R.V. Isakov –  Ph.D. (Eng.), Associate Professor, Department of Biomedical and Electronic Means 

and Technologies, Vladimir State University n.a. A.G. and N.G. Stoletov E-mail: Isakov-RV@mail.ru

I.I. Alexeeva –  Undergraduate, Department of Biomedical and Electronic Means and Technologies,  Vladimir State University n.a. A.G. and N.G. Stoletov E-mail: aly33xa@mail.ru

The human factor is one of the key indicators affecting the efficiency and quality of work of operators of various activities. Therefore, neglect of official duties and violation of labor discipline can not affect the level of safety. The use of current control changes in the functional state of a person can significantly improve the efficiency of its activities.

The brain is the key organ responsible for the functioning of the human body. In modern conditions, it is very important to create an inexpensive non-invasive and easy-to-install technology that would allow monitoring the functional state of the brain of the operator. It is known that any human organ (including the brain) is in constant micromovement. Any slightest disturbance of normal mobility may affect its performance.

Currently, some scientists are trying to improve the diagnosis of the central nervous system. However, they do not reveal all the informational possibilities of this signal, and in particular, they do not provide an assessment of the functional state of the brain in real time. The brain's vibration signal, a hardware-software complex of vibroacoustic studies was developed.

Considering that information about the functional state of the brain in this signal is in the frequency domain, the time-frequency transform (window Fourier transform) was used. The decision on the presence of a functional state according to the results of neural network analysis. As part of this work, an experimental study was conducted to assess the ability to distinguish between two functional states: sleepy and anxious. 

The spectrum of the vibration signal of the brain has a fairly stable structure of harmonics, but it responds to changes in the functional state of the brain in two ways: the signal level at a certain frequency, i.e. amplitude modulation and frequency variation in a certain range, i.e. frequency modulation. For different functional states, different patterns can be distinguished in the spectrogram of the signal of brain vibrations. Also, the typical shape of the spectrum may vary among different subjects. Therefore, the use of machine learning is required to calibrate the system for a specific operator.

As the basis for such a system, perceptron-type artificial neural network technology (multilayer perceptron) was used.

The choice of the number of neurons in the hidden layer was made on the basis of the results of a computational experiment. For both states, the best were neural networks with 50 neurons of the hidden layer.

Both neuromodules show maximum sensitivity and slightly reduced specificity. To enhance the specificity, alarm is generated when several successive receipt of signals from neuromodule. 

The test results showed that the neural network successfully separates the functional states.

The results of this research and development show the possibility of creating a new system for monitoring the functional state of the brain, allowing differentiation of the operator's extreme states, such as sleepy and anxious. This additional information will help reduce human error.

1. Goryuhin V.V., Dubinin N.M. Sistema kontrolya operatora v slozhnyh tekhnologicheskih processah // Vestnik UGATU. Upravlenie, VTiI. 2007. T. 9.  № 5(23). S.16 – 21.
2. Aktual'nye problemy psihologicheskoj bezopasnosti. Sb. trudov regional'nogo soveshchaniya, g. Sankt-Peterburg, 2012 g. SPb.: Svoe Izdatel'stvo. 2012. 110 s.
3. Sistema kontrolya ustalosti operatora DSS kompanii Seeing Machines https://www.vost-tech.ru/technology/vysokotehnologichnye_ produkty/kontrol_ustalosti_operatora/ Data obrashcheniya: 11.11.2018.
4. Patent № 2417053 (RU). Sposob kontrolya i upravleniya funkcional'nym sostoyaniem operatora i ustrojstvo ego osushchestvleniya / E.N. Stradnikov,  A.S. Sliva, S.S. Sliva, N.E. Stradnikova, V.Eh. Kosteckij 17.04.2009. RF.
5. Jian-Hua Zhang, Xiao-Di Peng, Hua Liu, Jörg Raisch, Ru-Bin Wang. Classifying human operator functional state based on elec-trophysiological and performance measures and fuzzy clustering method // Cognitive Neurodynamics, 2013. V. 7, Is. 6. P. 477–494.
6. Efimov A.P. Novyj biomekhanicheskij metod neinvazivnoj ocenki vnutricherepnogo davleniya: verifikaciya sposoba i cifrovogo apparata // Rossijskij zhurnal biomekhaniki. 2011. № 4. S. 47–59.
7. Patent № 2621580 (RF). Sposob neinvazivnogo opredeleniya vnutricherepnogo davleniya / A.V. Gribkov, A.S. Kanyshev, A.A. Kirpichev, A.N. Cyplenkov, O.V. SHukov.
8. Titov V.A. Psihofiziologiya: konspekt lekcij. M.: Prior. 2003. 176 s. 
9. Fast Artificial Neural Network Library http://leenissen.dk/fann/wp/ Data obrashcheniya: 11.11.2018
10. Al'-Hulejdi N.A., Isakov R.V., Sushkova L.T. Issledovanie metodov kodirovaniya vyhodov iskusstvennyh nejronnyh se-tej pri klassifikacii variabel'nosti serdechnogo ritma // Nejrokomp'yutery: razrabotka, primenenie. 2013. № 6. S. 48–54.