No. 7

Abstracts. No. 7, 2022 year
 
ABSTRACTS
Leonov S.A., Mikhailova Yu.V., Sudarikov V.K. (e-mail: leonov@niiteplopribor.ru)
Computer research of a magnetic flowmeter

Computer simulation possibility of a magnetic flowmeter for liquid-metal heat carries and its calculation method are considered. Numerical research of the flowmeter sensitivity using the example of the magnetic flowmeter IRMU-3-300 design was carried out. The influence of deviations in dimensions and relative positions of the flowmeter primary converter on the signal has been explored, the temperature error of the flowmeter has been estimated.

Keywords: magnetic flowmeter, Reynolds magnetic number, magnetic induction, voltage difference between electrodes, liquid metal, flowmeter primary converter.

Kaziuchyts V.O., Borovikov S.M., Shneiderov E.N. (e-mail: vladkaz@bsuir)
Method for predicting the thermal resistance of the junction-case of high-power semiconductor devices by indirect electrical parameters

The length of the procedure for measuring the thermal resistance of the junction-case of high-power semiconductor devices and the need for expensive specialized measuring equipment limit the use of this parameter as an informative one in solving problems of individual prediction of the reliability of semiconductor devices. The thermal resistance of the junction-to-case semiconductor devices of high power is proposed to be determined by two indirect, easily controlled electrical parameters, which are statistically correlated with the thermal resistance of interest. As a model for predicting thermal resistance, a previously obtained regression equation is used, including selected indirect electrical parameters as factors (arguments). On the example of field-effect transistors of the KP744A type, it has been established that the use of electrical parameters that have a noticeable and / or moderate correlation with the thermal resistance of the junction-case on the Chaddock scale (modulus of the correlation coefficient 0.4... 0.7), allows you to obtain predictive estimates of thermal resistance with a relative error not exceeding 7... 10 percent.

Keywords: semiconductor devices, reliability, individual prediction, informative parameters, junction-case thermal resistance, regression equation.

Sidorenko M., Petrosyants V. (e-mail: sidorenko.man@dvfu.ru)
Device of targeted drug delivery by magnetic nanoparticles

An improved method for drug delivery using magnetic activation of nanoparticles inside the body has been proposed. The design of the device is proposed, which makes it possible to generate constant and variable fields inside the body. A study of the operating modes of the device was carried out. The dependences of the device parameters on the modes of its operation are obtained. Recommendations are given on the choice of effective operating modes of the device, taking into account the limitations of invasive impact on the core.

Keywords: targeted drug delivery, magnetic nanoparticles, invasive device, controlled magnetic activation.

Khatsevich T.N., Izhbuldin D.A., Grechenevskiy A.S. (e-mail: khatsevich@rambler.ru)
Designing an LWIR zoom lens

Current trends in IR lens design are outlined. The characteristics of continuous zoom LWIR lenses are analyzed. The zoom lens optical scheme with a variable focal length in the range of 200 to 40 mm and relative aperture 1 : 1.5 is designed. It is shown that it is possible to minimize the longitudinal chromatic aberration in the continuous zoom LWIR lens system with a limited number of materials used. It is shown that the optical system of a four-component zoom lens can compensate manufacturing errors of optical and mechanical components in the process of adjustment and provide an active temperature compensation in operation.

Keywords: zoom lenses, longwave infrared, LWIR, uncooled matrix receiver, image quality assessment, active temperature compensation.

Biryukov S.V. (e-mail: sbiryukov154@mail.ru)
Investigation of a double spherical electric field strength sensor with overhead sensing elements

A new type of electric induction spherical electric field strength sensors of a closed type is being investigated. Along with the well-known types of electro induction sensors, which are single and double according to their design classification, a new type of sensors is assigned to the newly introduced classification feature – dual sensors. In the work, attention is paid to one of the possible options for constructing dual sensors – a sensor with overhead sensitive elements. The metrological characteristics of the sensor, namely, the additional error from the inhomogeneity of the electric field, were studied. The interrelation of the error from the constructive dimensions of the sensitive elements of the sensor and its spatial measurement range, in which this error does not go beyond the specified one, is established. The research results show the feasibility of building two versions of dual sensors with overhead sensitive elements. The first option makes it possible to make dual sensors with an error selected from the inequality 2,77 % < δ < 4,02 % and maintained over the entire spatial measurement range 0 < a < 1. The second option provides dual sensors with errors selected from the inequality 0,02 % < δ < 2,55 %, but in a limited upper value of the spatial measurement range of 0,33 ≤ a ≤ 0,98. The results of the conducted studies show the promise of using dual sensors with overhead sensing elements.

Keywords: electric field strength, measurement, single sensors, doubles sensors, dual sensors, overhead sensing elements, error due to field inhomogeneity.

Chertilin K.E., Ivchenko V.D. (e-mail: kchertilin@yandex.ru)
Development and transfer training of an artificial neural network for image classification

To solve machine learning problems with the allocation of a large number of features, it is necessary to use deep neural networks, which require large time costs. To solve this problem, the article suggests using pre-trained neural networks. The article provides an example of transfer training of a deep neural network for solving the problem of image classification in Python. The analysis of the obtained results is carried out, the results and advantages of the proposed method are demonstrated.

Keywords: artificial neural networks, convolutional deep networks, transfer learning, ResNet, Python, Pytorch library, image classification.

Kolmogorova S.S., Biryukov S.B., Kolmogorov A.S., Baranov D.S. (e-mail: ss.kolmogorova@mail.ru)
Computerized software and hardware complex of the system of data collection and intelligent processing

The research work presents an automated collection and primary processing of electromagnetic field parameter estimation data in a distributed measuring complex. The developed intelligent set of early warning safety system devices is based on non-contact measurement of the electromagnetic field, to ensure life safety and stable operation of equipment. Experimentally measured electric field values agree well with simulated values. The prototype digital data system with safety warning devices has no false positive or false negative results when testing the model behavior of the electromagnetic analysis software modules of the primary data processing. The software modules include a configuration degradation scenario, e.g., in the event of loss of communication or failure of one of the distributed sensors. In this case, the neuroconfiguration adapts to the current conditions, while maintaining the maximum efficiency of the complete system.

Keywords: software and hardware system, measurement process automation, measurement signal, Internet of Things, sensor signal processing, program processing, electrometric measurements.

Вack to content