=Paper=
{{Paper
|id=Vol-2387/20190448
|storemode=property
|title=None
|pdfUrl=https://ceur-ws.org/Vol-2387/20190448.pdf
|volume=Vol-2387
|dblpUrl=https://dblp.org/rec/conf/icteri/IvchenkoK19
}}
==None==
https://ceur-ws.org/Vol-2387/20190448.pdf
Methodology for the Construction of Predictive Analysis
Systems as Exemplified by the Mining Equipment in the
Conditions of Big Data and Simulation Methods
Rodion Ivchenko1[0000-0003-4252-4825] and Andrey Kupin2[0000-0001-7569-1721]
1 Department of Automation Computer Science and Technologies
2 Department of Computer Systems and Networks
Kryvyi Rih National University
11, Vitaliya Matusevycha str., Kryvyi Rih, Ukraine
kupin.andrew@gmail.com, ivchenko.ra@gmail.com
Abstract. Currently, almost all elements of economic activity somehow exist ac-
cording to the laws of macroeconomics. This is facilitated by the rapid develop-
ment of international relations, the acceleration of logistics operations, political,
religious, cultural integration and unification at the level of interstate relations
and interactions (as a result of the evolution of state-formed market systems). At
the same time, every year more and more significant influence on macroeconom-
ics is played by minor global factors that were previously practically not taken
into account, such as climate change, growth of the world population, etc. Thus,
today, every company seeking to be efficient and profitable needs to focus not
only on the laws of the domestic market, but also on global trends when building
it’s strategies and implementing tactical tasks [1].
All the above prerequisites gave rise to a new concept of production develop-
ment, called Industry 4.0. Previous scientific and technological revolution led to
the automation of individual processes and devices, while Industry 4.0 provides
for the end-to-end digitization of all physical assets and their integration into the
digital ecosystem along with the assets of partners involved in the value chain.
The creation of the concept Industry 4.0 in the framework of solving the problems
of managing modern technological processes and production had several basic
prerequisites. One of them is related with the fact that the complication of the so-
called material part of production, of course, also leads to the complication of the
organizational component. It is becoming more difficult for a modern manager
to make the right management decisions. In the progression, the variability of the
applied goals, conditions, restrictions, and with them the scale of possible conse-
quences, increases. Another important reason is the fact that in modern manage-
ment conditions it is necessary not only to obtain statistics and analytics of pro-
duction, but also to be able to predict using the obtained data. High-performance
methods should be applied to isolate the most important and relevant information
at the time of the decision, with the possibility of a predictive analysis of possible
options for events.
Keywords: Industry 4.0, Big Data, Predictive analysis, Simulation.
1 Object Research
In general, the mathematical model of each stage of the hierarchy can be viewed as a
complex object of the mineral dressing technology and presented as a function of vari-
ables. Here, three types of actions serve as the input of the object (Fig.1) [2]:
1. Uncontrolled (but monitored) input variables Y = {у1,...уr} constitute a disturbance
vector and, as a rule, characterize, as far as concentrating production is concerned,
quality indicators of the source material to be processed and those of its intermediate
products obtained during the concentrating process;
2. Controlled input variables U = {u1,...,un} constitute a control vector and character-
ize, as a rule, quantitative indicators (expenditure) of material and energy flows;
3. The uncontrolled factors Z = {z1,...,zk} constitute an interference vector. Basically,
this is a disturbance vector, about which the developer of the control system knows
very little or nothing at all. Most often that vector is not taken into account at all.
Y
u1
y1 y2 y3 . у r
x1
u2 x2
u3 x3
U
. Object . X
.. ..
un xm
z1 z2 z3 . zk
Z
Fig. 1. The structure of a complex control object.
1.1 Monte Carlo Method as a Type of Simulation
The idea of the method is as follows. Instead of describing the process with the help of
an analytical apparatus (differential or algebraic equations), a random rally is performed
using a specially organized procedure that includes randomness and gives a random
result.
The essence of the Monte Carlo method is as follows: it is required to find the value
A of a certain quantity under study. To do this, choose a random variable X, the math-
ematical expectation of which is A, that is, M (X) = A. The probability of events in this
case is estimated using the frequency of outcomes in numerous sessions of simulation.
Practically do the following: produce N tests, in which they receive N possible val-
ues of X, calculate their arithmetic average and take it as an estimate (approximate
value) of the desired number A.
1.2 At the Application Level
In terms of value, this criterion (1) can be expressed in terms of the amount of losses
that a combine may incur as a result of a possible failure of an individual equipment
segment(s).
S у (1 K1 ) A K GК (1 K 2 )QК T min (1)
where K1 , K 2 are the coefficients characterizing the share of costs due to a decrease
in the quality of products and their quantity, respectively (due to failures or failures)
KGК - the coefficient of readiness of operation of the plant; A is the unit cost (UAH /
t); Q К - planned capacity of the plant (tons / hour); T - time of the plant for the year
[3].
It should be noted that the exact calculation of the coefficients K1 , K 2 is quite dif-
ficult, due to the multi-factorial nature of the latter. Therefore, we investigated the pos-
sibilities for the approximate calculation of these quantities, by predicting the probabil-
ity of equipment failures, as well as their possible consequences using simulation and
statistical methods.
