CC BY
6constants or equal to 0 indicate a disparity of mechanism to physic-chemical catalytic basics of reaction, but it is possible that b'c = 0.
For the chosen mechanism, process data for temperatures equal 463 K using the algorithm mentioned above.
Calculate rate constant k and adsorption coefficients bA, bB, be for both temperatures using the relevant equations - see (15a - 21a).
Using values of calculated reaction rate constants and component adsorption coefficients for specified temperatures, calculate:
- activation energy, J/mole:
R ■ T ■ T KT E = R T T ■ ln- t
a (T2 - Ti )
Kr
- pre-exponential factor, s-1
k0 - ■
K-r
K-r
-E/ - E/
'R-T „ /RT-)
e ' 1 e
- heat of adsorption for each component J/mole:
Q -
R■ T ■ T bT
R T1 T2 _J _ lTl
(T - T) biT 2
The best results (the minimum total sum of squared deviations) among the four mechanisms are obtained for mechanism 1 - equation kinetics (15). Calculated reaction rate W practically do not differ from the experimental values We. Table 1 provides calculations of activation energy and heat of adsorption of components.
Conclusions. The proposed method for calculating kinetic parameters will facilitate better understanding by students and researchers of possible mechanisms of heterogeneous-catalytic reactions and performing calculations according to the proposed algorithm will deepen their skills in the advanced Excel.
References
1. G.S. Yablonskii, V.I. Bykov, V.I. Elokhin, A.N. Gorban (1991), Kinetic models of catalytic reactions, Elsevier Science, p. 391
2. Anna Kulik, Ramona Saliger. Seminararbeit «Heterogene Katalyse an Festkörperoberflächen», Technische Universität Braunschweig Institut für Physikalische und Theoretische Chemie, Braunschweig, 2006, 27 s.
3. Devanand Pintoa, Edgar A. Arriagab, Regine M. Schoenherrc, Shirley Shinn-Huey Chouc, Norman J. Dovichic (2003), Kinetics and apparent activation energy of the reaction of the fluorogenic reagent 5-fu-roylquinoline-3-carboxaldehyde with ovalbumin, Journal of Chromatography B, 793 (1), 107-114.
4. Koncevoj A.L. Issledovanie processa gidriro-vaniya primesi dioksida ugleroda v prirodnom gaze /A.L. Koncevoj, B.A. ZHidkov, O.G. CHernickij // Ki-netika i kataliz. - Kiev: Naukova dumka, 1982. -vypusk 20. - S.78-82.
5. Langmuir-Hinshelwood Kinetics. https ://www.sciencedirect.com/topics/chemistry/lang-muir-hinshelwood-kinetics (date of the application 06.01.22)
FORMATION OF A SET OF ALTERNATIVES FOR PERSONNEL DECISION ON COMPLETING VACANT POSITIONS OF MILITARY ORGANIZATIONAL STRUCTURES
Prokopenko O.,
ORCID 0000-0002-5482-0317 Adjunct of the Center for Military and Strategic Studies of the National Defence University of Ukraine named after Ivan Chernyakhovsky
Ukraine, Kyiv Rybydajlo A. ORCID 0000-0002-6156-469X Candidate of Technical Sciences (Ph.D), Senior Researcher, Leading Researcher of the Center for Military and Strategic Studies of the National Defence University of Ukraine named after Ivan Chernyakhovsky
Ukraine, Kyiv
Abstract
The proposed method of forming a set of alternatives for personnel decision on completing vacant positions of military organizational structures on the basis of the application of the developed model of artificial neural network.
Keywords: personnel decision; career management; military organizational structure; artificial neural network, decision support system, rating, classification.
Formulation of the problem. The modern stage of reforming the Armed Forces of Ukraine is carried out in the conditions of a difficult military-political and economic situation that has developed as a result of the armed aggression of the Russian Federation. This causes requirements for guaranteed and high-quality
staffing of military organizational structures by trained and motivated personnel.
The most effective mechanism for solving this problem is the development and implementation of appropriate information and analytical support for:
personnel accounting taking into account the individual professional competencies and moral and business qualities of each individual;
carrying out constant monitoring of vacant positions and the scarcest specialties;
forecasting the need for human resources in accordance with real and potential threats in the field of national security and defense;
implementation of virent and transparent procedures for career advancement of personnel [1].
Special attention should be paid to Information Decision Support Systems in the field of Human Resource Management, which provide the necessary information support for making informed, virtuous and transparent personnel decisions.
