AIoT-Driven Closed-Loop Thermal Regulation for Photopolymerization: A CNN–TCN Edge-Computing Framework

Imagine you are a chef trying to bake the perfect soufflé. You know that if the oven gets too hot, the delicate mixture will collapse or burn, ruining the dish. But what if your oven had no thermometer and no way to adjust the heat based on what's happening inside? You'd be guessing, hoping for the best, and likely ending up with a disaster. This is the kind of problem scientists face when working with "photopolymerization"—a fancy word for turning liquid gels into hard solids using light. It's the magic behind 3D printing and, in this story, the hardening of gel nail polish. The process creates heat as a side effect, and if that heat isn't managed, it can damage the material or even burn the person wearing the nails. For years, the machines doing this work have been like that blind chef: they just blast light and hope the temperature stays safe. But recently, a new wave of technology called AIoT (Artificial Intelligence of Things) has arrived. Think of AIoT as giving your oven a brain and a nervous system, allowing it to "feel" the heat and think ahead, rather than just reacting after things go wrong.

This is exactly what a team of researchers from Taiwan set out to do. They built a "Nail Curing Intelligent Testing System" (NCITS) that acts like a super-smart, proactive guardian for gel nails. Instead of waiting for the gel to get too hot and then panicking, their system uses a special kind of computer brain—a mix of two AI models called a CNN and a TCN—to predict exactly how the temperature will behave in the next few seconds. It's like having a weather forecaster inside your oven who says, "Hey, the heat is going to spike in 5 seconds, so let's turn down the flame now." By doing this, the system can adjust the power of the UV light in real-time, keeping the temperature just right. The results were impressive: the new system reduced the peak temperature by up to 35% compared to old methods, and it was much better at stopping the temperature from overshooting the target. In fact, it was so good at predicting the future that it made the old-school control method (called PID) look a bit slow and clumsy, cutting down on errors and settling time significantly. The best part? They managed to shrink this smart brain down so it could run on a small, cheap computer chip right next to the nail lamp, proving that high-tech intelligence can fit into everyday beauty tools.

The Problem: The "Hot Head" of Gel Nails

When you get a gel manicure, a special lamp shines ultraviolet (UV) light on the polish to turn it from a gooey liquid into a hard, shiny solid. This process is called photopolymerization. However, there's a catch: as the chemical reaction happens, it releases heat. It's an exothermic reaction, which is just a scientific way of saying "it gives off heat."

If this heat builds up too fast, it can cause two big problems. First, the gel might not cure properly, leaving you with a sticky or weak nail. Second, and more importantly for the person getting the manicure, it can feel like a burning sensation on the finger. In the past, the machines used to cure these nails were "open-loop," meaning they just turned on the light for a set time and turned it off, with no way to check if the gel was getting too hot. It was like driving a car with your eyes closed, hoping you don't hit a wall.

The Solution: A System That Can "See" the Future

The researchers, led by Kai-Chao Yao and Sumei Chiang, decided to fix this by giving the curing lamp a brain. They built a system called NCITS (Nail Curing Intelligent Testing System). Here is how it works, broken down into simple steps:

1. The Senses: The system uses sensors (like tiny thermometers) to constantly measure the temperature of the gel as it cures.
2. The Brain (AI): This is the cool part. The system uses a hybrid AI model made of two parts: a 1D-CNN and a TCN.
   • Think of the CNN as a detective that looks at the immediate temperature changes. It asks, "Is the temperature rising fast right now?" It calculates the speed of the heat rise (the derivative).
   • Think of the TCN as a time traveler that looks at the history of the temperature. It asks, "Based on the last few seconds, where is the temperature heading?" It understands the pattern over time.
   • Together, they don't just see the current temperature; they predict what the temperature will be in a few seconds.
3. The Action: Because the system can predict the future, it doesn't wait for the gel to get too hot. If it sees a spike coming, it instantly lowers the power of the UV light or turns on a cooling fan to stop the heat before it gets out of control. It's like a goalie who jumps to catch a ball before the striker even kicks it.

The Results: Smarter, Cooler, and Faster

The team tested their new system on four different layers of gel nail polish: the base coat, the color gel, the builder gel (which is thick and creates the most heat), and the sealer coat. They compared their AI system against the old standard method (called PID control) and against running the lamp with no control at all.

Here is what they found:

• Stopping the Burn: The old, uncontrolled system let the "Builder Gel" layer heat up to a scorching 39.5 °C. The new AI system kept it down to a much safer 34.7 °C. That is a reduction of about 35% in peak temperature.
• No More Surprises: The AI system was much better at avoiding "overshoot." Overshoot is when the temperature goes way past the target before the machine realizes it needs to cool down. The new system reduced overshoot by 39.7% compared to the old PID method.
• Faster Stability: The time it took for the temperature to settle down and become stable was 12.5% faster with the AI system.
• Better Accuracy: The system's predictions were incredibly accurate, with an error rate (RMSE) of only 0.89 °C. This means the system knew the temperature almost exactly.

Why This Matters

The researchers didn't just build a fancy lab experiment; they made sure it could run on a small, affordable computer chip (an Arduino Uno) right at the edge of the device. They even shrunk the AI model down using "8-bit quantization," making it 4.2 times smaller so it could run instantly without needing a massive server in the cloud. This means that in the future, your nail lamp could be smart enough to protect your fingers automatically, without you needing to be a computer expert.

While the study focused on gel nails, the researchers note that this same "smart thermostat" idea could be used for other things that involve heating materials with light, like certain types of 3D printing or coating processes. They showed that by using AI to predict heat instead of just reacting to it, we can make manufacturing safer, more consistent, and much more comfortable.

In short, this paper proves that giving a machine the ability to "think ahead" about temperature can solve a problem that has plagued the industry for years. It turns a blind, reactive process into a proactive, intelligent one, ensuring that whether you are making a nail or a new material, the heat stays exactly where it should be.