSIGHTSBy Shatil Haque, SVP, Global Operations, Worldwide Quality, LumiledsIN MY OPINIONHOW BIG DATA AND AI SHINE A LIGHT ON LED QUALITY IMPROVEMENTThe LED industry has grown its production capacity at a remarkable speed since LEDs were first used in commercial illumination products barely 20 years ago. In that time, virtually all the lighting industry has shifted from relying on filament bulbs and other incandescent light sources to LEDs, a type of semiconductor device. Today, across the commercial, residential, automotive, and specialty lighting segments, many billions of LEDs are assembled into new lighting fixtures every year. Lumileds, based in Silicon Valley, pioneered the design and production of LEDs for lighting and today is one of the world's top five manufacturers. Maintaining its leadership position requires the implementation of advanced manufacturing technologies and methods that enable it to continually improve product performance and quality while reducing the cost of operations. Big Data reveals patterns across billions of production unitsLumileds counts its annual production volumes in the billions of units. The quality across this production output is stratospherically high. The brand value of Lumileds depends on maintaining its leadership position in quality: customer shipments have a typical defect rate of fewer than 20 parts per billion. An important innovation in production operations that helps the company to continually improve this defect rate is the application of artificial intelligence (AI) to the vast data sets generated by testing and inspection of raw epitaxy, finished wafers, and packaged LEDs. `Maverick detection' is the process for analyzing Lumileds' `Big Data' ­ establishing the recurring patterns which characterize good units, and identifying the outliers, or maverick units, which do not match the good pattern. Shatil Haque
< Page 7 | Page 9 >