The task of appointing a serviceman to a vacant position of a certain military organizational structure is currently being solved in the following order:
formation of a list of candidates for a vacant position based on the results of an annual assessment; rating of candidates;
development of recommendations by the selection committee based on taking into account additional characteristics of candidates - this process is not automated;
providing an updated rating list of candidates to an authorized person for making a personnel decision.
The task of automating the process of clarifying the rating of candidates for a vacant position, taking into account their additional characteristics, is considered relevant.
Analysis of recent research and publications. In recent years, the leadership of the Ministry of defense of Ukraine and the General Staff of the Armed Forces of Ukraine has paid considerable attention to solving the problems of creating a modern personnel management system based on automation of Personnel Management processes [2, 3].
At the moment, the special software of the information and analytical system (IAS) "Personnel" allows you to carry out the main processes of accounting and Personnel Management in automated mode. The architecture, software environment, and process development and configuration tools allow you to create new and improve existing career management processes.
However, when calculating the rating of candidates for promotion to higher positions only the results of the annual assessment of military personnel of the Armed Forces of Ukraine are taken into account [4]. In other words, when forming a set of possible alternatives for personnel decisions, additional characteristics of candidates (term of office, experience of combat oper-
ations / peacekeeping operations, motivation and personal preferences of military personnel to develop an individual professional career, and so on) are not taken into account.
The use of Information Technologies based on artificial intelligence has become a promising direction in the theory of decision-making. The main tasks of their application in decision support systems are solving classification problems, where artificial intelligence is a kind of tool for performing functions associated with human intelligence - logical thinking, learning and self-improvement [5].
The issue of neural network assessment of personnel competencies is investigated in [6]. Studies were conducted on the classification of personnel based on the assessment of their acquired professional competencies, due to the developed artificial neural network (ANN). A methodological approach to teaching ANN is proposed, which can be used to solve a different range of management issues.
In [7], it is proposed to provide the possibility of an objective and complete analysis of the current state of an object, through the use of improved fuzzy temporal models of the object state, an improved procedure for predicting the object state, and an improved procedure for training artificial neural networks that evolve. The ability to clarify information about the state of the monitoring object is achieved through the use of an improved training procedure. Its essence lies in the fact that synoptic weights of an artificial neural network, the type and parameters of the membership function, as well as the architecture of individual elements and the architecture of the artificial neural network as a whole are trained.
The purpose of the article is to highlight the methodology for forming rating lists of candidates for appointment to vacant positions using neural network technology, which, unlike the existing one, provides an opportunity to automatically take into account additional characteristics of candidates.
Research results. The principle of building decision support systems based on artificial intelligence is based on the use of artificial neural networks (ANN).
ANN is a mathematical model and its software and hardware reproduction, which simulates the activity of the human nervous system, built on a system of interconnected artificial neurons. Each artificial neuron is a kind of processor, which lays the simplest algorithm for converting input parameters from the neurons of the previous layer, the formation of output parameters and their transfer to the neurons of the next layer [8]. ANN consists of three types of elements (Figure 1).
Figure 1 - Scheme of an artificial neural network
The first type of elements includes input signals, each of which at the input takes the value of a certain characteristic of the object of observation
X,X,...,Xk | k = 1,K (where: K - the number
of characteristics).
The second types of elements are associative elements - neurons of the hidden layer
n11,n12,...,nji | i = 1,1, j = 1, J (where: J -
number of layers, I - number of neurons in the layer), which combine with the elements of the first type and transmit signals to the responsive elements. There may be one to several hidden layers at a time, depending on the complexity of the model and the results to be achieved. The artificial neurons in each layer are also associatively related.
The third type includes elements of the output signal of the ANN y,y,...,ym | m = 1,M (where:
M - is the number of elements of the output signal), ie the result expected from the ANN.
A system of interconnected artificial neurons in a network with a controlling influence, capable of solving problems of different nature and level of complexity. A distinctive feature of ANN from conventional algorithms for solving problems in similar processes is the need for their training.
The simplest element of ANN is the perceptron [9] (Figure 2).
The input of the perceptron receives input data
Xj,X2,...,xk . The artificial neuron receives the input
information through synapses, and the output information in the form of the result is received through the axon.
Figure 2 - Scheme of a simple perceptron
Each synapse has its own weight w, W,-., W
, which determines how much the appropriate input of the neuron affects its state. Therefore, the state of the artificial neuron is calculated as follows:
K
s=Z
xkwk + b
(1)
k=1
where: K - is the number of input values of the artificial neuron;
Xk - the value of the k-th input of the artificial neuron;
Wk - the value of the k-th synapse;
b - bias value [9].
