Ever wonder if you could listen to a problem and solve it? That's precisely what a Delaware-based startup, Sonibel Instruments, is doing, but instead of cracking codes like in the movie "Sneakers," they're listening to welds!
Remember the scene in the 1992 film where Whistler, the blind audio expert, identifies the location of a hidden black box just by listening? Well, Sonibel is using a similar concept, but with a modern twist: artificial intelligence and acoustic sensing to detect defects in MIG (Metal Inert Gas) welds.
Imagine a world where you could instantly know if a weld is flawed, before it even leaves the welding station. Sonibel Instruments is making this a reality. The team, led by CEO Sophia Millar, CTO George Hallo, and CPO Hooman Piroux, got the initial idea from a job shop owner who could identify bad welds just by their sound.
But here's where it gets interesting... Sonibel's system doesn't rely on cameras or visual inspections. Instead, it uses a specially designed sensor, encased in aluminum, that attaches directly to the welding torch. This sensor listens to the vibrations of the molten metal droplets as they hit the weld puddle. Hallo explains that the variations in these vibrations are the key. They indicate whether the weld is solid or if there are issues like perforations or incomplete fusion.
Using a unique algorithm developed by Hallo and Piroux over nearly a year, the sensor's software analyzes the sound, interprets it, and displays the results on a small monitor. The system currently classifies a weld as either "good" or "defective" once it's complete. However, if a defect is detected, the system pinpoints the location of the problem within the weld.
"Sometimes, it's very clear that this is a terrible, porosity-filled weld," Millar says. "We can just label that right away. But other times, it's subsurface, and that's where our tool would help most."
And this is the part most people miss... The secret sauce? A comprehensive database of welds and inspection results. This database is the foundation of the system's algorithm, allowing it to accurately identify even subtle defects. Hallo emphasizes the importance of this database in capturing the full range of weld variability. The goal is to catch not only the most obvious flaws but also those hidden beneath the surface, such as subsurface porosity.
Sonibel is still in its early stages, with only a few units currently being tested in the field. The company is actively gathering data and feedback from early test projects to further refine its AI software. They even had a waiting list of potential customers eager to purchase the product before it was even fully ready! In exchange for beta-testing the system, these partners provided valuable feedback and data, which the team needed.
"Right now, a big part of what we’re doing is going back to that wait list and seeing who wants to convert, who wants to actually buy the unit," Millar said. They are currently focused on demos and pilot programs to showcase the technology.
What do you think? Could acoustic sensing revolutionize the welding industry? Do you foresee any potential challenges or limitations with this technology? Share your thoughts in the comments below!
