=Paper=
{{Paper
|id=Vol-2105/10000227
|storemode=property
|title=Optimization of the Activity of Operators of Critical Systems by
Methods of Regulating Operational-Tempo Tension
|pdfUrl=https://ceur-ws.org/Vol-2105/10000227.pdf
|volume=Vol-2105
|authors=Evgeniy Lavrov,Nadiia Pasko
|dblpUrl=https://dblp.org/rec/conf/icteri/LavrovP18
}}
==Optimization of the Activity of Operators of Critical Systems by
Methods of Regulating Operational-Tempo Tension==
https://ceur-ws.org/Vol-2105/10000227.pdf
Optimization of the Activity of Operators of Critical
Systems by Methods of Regulating Operational-Tempo
Tension
Evgeniy Lavrov
Sumy State University,
Sumy, Ukraine
prof_lavrov@hotmail.com
Nadiia Pasko
Sumy National Agrarian University
Sumy, Ukraine
senabor64@ukr.net
Abstract. The model of the influence of the available time on the tension of op-
erators and the infallibility of their activity is considered. The productions and
methods of solving possible optimization problems are developed to search for
ergonomic reserves to increase the efficiency of critical systems.
Keywords. Critical system, IT resources, ergonomics, poly-ergative system, in-
cident management, human operator, algorithm of activity, operational-tempo
tension, self-control, optimization of activities.
1 Introduction
Modern technological, transport, aerospace and energy systems are inherently critical
[1,2] because even minor disruptions to their operation can lead to accidents, being
catastrophic in most cases. Decision-making processes in critical infrastructures in-
volve the need to process large amounts of information in real time [1,2]. Mainte-
nance of IT resources of critical systems in the effective state and the ensuring their
high reliability and efficiency are becoming an increasingly serious problem [1,2].
The analysis of the real operators’ activity made it possible to determine the fol-
lowing failure conditions for most cases [3–5]:
Operational-tempo tension of the activity is the cause of erroneous reactions;
There are no real mechanisms in place to ensure ergonomic quality, aimed at
providing the standards of operator’s activity;
Management of time constraints on the implementation of activity algorithms can
be one of the main reserves of increasing ergonomics.
Scientists working within the framework of the functional structural theory of er-
gotechnical systems of Prof. A.I. Gubinsky, have traditionally paid much attention to
the study of optimization problems for operators [6]. Characteristics of some new
problems, specific for polyergatic systems, are given in the paper [7]. Unfortunately,
despite the huge scientific reserve of that school, in the field of optimization of the
human-machine interaction, tasks, as a rule, are solved with the assumption of invari-
ability of the characteristics of the operator in the process of activity and without tak-
ing into account the effect of the time resource on the quality of implementation of the
algorithms available.
2 Problem Statement
The task is to identify current problems and develop appropriate models that allow
solving the problems of optimizing the discrete activity of operators by varying the
values of parameters that characterize time resources, i.e. by means of regulation of
operational-tempo intensity.
3 Approach
3.1 Models for Managing the Operational-Tempo Tensity
A Model for Assessing the Impact of the Operational-Tempo Tension on the
Quality of Activities. Professor P.P. Chabanenko [8] succeeded to identify and for-
malize the mechanism of the operator’s flexible response to the available time re-
source (based on the study of real engineering and psychological data).
Tension, as a psychological fee for achieving the goal of activity, includes several
components, the main ones being the tempo, determined by the time deficit to solve
the problem, and the operational one, determined by the nature of the operations of
the activity algorithm. The joint operational-tempo tension is determined by the sim-
ultaneous influence of two noted factors: The tension H defines the probability P0 of
switching on the self-monitoring of the current operation by the operator [8]:
P0 1.836H 2 0.962H 0.874, (1)
where H is a ratio of the time required to complete the operation at the maximum rate
to the allowable time actually given to the operator to perform this operation.
Optimal tension of the operator’s activity: Hopt = 0.262. The area H < Hopt corre-
sponds to an insufficient load on the operator, and the area H > Hopt corresponds to an
excessive load on the operator.
When H = Hopt, there is a probability to start a self-monitoring function P0 = 1,
which corresponds to the case of setting error-free operation.
When H = 1, the probability P0 = 0, which corresponds to the speed setting. An il-
lustration of settings types in the operator’s activity is shown in Fig. 1.
