SPbPU: Intelligent infrared minispectrometer

The main articles are:

• MES Systems - Features and Benefits
• Machine Learning

2023: Smart Infrared Minispectrometer Announcement

At St. Petersburg Polytechnic, a team of scientists has developed a sensitive intelligent infrared minispectrometer based on the LMS spectrometer matrix combined with a robotic platform. The use of this device will allow you to move from selective control of raw materials and products to complete, which will optimize the efficiency of use of materials and eliminate scrap in production. This was announced on July 12, 2023 by representatives of the St. Petersburg Polytechnic.

Scientists SPbPU created a spectrometer on a robotic platform to exclude scrap in production

As reported, the device, based on the spectral characteristics of reflected light, receives certain information from the surface of materials and conducts their qualitative and quantitative analysis - it determines the moisture content in cereals, the fat content of dairy products, determines the quality of gasoline, diesel fuel, plastics.

Polytechnics managed to amplify the optical signal thanks to the immersion technology of forming microlenses on a board with chips, with lens sizes of only 500 microns in diameter, and a height of 320 microns. This made it possible to optimize the output power of the device by 4 times, which in turn increased the sensitivity of the device.

This device has revealed its capabilities thanks to specially developed software that allows you to work both in data accumulation or collection mode and in classification mode. The latter mode allows recognition of objects by reflection spectra using previously accumulated information.

In addition to the minispectrometer, a robotic platform has been developed that allows you to collect a large amount of spectral data in an automated mode. The model trained on this data more correctly describes the object and minimizes classification errors.

The developed complex, consisting of a semiconductor matrix of an industrial partner, an enhanced optical unit, a robotic platform and software with machine learning elements, allows you to provide ultra-low power consumption, automate the collection of spectral data and training of the device. The intelligent "filling" of the device allows you to create your own material bases and sort objects based on the accumulated data.

{{quote 'author
= commented Alexander Semencha, Project Manager, Associate Professor, Director of the Nano-Technologies and Coatings REC|The use of the developed complex allows enterprises to independently "train" the device for their technological process and expand the range of applications, which will allow it to be implemented in small enterprises where the organization and content of a standard chemical laboratory is not economically justified. The introduction of the technology will optimize the efficiency of small businesses and reduce warehouse costs. For example, for cheese making, the protein content in milk is important and private cheese factories face the problem of incoming control. Moreover, with the wrong choice, the whole batch can be scrapped, which increases the risk of this business and limits the entry into the market of cheese varieties, as for example it is developed in Belgium.}}

It is possible to use the developed complex in the laboratories of bakery, flour mill and bakery enterprises, feed mills, poultry farms, livestock complexes, oil extraction and fat and fat plants, bread inspections, standardization and metrology centers, sanitary and epidemiological supervision.

This development is based on close cooperation between SPbPU and Microsensor NT LLC, where SPbPU. The project is supported by the Priority-2030 program.