Electronic Taster
Development of a technology and device made up of a molecular sensor that allows to determine the quality and taste characteristics of products by spectral analysis, as well as data analysis and machine learning.
Motivation
Taste sensations are subjective and individual, though they are related to chemical composition of substances. That is why, until now, despite the development of analytical chemistry methods, there is no "electronic nose " or "electronic taster " at least remotely similar to the results on our senses. Within the framework of this project we are trying to make the subjective objective using modern approaches of big data analytics and artificial intelligence. These methods refer to the section of science at the junction of Chemistry, physics and Informatics called chemometrics.
As input data for analysis we use optical spectra in the near infrared range as a source of information about the chemical composition of the substance. Use of neural networks should allow to build individual cards of taste sensations, and connection of these taste cards with the analysis of spectral data will allow to create an "electronic taster ".
As input data for analysis we use optical spectra in the near infrared range as a source of information about the chemical composition of the substance. Use of neural networks should allow to build individual cards of taste sensations, and connection of these taste cards with the analysis of spectral data will allow to create an "electronic taster ".
Principle of the technology
Sample sensor
The principle of operation of the sensor is based on the presence of individual spectral features of each substance. The most vivid of these features are manifested in the middle IR spectrum area. Obtaining experimental data from this range is a technically difficult task requiring costly and cumbersome equipment. However, the use of data analysis algorithms allows the extraction of the required data from the near IR range and even from the photos of the substance. The equipment used for such a task is considerably cheaper and compact. Thus, a sensor capable of extracting the desired information from the near IR range is a perspective device. The sensor itself consists of three main elements - spectrometer, cloud storage, and data processing algorithm. The sensor allows to determine quality (freshness) and taste characteristics (level of sugar, acidity, etc.) of both solid and liquid food products. In addition, the sensor allows the analysis of products placed in transparent packaging.
The spectrum of reflection of food products is initially measured. An ordinary incandescent bulb (or LED bulbs) is used as the source of radiation. The resulting spectrum is sent to the cloud storage where it is processed using the WiFi module. A step-by-step processing algorithm is used – primary processing is carried out by the main component method, its result is fed to the input of classification trees, which allows to distinguish the peculiarity (the same taste characteristic) for each The product being analyzed.
The spectrum of reflection of food products is initially measured. An ordinary incandescent bulb (or LED bulbs) is used as the source of radiation. The resulting spectrum is sent to the cloud storage where it is processed using the WiFi module. A step-by-step processing algorithm is used – primary processing is carried out by the main component method, its result is fed to the input of classification trees, which allows to distinguish the peculiarity (the same taste characteristic) for each The product being analyzed.
How do we do it?
Areas of application
The technology can be demanded by ordinary buyers as well as producers of food products, food retail, kindergartens, schools. In addition, the sensor can also be of interest to enterprises like restaurant business as a tool to determine the taste characteristics of elite food products (such as coffee or wine).
Stage of Project
The sensor is at the prototype development stage. At the moment the written algorithms of data processing, allowing to extract the necessary information from spectra are tested. It also produces a database set for different products.
Coffee and wine are used as test objects for training.
The technology can be demanded by ordinary buyers as well as producers of food products, food retail, kindergartens, schools. In addition, the sensor can also be of interest to enterprises like restaurant business as a tool to determine the taste characteristics of elite food products (such as coffee or wine).
Stage of Project
The sensor is at the prototype development stage. At the moment the written algorithms of data processing, allowing to extract the necessary information from spectra are tested. It also produces a database set for different products.
Coffee and wine are used as test objects for training.