Fig. 1. An illustration of setting types in the operator’s activity: a) error-free; b) speed; c) flexi-
ble response to the resource of time; R – working operation; K – control operation (notations by
[6]).
Models of Activity Optimization by Means of Management of Operational-
Tempo Tensity. Time limitations are determined by the characteristics of the flows of
incoming signals or orders, or various organizational decisions (including the number
of operators in the shift and distribution of functions among them). They may be set
instructively by the managing operator or they can be established by software and
technical means of activity management.
Depending on the nature of the activity and the level of optimization (function,
complex of functions, etc.) we distinguish 2 classes of possible tasks.
Choosing the optimal time constraint for the implementation of activities (1st
class of tasks). Task 1.1 (single criterion):
There are given: The structure of the activity algorithm; characteristics of reliabil-
ity and execution time of operations, permissible time limits for the implementation of
the activity algorithm in the form of the lower limit of Tmin and the upper limit of Tmax
(may not be available); and the maximum permissible activity tensity H0. It is neces-
sary to choose the limit value T, imposed on the implementation time, which provides
the maximum probability of the error-free execution:
B(T ) max (2)
H (T ) H0 (3)
Tmin T Tmax (4)
where T is the allowable time for activity implementation; B (T) is a probability of
error-free execution; H(T) is an intensity of the activity; H0 is a maximum permissible
intensity; Tmin, Tmax are minimum and maximum times to implement activity.
Task 1.2 (multiple criterion):
The criterion for minimizing the intensity is added to the formulation of the Task
1.1. (2) – (4) instead of, or in addition to the restriction on the intensity (3).
H (T ) min (5)
Task 1.3 (optimization of income from activities):
The statement appears due to the need to solve the problem of the type "what is
better: a greater number of implementations of activity algorithm with low infallibility
or high infallibility with fewer implementations". The economic consequences of
correct (incorrect) implementations of activities can be evaluated in various ways, for
example, "income from correct implementation" and "damage from improper imple-
mentation".
Consider one of the possible formulations of the problem, taking into account the
economic consequences of the activity. The following statements are given in addi-
tion to the initial data of tasks 1.1 and 1.2, where C1 is an income from a single error-
free implementation of the activity; C2 is a damage from a single implementation of
the algorithm with an error; T’ is a total time during which the activity algorithm
should be repeated (taking into account the deduction of time for various breaks:
technological, for leisure, etc.). If we consider that the number of applications realized
in time is determined by the time allocated for a single implementation n=T’/T, the
target function will have the form
T'
(C1B(T ) C2 (1 B(T ))) max
T (6)
under the constraint (3)-(4).
Task 1.4 (multiple criterion analogue of task 1.3):
The criterion of intensity minimization (5) is added to the statement of task 1.3, in-
stead of, or in addition to the constraint (3).
In the formulations of tasks 1.1-1.4, there is no explicit restriction on the timely
implementation of the activity algorithm, since, as follows from the initial assump-
tions, the activity is pre-configured and the operator provides (at Tmax not less than the
minimum required time for implementation of the activities) timely execution (by
enabling or disabling the self-monitoring of operations).
The way to solve the tasks of choosing the optimal time limitations for activity im-
plementation:
Since the analytic dependence for B(T) cannot be defined in principle, and the
mathematical model, algorithm and program for calculation of the probability of er-
ror-free execution at a given point (for a given T) are developed (see [3–5]), we pro-
pose a numerical approach to the solution, based on determining values
T1,T2,…,Ts ϵ [Tmin,Tmax], the calculation of the criterion function for these values (in
case of single criterion optimization) or the "convolution" of the criteria functions (in
case of multi criterion optimization) and the values of the indices to which the con-
straint is imposed (H), and determining the optimal value or search area narrowing
(and the repetition of the procedure).
Optimization algorithms based on one-step or multi-step choice are possible. The
one-step choice consists of determining the number of points N, calculating values
T T
Ti Ti 1 max min , (7)
N
where i=1,…,N, T0=Tmin, computing B(Ti), H(Ti) (using algorithms and programs [3–
5,9,10]) or other criteria functions (depending on the task), and determining the opti-
mal value by a simple search. The disadvantage of the one-step choice is that a large
number of points N are needed to ensure high accuracy.
