Recently, there has been a double success in the field of sensors! Smart Matchstick stood out in the 9th China (International) Sensor Innovation and Entrepreneurship Competition national preliminaries organized by the China Instrument and Control Society, successfully breaking through from over 300 participating units and 819 projects (with a selection rate of about 11.5%), and the “Dynamic Image Particle Sensor” project has entered the national finals. The attachments are as follows:
Subsequently, at the academic annual meeting of the Guangdong Provincial Measurement and Control Technology and Equipment Application Promotion Association, the project “Research and Application of Dynamic Fluid Particle Image Sensing System” completed in collaboration with Harbin Engineering University won the first prize in the technology innovation category.
The two honors reflect Smart Matchstick’s continuous efforts in the field of intelligent perception and industrial equipment status monitoring, and are authoritative recognition of its innovative breakthroughs.
01
Confronting Industry Pain Points, Overcoming Multiple Technical Bottlenecks
In industries such as manufacturing, healthcare, and renewable energy, particle monitoring is a key aspect of ensuring equipment safety and product quality. Traditional particle monitoring devices have long faced industry pain points such as blurred imaging of dark fluids, weak interference resistance in complex conditions, low levels of intelligence, and data processing delays—small particles in industrial equipment lubricating oil are difficult to capture accurately, leading to early wear of equipment without timely warnings; the accuracy of detecting particulate contamination in medical injections is insufficient, posing medication safety risks; the monitoring range of particles in new energy lithium battery slurries is limited, affecting battery performance optimization.
Smart Matchstick Technology, in collaboration with Associate Professor Zhao Bin’s team from Harbin Engineering University, has spent years addressing these pain points to achieve comprehensive technical breakthroughs, establishing a complete solution from mechanism research to engineering application, which has been evaluated by industry experts as “an innovative achievement with real engineering value.”
At the meeting, Smart Matchstick’s General Manager Zou Jun, as the project leader, shared key technical breakthroughs in optical imaging chain design, light compensation for dark fluids, multi-dimensional feature recognition of particles, and embedded high-speed processing architecture.
02
Six Core Technical Highlights Overcoming Bottlenecks
1. Adaptive Light Field Control Technology for More Accurate Imaging of Complex Media
To address the challenges of poor light transmission in complex media such as dark oils and high-viscosity fluids, a dynamic light field control mechanism was constructed based on the phase of the light source, polarization characteristics, and the scattering laws of the medium. By real-time monitoring the light transmission rate of the fluid and the intensity of scattered light from particles, the angle, power, and polarization state of the laser light source are automatically adjusted, combined with the optical parameter optimization of dynamic high-precision lenses, to achieve real-time optical compensation in different media environments, ensuring that the morphological features of particles ranging from 1μm to 800μm are clearly distinguishable, with a signal-to-noise ratio improvement of over 40%.
2. First Embedded Edge Real-Time Detection Architecture
The project adopts a lightweight real-time computing architecture based on Linux, achieving:
The first lightweight edge real-time computing platform based on Linux, using a “MCU + AI chip + FPGA” collaborative processing mode, integrating the entire link process of image acquisition, filtering and denoising, and particle recognition into a compact body (weight ≤ 500g). It does not rely on external servers, achieving hardware-level flow processing, capable of processing thousands of particle data per second, outputting dynamic images in 2-5 seconds, and generating complete detection reports in 30-50 seconds, with data transmission delays reduced by 60%, perfectly adapting to the long-term on-site monitoring needs of industrial environments.
3. Breakthrough in Imaging of Dark Fluids, Overcoming Industry Bottlenecks
A mathematical model for light transmission attenuation in dark media was constructed, and an innovative brightness adaptive compensation algorithm was developed. Through AI-driven automatic exposure adjustment and linear light intensity compensation technology, the edge contours and surface texture signals of particles in dark lubricating oil are accurately restored, solving the pain points of traditional devices in imaging blurriness and detail loss in dark fluids. In practical tests, the particle detection accuracy remains above 98% in dark fluids with a viscosity of 1500cp, which has been evaluated by experts as “a milestone engineering breakthrough.”