The value of the axon of the neuron is the value of the activation function, which can be represented as:
Y = f ( S )
(2)
To model the neural network, it is necessary to determine its hyperparameters:
the number of network layers, hidden layers and neurons in each layer;
activation functions that will be used in the neurons of each layer.
These parameters are determined experimentally and in the process of learning the neural network can be changed.
The practice of constructing and using ANN does not provide unambiguous answers to how many neurons each layer should contain and how many hidden layers there should be, but their number must be greater
than the number of input and output parameters. Further, in the process of learning the neural network, the number of these components may increase or decrease, depending on the reliability of the original data.
To solve the problem of automating the procedure of classification of candidates for appointment to a vacant position, taking into account their additional characteristics, the architecture of the artificial neural network is built on two hidden layers, the parameters of which include:
1st layer - 9 neurons, which through synoptic connections receives input from 6 parameters
X = {x1...x6};
2nd layer - 3 neurons that will connect the synopses of the 1st hidden layer and the source layer, the neurons of which will output the original data Y through the axons.
The rectified linear unit (ReLU) function will be used as the activation function for the first hidden layer (figure 3):
f (S ) = max(0,S) =
'0| S < 0 S| S > 0
(3)
f(S) 1
J 0 s
Figure 3 - The rectified linear unit activation function (ReLU)
The ReLU function is a linear identity for all positive values and a zero for negative values. The advantages of using this function are faster calculations, faster and more efficient learning of ANN, better gradient propagation, and invariance with respect to scaling [8, 9].
The following personal quantitative and qualitative characteristics of servicemen are used as input data:
X - integrated assessment of the candidate, calculated using the method of determining the rating [4];
X2 - the level of motivation of servicemen, which is assessed using the methodology covered in [10];
X - term of office;
X4 - personal preferences (position in which the serviceman wishes to perform military service);
X - recommendations of the direct supervisor,
based on the results of the annual evaluation;
X6 - profile of service activity of the serviceman.
The input to the neural network can be represented as a column vector, where the numerical value of each element is a natural number, which is the corresponding estimate for each criterion:
X = ( X tu
(4)
where I = 6 - the number of input parameters of the neural network.
As initial data we will define three classes of correspondence of the serviceman to a vacant position
(Y):
y - it is advisable to appoint a candidate to a vacant position;
y2 - it is permissible to appoint a candidate for a
vacant position (it is necessary to acquire additional professional competencies, etc.);
y - it is not advisable to appoint a candidate to
a vacant position.
The source data from the neural network can be represented as a column vector:
7=( yk )K=1K
(5)
where K = 3 - the number of output parameters of the neural network.
The numerical value in one of the classes of source data should tend to one, while in other classes this value tends to zero, and the sum of the values of the output parameters of the neural network is equal to one:
iL yk =1
(6)
softmax(7 ) =
?yk
z
K k=1e
yk
(7)
Thus, on the basis of the neural network's assessment of the personal characteristics of candidates, the classification of candidates for the vacant position is carried out, which are placed in the appropriate classes of alternative solutions.
The decision-maker considers the list of candidates for each class and, based on personal professional experience and authority, selects a candidate to make a personnel decision on the appointment.
The second hidden layer will be the source, so it is not necessary to use the function of activating neurons.
The structural model of the proposed neural network is presented in Figure 4.
The algorithm of artificial neural network is built on the basis of the method of Forward propagation.
The weighted sum on one neuron can be represented as a scalar product of vectors, which are the values of the elements of the vector of input values and the vector of weights of the synoptic relationship of each neuron with the input data.
For the first layer of ANN, the input values supplied to each neuron can be represented as:
H1 = W x X + B1 = (hj )j=J, (8)
where H - vector-column of input values of the first layer of ANN;
X - input parameters to ANN; W - matrix of weights, where
)J I
'j=1
W = (w.. )J -1__;
B - bias vector, where B1 = (bj )J_yj ;
J - the number of neurons in the first layer (rows of matrix elements);
I - the number of input parameters to the ANN (columns of matrix elements).
The output values of the neurons of the first layer will be converted by the function ReLU input values:
The Softmax function is applied to the initial values, which converts an arbitrary vector into a set of probabilities:
S1 = f (H1) = (Sj ) J
(9)
where Sj - vector-column of the initial values of
the neurons of the first layer.