For this, as applied to a specific scheme, it is a set of a finite number N of servicing
devices. Those any technical device is, in turn, a device for maintenance in the system.
The structural model of such a network is presented (see Fig. 12). Here, the designations
P1, P2, P3, ..., PN correspond to the numbers of instruments for maintenance.
P1 P2 Pn
P2 P3
Fig. 2. Structural model example.
It should be noted that absolutely all technical components may not be included in this
scheme. It is enough to enter only the most important (key) devices, the failure of which
can directly affect the work of the main divisions of the plant.
As the main indicators characterizing the reliability of a device, we will use the time
between failures t ( t - expectation of time between adjacent failures of the system
being restored) and the average time the device can recover after a failure t В . These
characteristics are usually determined by statistical tests and must be indicated by the
manufacturer of the equipment in the passport of the specific equipment.
Thus, for a given structural scheme of the designed IC, we have two vectors in the
form:
T t1 , t2 , t3 ,..., t N
(2)
TВ [tв1 , tв 2 , tв 3 ,..., tвN ]
where t , t , t ,...,t - the values of MTBF for the corresponding devices in the dia-
1 2 3 N
gram of Fig.1; tв1 , tв 2 , tв 3 ,..., tвN - average recovery time for each of the devices.
Next, a simulation model is built in which the failures of the service devices are
reproduced (simulated) and their possible consequences are predicted. The block dia-
gram of the algorithm that implements such a model is shown in Fig. 3.
Fig. 3. General algorithm for simulation model of failures.
1.3 The Use of Spreadsheets in Simulation
Imitation using tabular processors is a separate area with its own characteristics. Its
proponents argue that using these systems improves the understanding of the processes
taking place much better than using specialized software that has a high cost and takes
time to study, and also hides the mechanisms used (although such environments are
quite widely used because they provide more features and allows simulating complex
systems).
One of the advantages of modeling by means of spreadsheets is that the tabular
model allows you to visually reflect (see Fig. 4) the behavior of the system under study
with the help of charts and graphs.
Using spreadsheets, you can perform simulation modeling of discrete and continu-
ous non-deterministic dynamic systems. The basis for constructing a dynamic model of
a system is a description of the system using a system of differential equations or a
system of recurrent equations with given initial conditions.
When constructing system-dynamic models using spreadsheets, the time delay is
modeled using links to cell ranges corresponding to previous points in model time.
The technology of simulation using spreadsheets includes [4,5]:
setting discrete points in time;
description of the state of the system in a certain range of the table;
setting the rules for the transition of the system to the state corresponding to the next
point in model time.
To build non-deterministic models using MS Excel, the functions
RANDOMBETWEEN, RANDOMBETWEEN are used.
When creating a dynamic model using a spreadsheet, the modeling horizon is deter-
mined by building the required number of ranges (replicating ranges) corresponding to
discrete points in time.
If the capabilities of the tabular processor do not allow to cover the entire modeling
horizon, then the behavior of the system on a shorter modeling horizon should be in-
vestigated to put forward a hypothesis about its behavior throughout the modeling hori-
zon.
A multiple start of a simulation session in a tabular processor is performed by recal-
culating formulas in all cells of the table, which is started by pressing the F9 key.
FROM TO Device 1 Device 2 Device 3
Months of operation (warranty) 0 120 Detail 1 72 97 67
Detail 2 112 21 46
Detail 3 64 16 106
Simulated time (months) 50 Detail 4 45 42 101
Price 150 Detail 5 92 26 44
Detail 6 79 53 6
Detail 7 118 104 10
Detail 8 104 51 64
Detail 9 23 2 111
Detail 10 6 51 78
Detail 11 48 78 21
Detail 12 3 43 77
Number of breakdowns 5 6 5
Finance losses 750 900 750
Breakdown diagram
150
100
50
0
Detail Detail Detail Detail Detail Detail Detail Detail Detail Detail Detail Detail
1 2 3 4 5 6 7 8 9 10 11 12
Device 1 Device 2 Device 3
Fig. 4 Simulation in spreadsheet with breakdown diagram.
References
1. Magazine "neftegaz". “Industry 4.0: Key Tasks of Modern Enterprises”. [Electronic re-
source] – Available: https://neftegaz.ru/science/view/1414-Industriya-4.0-klyuchevye-
zadachi-sovremennyh-predpriyatiy-i-resheniya-kompanii-Yokogawa
2. A.M. Mariuta, Yu. G. Kachan, V. A. Bunko: “Automatic Control of Technological Pro-
cesses of Concentrating Plants”. Moskow «Nedra». p 1983.
3. Molchanov A.A. Modeling and design of complex systems .- K .: Vyshcha school, 1988.-
359p. 5. Denisov A.A., Kolesnikov D.N. Theory of large control systems. - L .: Energoizdat.
Leningr. separation, 1982.- 287p
4. “Simulation of dynamic systems”. [Electronic resource] – Available:
edu.tltsu.ru/sites/sites_content/site216/html/media67140/lec7_ts.docx
5. A. Kupin, I. Muzyka, “Information Technologies of Processing Big Industrial Data and De-
cision-Making Methods”, IEEE “Problems of infocommunications science and technology
2018”, p. 4, 2018.