A multi-step search is implemented in a coherent strategy related to the following
actions:
1) The use of the selection rule for the first few points T1,T2,…,Ts ϵ [Tmin,Tmax], and
localization of the minimum point on the segment [T’min,T’max];
2) Choosing points TS+1,…,TK on the localized segment and determining the next
segment;
3) The analysis of the solution that was obtained for an admissible approximation
to the optimal one, and the optimal sequential search, carried out by one of the known
methods, for example, "Fibonacci search", "Golden section" method, etc.
Distribution of the directive time for realization of activities between their frag-
ments (2nd class of tasks):
Task 2: The statement makes sense in the case when the algorithm of the opera-
tor's actions is divided into several parts (fragments): each part is executed in the in-
terval between two events. And the first event (signal) is synchronized with the be-
ginning of the implementation of the algorithm corresponding to the activity frag-
ment, and before the second event (signal) arrives, the algorithm must be completed.
Such a situation usually occurs when fragments of system operation algorithms are
realized in parallel by a human operator and a machine (with a starter and a function-
ary of type "AND"[6]).
The fact that it is not always expedient to operate the automation machine with the
maximum possible speed is noted in a number of ergonomic studies: For example, in
some cases it is recommended to introduce a delay in the "computer response" in
order to ensure the comfort of the operator [8].
However, there is no way to determine the optimal reaction time in each particular
case. We assume that for technological or economic reasons the maximum permissi-
ble time T for the implementation of the algorithm of functioning (AF) of the entire
system is set. The task is to determine the directive times for the operator to perform
individual fragments performed between the signals of automatic means, ensuring the
maximum probability of error-free realization of the entire AF. At the same time,
restrictions on the intensity of activity on each of the fragments can be introduced.
Here is the definition of vector (T1,T2,…,Tn) that provides:
FB ( B1 (T1 ), B2 (T2 ),..., Bn (Tn )) max (8)
FT (T1 , T2, ..., Tn ) T (9)
H i (Ti ) H i0 , i 1,..., n (10)
where FB is a dependence of the probability of error-free execution of AF from Bi(Ti),
i=1,…,n, determined by the structure of AF; Ti is a directive execution time of the i-th
fragment of AF; Bi(Ti) is a probability of error-free execution of the i-th fragment of
AF; Hi(Ti) is strength of the i-th fragment; H0i is maximum permissible tension for the
i-th fragment; FT(T1,T2,…,Tn) is a dependence of AF execution time on the execution
time of its individual fragments, determined by the structure of AF; n is a number of
AF fragments. FB and FT are determined by the models developed for the typical
operational structures [3, 10].
In the particular case, when AF fragments are sequentially performed, the state-
ment (7)-(9) takes the form:
n
B (T ) max
i 1
i i (11)
n
T T
i 1
i (12)
H i (Ti ) H i0 (13)
The way of solving the task of directive time distribution to realize the activity be-
tween their fragments:
Step 1: Generation of a set of directive time values of the fragment implementa-
tion for each i-th fragment of i=1,…,n, for example, as follows:
1a. Definition of the range of research [Timin,Timax],
where Timin is the time required to implement the AF fragment without performing
functional control for all operations (for example, for a sequential chain of operations,
ni
i
Tmin TPij , (14)
j 1
where ni is the number of basic work operations in the i-th fragment, TPij is the math-
ematical expectation of the execution time of the j-th operation in the i-th fragment);
Timax is the maximum time value for the AF fragment implementation, provided
that all operations are performed with self-monitoring and taking into account the
variance of the execution time, for example, for a sequential chain of operations
ni ni
i
Tmax TPKij 3( DPKij )1 / 2 , (15)
j 1 j 1
where TPKij and DPKij are the mathematical expectation and a variance of the execution
time in the i-th fragment of AF of the j-th self-monitoring operation.
1b. Definition of the number of investigated values of the directive execution time
Ni for each i-th fragment i=1,…,n, and generation of the following values:
Ti,1,Ti,2,…,Ti,Ni ϵ [Timin,Timax], in the simplest case
i
Tmax Tmin
i
Ti , k Ti , k 1 , (16)
Ni
where i=1,…,n and Ti,0=Timin.
Step 2. Evaluation of error-free and activity intensity values on each i-th fragment,
i=1,…,n, with all the generated values of the directive time: Bi(Ti,j), Hi(Ti,j), i=1,…,n,
and j=1,…,Ni. The evaluation is carried out by using models and programs developed
in [3–5, 9 and 10].