Similarly, the calculation procedure is performed for the inputs of neurons of the second layer of the neural network:
H2 = W2 x S1 + B2 = (hk )K=IK, (10)
where H - vector-column of input values of the first layer of ANN;
S - vector of initial values of the 1st layer of the neural network;
W - matrix of weights, where
W = (w*)K, — —■
W (w k )k=1,1, j=1, J ■
B2 - bias vector, where B2 = (bk )^-Yk ■
K - the number of neurons in the second layer (rows of elements of the matrix);
J - the number of input parameters to the neurons of the second layer of the ANN (columns of the matrix elements).
H^W^X + B^ih^-5, =/(//,) = (sj)J-
Figure 4 - Structural model of an artificial neural network
The output values of the neurons of the second layer will be equal to its input values, because the activation function in this layer is not used: = H2
The results of the ANN calculations will be the values of the probability distribution for each class of the vector of the input values of the neurons of the second layer obtained using the Softmax function (7), where the maximum value y will indicate the suitability of an individual candidate for a particular class.
Training of an artificial neural network is carried out by means of a method of back propagation. This is an iterative gradient algorithm that is used to minimize the error of the multilayer perceptron and obtain the desired yield. The main idea of this method is to propagate error signals from the outputs of the network to its inputs, in the direction opposite to the direct propagation of signals in normal operation.
A detailed description of the neural network learning procedure is beyond the scope of this article.
The method of forming alternatives to personnel decisions based on the classification of candidates for the vacant position, using the technology of neural networks, in the form of a block diagram is presented in Figure 5.
Block 1. The architecture and hyperparameters of the neural network are determined (parameters that are not subject to learning):
vector X (input data) (pos. 1.1) - personal quantitative and qualitative characteristics of servicemen, according to which it is necessary to classify servicemen;
vector Y (initial data) (pos. 1.2) - determination of the number of classes to which the ANN classifies each serviceman;
the number of hidden layers of ANN (pos. 1.3) Hx, H2,..., Hp \p e □ , where p - the number
of hidden layers, as well as the number of neurons in each layer (pos. 1.4, 1.5). Determined experimentally, based on the number of elements of the vectors X and
Y;
selection of neuronal activation functions in the hidden layers of the ANN (pos. 1.6).
Block 2. The primary weights of synoptic connection in the hidden layers of ANN (pos. 2.1, 2.3), as well as the weights of the displacement vectors of neurons of each layer (pos. 2.2, 2.4) are determined. In the future, their true value is calculated by using the teaching methods of ANN.
Block 3. The input data to the ANN are in the form of a list of servicemen selected by special software, with certain personal quantitative and qualitative characteristics.
Block 4. ANN provides for the automated formation of alternatives to personnel decisions by the method of Forward Propagation. Input data (X), transformed by weights (), fall on the neurons of
the first layer of the ANN (pos. 4.1). At the output of the first layer of the ANN, from the active neurons determined by the activation function (^) (pos. 4.2), the
data converted by weight coefficients (W2), fall on the
neurons of the second layer of the ANN (pos. 4.3). In the second layer of ANN, in the absence of the function of activation of neurons, the initial values of this layer are equal to the input. The data is converted into a probability distribution using the Softmax function, which divides them into classes, thus forming the source vector (Y) (pos. 4.4).
Block 5. The transformed ANN input data are classified into three classes, which form lists of servicemen who should be appointed to a vacant position, for whom there is a possibility of appointment and who should not be appointed. The official, or attestation commission, considers each class of the formed lists, on the basis of which the personnel decision on appointment of the
best serviceman on a vacant position is made. In addition, based on the established classes, it is possible for the decision-maker (DM) to use regulatory functions in the form of administrative, disciplinary and stabilizing decisions for the rest of the military. For example, it is advisable to send second-class servicemen to advanced training courses, and to consider third-class servicemen before being assigned to another service profile.
