Published on: 2026-07-15
Source: Peter the Great St. Petersburg Polytechnic University –
An important disclaimer is at the bottom of this article.
Specialists at Peter the Great St. Petersburg Polytechnic University are developing a hardware and software complex for early fire detection, which allows identifying hazardous gases and the beginning of a fire at the smoldering stage.
The project is carried out with the support of the Scientific and Technical Development Fund of SPbPU within the framework of the University’s Development Program, implemented with the assistance of the “Priority-2030” program.
There are currently two main approaches to studying the composition of the gas phase: spectrometry/chromatography and the use of sensors. The first is quite accurate but expensive and requires special sample preparation, complex equipment, and highly qualified personnel. The second typically responds only to a specific type of gas, but can also react to others, meaning it has cross-sensitivity. Moreover, existing gas analysis systems require prolonged accumulation of experimental data and are incapable of rapid adaptation to new tasks.
The hardware-software complex (HSC), developed by specialists at SPbPU, allows for detecting and identifying the type of gas at the moment of its slight emission — before the onset of open combustion, at the smoldering stage. The polytechnic’s development is an intelligent system that combines a dual-membrane MEMS sensor with a nanostructured oxide layer, a predictive sensor response model, a synthetic data generator, and a transformer neural network. The dual-membrane architecture provides independent thermal profiling (the ability to separately program and control the temperature of two sensitive elements on one microchip in real time); the response time is 1–3 seconds, and power consumption is less than 50 mW.
The system also has an architecture open for further training and localization exceeding 90%, so the customization cycle of the PAC will be only 1–3 days, unlike existing foreign counterparts where it reaches 6–12 months.
In fact, the project by the polytechnic students is creating a domestic platform for intelligent gas sensing, reducing import dependency of critically important safety systems. The software-hardware complex does not compete with mass smoke detectors but creates a new segment — predictive gas analytics for facilities where downtime or damage costs are measured in tens of millions of rubles, such as energy industry enterprises, railway and IT infrastructure nodes, and marine vessels.
Our project is a transition from the data accumulation paradigm to the data design paradigm. We didn’t just create a sensor; we developed an entire ecosystem: a physical model of processes within the sensor that generates training data for AI, a verified generator of training data based on real signals, and electronic hardware components. This allows our system to learn from past experience and recognize new threats without complete retraining. As a result, the work is expected to achieve model transferability from real data to mixed data with degradation of accuracy no more than 5%, which corresponds to the global standard. Then Russia will have its own platform for intelligent gas sensing, which not only matches but in many aspects surpasses foreign analogs,” comments the significance of the development Anastasia Kondratyeva, project leader and senior researcher at the Research Laboratory for Nano- and Microsystems Technology.
The zero-shot detection for previously unseen analytes is being implemented in the SPC Polytech’s PAC — an approach that allows the system to detect substances it has never encountered during training or calibration.
According to the developers’ plans, the innovative hardware and software complex for early fire detection will be ready in 2027.
The further plans of the polytechnic researchers include scaling the PAC to related areas: environmental monitoring (emission control), olfactometric control (measuring odor by its impact on humans) in the food and perfume industries, as well as medical diagnostics (non-invasive analysis of exhaled breath).
In addition, the project’s materials will be integrated into SPbPU’s educational programs in the fields of nanotechnology, AI, and digital materials science.
Please note; This information is raw content obtained directly from the information source. It is an accurate report of what the source claims and does not necessarily reflect the position of MIL-OSI or its clients.