Implementation of a Low-Temperature Monitoring and Alarm System for the Taishan Neutrino Experiment

This paper describes the development and successful six-month operation of an EPICS-based low-temperature monitoring and alarm system for the Taishan Antineutrino Observatory, which utilizes PT100 sensors and multi-level thresholds to ensure the safe and stable operation of the detector by tracking liquid scintillator temperatures.

The Gist

Imagine a giant, ultra-sensitive camera buried deep underground, designed to take pictures of ghostly particles called neutrinos. This camera, known as the TAO experiment, uses a special liquid that glows when hit by these particles. However, this liquid is very picky: it must stay extremely cold (around -50°C, which is colder than a deep freezer) to work correctly. If it gets even a little too warm, the "camera" gets blurry, and the data becomes useless.

The paper you provided describes the smart thermostat and alarm system the scientists built to keep this liquid perfectly chilled and to scream for help if it ever starts to warm up.

Here is how they did it, explained simply:

1. The "Thermometers" (The Sensors)

Instead of using regular thermometers, the team used PT100 sensors. Think of these as tiny, super-precise metal wires that change their electrical resistance slightly when the temperature changes.

• The Problem: If you connect a thermometer with just two wires, the wires themselves can get hot or cold, confusing the reading (like trying to measure a room's temperature while holding a hot cup of coffee).
• The Solution: They used a three-wire setup. Imagine a three-legged stool; it's much more stable. This design cancels out the "noise" from the wires, ensuring the temperature reading is accurate to within half a degree. They placed 20 of these sensors evenly around the detector, like placing 20 weather stations across a city to make sure every neighborhood is the same temperature.

2. The "Brain" (The Computer System)

The sensors send their data to a Yokogawa GM10 system, which acts like a high-speed mailman. It collects the temperature numbers and sends them to a central computer brain running software called EPICS.

• EPICS is like the "operating system" for big scientific machines. It takes the raw numbers and turns them into a format that humans and other computers can easily understand.
• The system updates the temperature every second, creating a live map of the detector's "body temperature."

3. The "Security Guard" (The Alarm Logic)

This is the most critical part. The system doesn't just watch; it acts like a vigilant security guard with a strict rulebook.

• The Rules: The liquid is supposed to be at -50°C.
  • Level 1 Alarm (The "Yellow Light"): If the temp goes above -49.5°C or below -51.5°C, the system says, "Hey, we're drifting a bit."
  • Level 2 Alarm (The "Red Light"): If it goes above -49.0°C or below -52.0°C, the system screams, "Emergency! Something is wrong!"
• Smart Filtering: To stop the guard from barking at every little breeze, the system has a "cooling period." If the temperature wobbles near the limit, it won't send a new alarm for 12 hours. This prevents the scientists from being spammed with the same alert over and over.

4. The "Siren" (How People Get Notified)

When a real problem happens, the system doesn't just sit there. It immediately pings the scientists:

• Instant Messages: It sends a message to WeChat (a popular messaging app in China) and an email.
• The Message: If it's just one problem, it says, "Probe #5 is too hot." If there are many problems at once, it sends a summary: "We have 10 alarms; click here to see the details."
• The Dashboard: Scientists can log into a website to see a colorful map of the detector. Green dots mean "all good," orange means "watch out," and red means "danger."

5. How Well Did It Work?

The team ran this system for six months and analyzed data from 53 days.

• Accuracy: The sensors were incredibly stable, with temperature fluctuations staying between 0.15°C and 0.25°C.
• Speed: Even when the system had to handle 20 alarms at the exact same time, it reacted in less than 52 milliseconds (faster than a human blink).
• Reliability: It processed over 1,000 alarm records without crashing or losing data. It successfully caught a specific sensor that was running slightly too hot, allowing the team to fix it before it caused a bigger issue.

The Bottom Line

This paper describes a high-tech, fail-safe guardian for a delicate scientific experiment. By combining precise sensors, a smart computer network, and a "don't panic unless it's real" alarm strategy, the team ensured their neutrino detector stayed frozen and ready to capture the secrets of the universe. It's a blueprint for how to keep sensitive scientific equipment safe, reliable, and always on the lookout for trouble.