DEFINITION
hyper-parameters of the artificial neural network {parameters that are not subject to learning):
Y = (yk)
- amount of input data:
1.2
- amount of output data: -amount of hidden layers of ANN:
- amount of neurons in the hidden layers of the ANN:
- neuronal activation functions in the hidden layers of ANN:
1.5
= max(0,// )
UPDATE
parameters of the artificial neural network during its training (parameters to be trained):
- values of synaptic weights of the first layer of ANN
- values of the bias vector of neurons of the first layer of ANN: -values of synaptic weights of the second layer of ANN:
- values of the bias vector of neurons of the second layer of ANN
"1 ^ Jl j-\ .J ,1—1./
»'. -(»v.»;;,,,
—
METHOD
learning artificial neural network
DOWNLOAD
input data by special software (characteristics of selected candidates for the vacant position)
Xi - integral assessment of candidates; x2 - level of motivation of candidates; x3 - term of office of candidates:
x4 - candidate preferences (the position that the candidate wants to occupy); x5 - recommendations based on the results of the annual evaluation of candidates; x6 - job profiles of candidates;
CALCULATION BY ARTIFICIAL NEURAL NETWORK by the method of the Forward Propagation alternatives to the personnel decision
Hi=WlxX + B]
S^fiHj
H2 = W2xSl + B2
Y - Softmax(//2)
ACCEPTANCE
by the decision maker (DM), the most rational version of the personnel decision on the appointment of a specific candidate for a vacant position, from the set of possible alternatives Y formed by an ANN
y-i - candidates who should be appointed to a vacant position; y2 - candidates for whom there is a possibility of appointment to a vacant position; y3 - candidates who should not be appointed to a vacant position
Figure 5 - Block diagram of the methodology for forming alternatives to personnel decisions
To test the efficiency of the proposed method, in the Python 3 programming language, a model of special software of the proposed model of artificial neural network was developed. Learning of ANN was carried out on the basis of the developed training sample by means of its division into three batch. The quality of ANN training was assessed by determining the percentage of
providing ANN correct answers to the total number of input parameters of the training sample:
V
z
Q =
^ z=1 ^
correct
z
x 100%.
(11)
where Q - the quality of learning ANN; correct
yz - correct ANN answer;
Z - is the number of rows of the training sample (Dataset length).
With the speed of learning ANN ( = 0,003 and the number of epochs of learning G = 600, the quality of learning is Q = 98,9% , which indicates its high learning.
This is also traced on the error drop graph (Figure 6), which shows certain fluctuations when loading each batch. The use of batch data loading in the training of ANN - increases the efficiency and accuracy of the calculation of the original data, but the number of epochs is tripled, because each new cycle of training will load a separate batch. Therefore, the full Dataset learning
cycle A will be A X 3 a cycle in which three packets of randomly mixed tuples of learning data with the correct answers will be loaded into the ANN.
Figure 6 - Graph of error drop at ß — 0,003 and G — 600
Conclusions. The use of neural network methods to take into account the additional characteristics of candidates for appointment to typical positions will provide an opportunity to automate the procedure of forming many alternatives to staffing.
In the future it is advisable to detail the method of learning an artificial neural network for the formation of alternatives to personnel decisions. The results of this study will be covered in future issues of this scientific journal.
References
1. On the decision of the National Security and Defense Council of Ukraine of August 20, 2021 "On the Strategic Defense Bulletin of Ukraine": Decree of the President of Ukraine of September 17, 2021 № 473/2021. Official Gazette of the President of Ukraine. 2021. 5 Oct. (№ 24). P. 27.
2. Tureychuk A. M Analysis of automated systems created for automation of personnel management processes of the Armed Forces of Ukraine. Collection of scientific works of the Center for Military Strategic Studies of the of the National Defence University of Ukraine named after Ivan Chernyakhovsky. Kyiv,
2016. № 3 (55). PP. 106-110. DOI: 10.33099/23042745/2016-1-56/106-110.
3. On approval of the Concept of Personnel Policy in the Armed Forces of Ukraine for the period up to 2025 : order of the Ministry of Defense of Ukraine dated 14.09.2021 № 280. Official website of the Ministry of Defense of Ukraine. URL: https://www.mil.gov.ua/content/mou_orders/ mou_2021/280_nm.PDF
4. Methodical recommendations on the formation and use of the Reserve of candidates for promotion in the Armed Forces of Ukraine, approved by the Director of the Personnel Policy Department of the Ministry of Defense of Ukraine from 19.03.2018 № 350: electronic resources / official website of the Ministry of Defense of Ukraine. 2018. URL: https://www.mil.gov.ua/ content / other / mrk_rezerv_2019.pdf.
5. Tadeusevich R., Borovik B., Gonchazh T., Lep-per B. Elementary introduction to the technology of neural networks with examples of programs / trans. from Polish. ID Rudinsky. Moscow: Hotline - Telecom, 2011. 408 p.