4. Proposal of an Intelligent Recognition Model Based on Morphology and Texture Fusion
Based on a database of particle wear and contamination types, a multi-modal large model for image-morphology representation was constructed, and an intelligent recognition algorithm that fuses morphology and texture was innovatively proposed. The recognition accuracy for six major types of particles, including fatigue wear, sliding wear, cutting wear, fibers, impurities, and bubbles, reaches 95.6%, improving by over 30% compared to traditional algorithms, allowing for precise tracing of contamination sources and equipment wear status.
5. New Laser Light Sheet Illumination Enhances Imaging Depth of Field and Uniformity
To address the issues of insufficient imaging depth of field, uneven light spots, and significant scattering interference in traditional illumination in strongly absorbing fluids, the project developed:
● A thinner, more uniform light sheet illumination scheme without obvious diffraction patterns;
● A new laser structure with a wide wavelength range, stable output, and continuously tunable wavelength;
Through customized optimization of optical components, precise control over light field energy distribution, illumination consistency, and refraction loss was achieved, resulting in significant improvements in field width, transmission efficiency, and planar uniformity.
6. Full-Scene Compatible Design to Adapt to Multi-Industry Application Needs
Using a multifunctional modular design, supporting both bypass installation and online integration, compatible with specialized needs in hydraulic, lubrication, and pharmaceutical applications. Equipped with HDMI visualization output and RS485 industrial bus dual interfaces, supporting various international standards such as ISO 4406 and NAS 1638, it can interact with external devices such as filtration systems and dehydration modules, achieving a transformation from “laboratory analysis” to “on-site decision-making.”
03
Collaboration Between Industry and Academia to Bridge the Entire Link from Mechanism to Engineering Application
This project is led by Smart Matchstick Technology, in collaboration with Associate Professor Zhao Bin’s team from Harbin Engineering University. Both parties have jointly built a particle analysis and tribology joint experimental platform, developed a particle database and recognition model, and continuously iterated technical achievements through mutual verification of theoretical modeling and real industrial data, promoting the application of research results in wind power, coal machinery, and gas turbine industries, achieving a complete connection from mechanism research to engineering application.
▲ Research on the Accuracy of Dynamic Image Technology Data
Through this project, we have independently mastered optical, algorithmic, and embedded capabilities, filling the gap in high-speed imaging and online analysis, promoting the application of research results in wind power, coal machinery, and gas turbine industries, achieving a complete connection from mechanism research to engineering application. This forms a complete R&D closed loop from wear generation mechanisms, data collection, image recognition to on-site diagnosis, providing a scalable deployment technical path for intelligent state monitoring throughout the equipment’s life cycle.
The successful entry into the finals of the sensor innovation and entrepreneurship competition and the provincial science and technology first prize is authoritative recognition of Smart Matchstick’s continuous efforts and innovative breakthroughs in the field of intelligent perception and industrial equipment status monitoring technology. It also marks that the company has reached an industry-leading level in key technologies such as online identification of dynamic fluid particles and intelligent oil monitoring, providing important technical support for the health management and intelligent operation and maintenance of equipment throughout its life cycle.
Previous Highlights
Coal Washing Plant Case | Belt Reducer Online Oil Monitoring Device, Building a Lubrication Operation and Maintenance Closed Loop
Paper Mill Boot Press Station Operation and Maintenance: Online Monitoring of Three Key Indicators of Oil Wear Particles
Exhibition Review | Smart Matchstick Appeared at the International Coal Mining Exhibition, Empowering Efficient Operation of Mines with Oil Monitoring Technology
Taking the Steel Plant Blast Furnace Hydraulic Station as an example: Early Warning of Oil Contamination, More Scientific Production and Operation Maintenance
The 2025 International Coal Mining Exhibition is coming, lock in booth E4317 of Smart Matchstick, unlocking safe lubrication solutions for mining equipment
Smart Matchstick Technology—committed to empowering the digitalization of manufacturing, providing smarter, safer, and more sustainable lubrication monitoring solutions for the world, and working together to achieve industry benchmarks!