Technical Summary

Technical Summary: Implementation of a Low-Temperature Monitoring and Alarm System for the Taishan Neutrino Experiment

Problem Statement
The Taishan Antineutrino Observatory (TAO), a near-site experiment for the Jiangmen Underground Neutrino Observatory (JUNO), utilizes a liquid scintillator detector operating at a critical temperature of −50 °C. Its primary scientific objective is to provide a precise reference reactor antineutrino energy spectrum to enhance the sensitivity of neutrino mass ordering measurements. To ensure the safe operation of the detector and the accuracy of data acquisition, real-time monitoring of the liquid scintillator's temperature is mandatory. The system must address challenges such as maintaining spatial temperature uniformity, ensuring high measurement accuracy in a cryogenic environment, and providing immediate alerts when temperatures deviate from safe operational ranges.

Methodology
The authors designed and implemented a distributed low-temperature monitoring and alarm system built upon the Experimental Physics and Industrial Control System (EPICS) framework. The system architecture consists of five layers:

1. Hardware Layer: Twenty sealed PT100 platinum resistance temperature sensors are evenly distributed on the outer surface of the TAO detector's copper shell (ten in the northern hemisphere, ten in the southern). To balance accuracy and cost, a three-wire bridge circuit design was adopted to compensate for lead resistance errors.
2. Data Acquisition Layer: A Yokogawa GM10 data acquisition system (specifically two GX90XA-10-U2 I/O modules) provides constant current excitation, converts resistance to voltage, and digitizes the signal via high-precision A/D converters. The modules transmit data via an internal backplane bus to a master station.
3. Data Conversion Layer: The GM10 master station publishes process data using the Modbus/TCP protocol. An EPICS Input/Output Controller (IOC) utilizes a Modbus/TCP client driver to read these registers and map them to EPICS Process Variables (PVs) with engineering units (°C).
4. Middleware Layer: An alarm program monitors PV updates via EPICS Channel Access (CA) callback mechanisms. It employs a trigger-based logic with multi-level thresholds. Data is archived using PyMySQL.
5. User Layer: A web-based interface allows users to view real-time data, trend curves, and spatial temperature distribution maps. Notifications are pushed via WeChat Work (Enterprise WeChat) and email.

The alarm logic utilizes a "state-transition trigger" mechanism where alarms are generated only when the status code jumps to a higher level. To prevent alarm flooding, a "cooling mechanism" suppresses repeated alarms from the same PV if the temperature fluctuates near the threshold within a 12-hour window. Alarm levels are defined as:

• General Alarm: Deviation of ±0.5 °C from the setpoint (−50 °C).
• Severe Alarm: Deviation of ±1.0 °C from the setpoint.

Key Contributions

• System Integration: Successfully integrated industrial hardware (Yokogawa GM10) with the EPICS control framework, overcoming previous limitations where data was only accessible via proprietary software.
• Algorithm Design: Implemented a robust alarm handling strategy that includes state-transition triggering, cooling mechanisms to suppress false positives, and intelligent message aggregation (summarizing multiple alarms vs. pushing single detailed alerts).
• Visualization: Developed a web-based interface featuring a real-time spatial temperature distribution map of the detector, allowing operators to monitor thermal uniformity visually.
• Scalability: The system supports 20 channels with a temperature measurement accuracy better than ±0.5 °C.

Results
The system has been operating stably at the TAO site for six months. Statistical analysis based on 53 consecutive days of data (January 1 to February 22, 2026) from 18 effective probes yielded the following:

• Accuracy and Stability: The mean temperatures of 14 probes were concentrated between −50.5 °C and −49.0 °C. The standard deviation of temperature fluctuations for the probes ranged between 0.15 °C and 0.25 °C, indicating high measurement stability.
• Alarm Performance: During the statistical period, the system processed over 1,000 alarm records. The 18 effective probes triggered a total of 191 level-2 alarms. The system successfully suppressed redundant alarms; for instance, one probe (N20-37) which was persistently slightly overheated triggered 66 alarms, but the cooling mechanism prevented continuous flooding.
• Real-time Performance: Response time tests under load (1 to 20 concurrent PV alarms) showed that the alarm push delay remained below 52 ms, even at maximum concurrency, demonstrating excellent real-time capability.

Significance
The paper claims that the system provides critical support for the thermal stability and performance of the TAO experiment. It has proven effective in ensuring safe and stable operation by enabling timely operator intervention through instant alerts. The authors position this work as a cost-effective and replicable reference model for constructing monitoring systems in small- to medium-scale physics experimental setups. Future work is identified as focusing on introducing machine learning methods for predictive alarms based on temperature trends to further enhance intelligent operation and maintenance.