6. Krichevsky ML, Dmitrieva SV, Martynova Yu. A. Neural network assessment of staff competencies.
Labor economics. Moscow: Creative Economy Publishing House LLC, 2018. Vol.5 №4. Pp. 1101-1118.
7. Romanenko I., Golovanov A., Khoma V., Shy-shatskyi A., Demchenko Ye., Shabanova-Kushnarenko L., Ivakhnenko T., Prokopenko O., Havaliukh O., Stu-pak D. Development of estimation and forecasting method in intelligent decision support systems. Eastern-European Journal of Enterprise Technologies, Information and controlling system. Kharkiv, 2021. Vol. 2 No. 4 (110). P. 38-47.
8. Kruglov V. V, Borisov V. V. Artificial neural networks. Theory and practice. 2nd ed. Moscow: Hotline-Telecom, 2002. 382 p.
9. Haykin S. Neural networks: full course, 2nd ed., Ed. / trans. with English Moscow: 000 "I. D. Williams", 2006. 1104 p.
10. Grom VA, Georgadze OA, Yakimenko IV Methodical approach to assessing the level of motivation of servicemen of the Armed Forces of Ukraine. Collection of scientific works of the Center for Military Strategic Studies of the National Defence University of Ukraine named after Ivan Chernyakhovsky. Kyiv, 2016. № 2 (57). Pp. 67-70. DOI: 10.33099 / 2304-2745 / 2016-2-57 / 67-70.
ЭКСТРАКЦИЯ ПРЕСНОЙ ВОДЫ ИЗ АТМОСФЕРНОЙ ВЛАГИ
Серебряков Р.А.
кандидат технических наук, ведущий научный сотрудник, Федеральный научный агроинженерный центр ВИМ (ФГБНУ ФНАЦ ВИМ),
Россия, Москва
FRESH WATER EXTRACTION FROM ATMOSPHERIC MOISTURE
Serebryakov R.
Candidate of Technical Sciences, Leading Researcher, Federal Scientific Agroengineering Center VIM (FGBNUFNATS VIM)
Russia, Moscow
Аннотация
Проблема дефицита пресной воды становится все актуальней для многих регионов мира - это становится одним из главных факторов, сдерживающих развитие цивилизации во многих регионах Земли. Её обострение связывают с ростом населения, климатическими изменениями и рядом других причин. Так в ХХ веке население земного шара выросло в три раза. За этот же период потребление пресной воды увеличилось в семь раз, в том числе на коммунально-питьевые нужды - в 13 раз. При таком росте потребления стало резко не хватать водных ресурсов в целом ряде регионов мира. По данным Всемирной организации здравоохранения более двух миллиардов человек в мире страдают сегодня от нехватки питьевой воды.
Предлагается современная альтернативная энергонезависимая установка с использованием технологий вихревой энергетики - «Воздушный Родник» для получения пресной воды из атмосферного воздуха.
Abstract
The problem of fresh water shortage is becoming more and more urgent for many regions of the world - this is becoming one of the main factors hindering the development of civilization in many regions of the Earth. Its aggravation is associated with population growth, climatic changes and a number of other reasons. So in the twentieth century, the world's population has tripled. During the same period, the consumption of fresh water increased sevenfold, including for communal drinking needs - 13 times. With such an increase in consumption, water resources have become severely scarce in a number of regions of the world. According to the World Health Organization, more than two billion people in the world today suffer from a lack of drinking water
Ключевые слова: вода, экстракция, атмосферная влага, конденсат, вихревая энергетика, родник, имитационное моделирование, биокаталическая активность, окислительно-восстановительный потенциал.
Keywords: water, extraction, atmospheric moisture, condensate, vortex energy, spring, simulation modeling, biocatalytic activity, redox potential.
Введение. Для многих мест на земном шаре проблема нехватки питьевой воды не является новой, так как она обусловлена климатическими особенностями, а именно небольшим количеством осадков. К засушливым областям относятся те территории, где выпадает менее 400 мм осадков в год. При таких значениях невозможно ведение сельского хозяйства без дополнительных источников воды. Наиболее засушливые области, где выпадает
менее 100 мм осадков в год, составляют 34% земной поверхности (без учета Антарктиды). На аридные области (100-200 мм осадков в год) приходится 15% поверхности суши. Столько же занимают се-миаридные области (200-400 мм осадков в год). Территории аридных земель в основном приходятся на развивающиеся страны, в которых нормы потребления воды отличаются от индустриальных стран. По данным Всемирной организации здравоохранения (ВОЗ) в развивающихся странах лишь