Step 3. An exception for each i-th activity fragment of Ti,j(i=1,…,n, j=1,…,Ni)
those values that do not satisfy the constraints on the activity tensity Hi(Ti,j)≤H0i.
Step 4. Choosing the best option. With the generated variants of the execution of
fragments with available estimates of the quality indicators, the task is reduced to the
standard problem of maximizing the probability of error-free execution of AF under
the constraint on the mathematical expectation of the execution time [6,7 and 10].
Step 5. Investigation of the vector obtained as a result of the solution of the opti-
mization problem. (T*1,T*2,…,T*n); generation of directive time values for each i-th
fragment in the range [T*i – ∆i, T*i + ∆i], (Δi can be determined, for example
i
Tmax Tmin
i
i (17)
Ni
(see 1b)), and transition to step 2 (steps 2-5 should be repeated until the solution with
required accuracy is obtained).
The convenience of the approach is that one of the steps reduces the task to the
known problem of optimizing a human machine system and can be solved by any
known method [6,7, 9,10] with any structure of connections between operations.
4 Applications
The models were used in the development of the following systems ensuring ergo-
nomic quality:
Automated processing plants for various purposes [3];
Systems of technical support of IT resources in telecommunications [9,10];
Distributed banking processing systems [4];
Relevant topics of the training courses on "Ergonomics" and "Information Sys-
tems" at the Sumy State University, the Sumy National Agrarian University and
the National University of Bio-Resources and Nature Management (Kyiv).
5 Conclusion
Operational-tempo tension is determined by time constraints on the activity of opera-
tors of critical systems and significantly affects the indicators of error-free function-
ing. It is convenient to evaluate the influence of time resources on the efficiency of
ergotechnical systems by using models based on formalisms introduced by Prof.
P.P. Chabanenko.
Solving the tasks of choosing temporal constraints for the implementation of ac-
tivities and the distribution of directive time for the execution of work between its
fragments makes it possible to ensure the fulfillment of specified requirements for the
operation of critical systems while observing the normative indicators of the intensity
of the operators’ activities. Solving the problems of choosing a temporary restriction
on the implementation of activities and the distribution of directive time for the im-
plementation of activities between its fragments can provide specified requirements
for the effectiveness of critical systems, while observing the normative values of the
activity intensity of operators.
The models should be used as elements of mathematical support of DSS for op-
erators-managers of critical systems.
References
1. Bloomfield, R., Lala, J.: Safety-Critical Systems. The Next Generation / IEEE Security &
Privacy 11(4),11–13 (2013). doi: 10.1109/MSP.2013.95.
2. Skatkov, A., Anikevich, K., Shevchenko, V.: Information model of management of IT re-
sources of critical systems. Collection of scientific works of Sevastopol National Universi-
ty of Nuclear Energy and Industry 1(37), 201–206 (2011). (In Russian)
3. Lavrov, E., Pasko, N., Krivodub, A., Barchenko, N., Kontsevich, V.: Ergonomics of IT
outsourcing. Development of a mathematical model to distribute functions among opera-
tors. Eastern European Journal of Enterprise Technologies 4(80), 32–40 (2016). doi:
10.15587/1729-4061.2016.66021.
4. Lavrov, E., Pasko, N., Tolbatov, A., Tolbatov, V.: Ergonomic Reserves for Improving Re-
liability of Data Processing in Distributed Banking Systems. In: Proceedings of 2nd Inter-
national Conference on Advanced Information and Communication Technologies-2017
(AICT-2017), pp. 79–82. Lviv, Ukraine, July 4–7, (2017).
5. Lavrov, E., Barchenko, N., Pasko, N., Borozenec, I.: Development of models for the for-
malized description of modular e-learning systems for the problems on providing ergo-
nomic quality of human-computer interaction. Eastern-European Journal of Enterprise
Technologies. Ser. “Information technology” 2(2/86), 4–13 (2017). doi: 10.15587/1729-
4061.2017.97718.
6. Adamenko, A.N., Asherov, A.T., Lavrov, E.A. et al.: Information controlling human-
machine systems: research, design, testing, Reference book, Gubinsky, A.I. & Evgrafov,
V.G., eds. Mechanical Engineering, Мoscow, (1993). (In Russian).
7. Lavrov, E., Pasko,N.: Approach to optimization models bank forming for distributing of
functions between the ACS operators. Eastern-European Journal of Enterprise Technolo-
gies. Ser. “Mathematics and Cybernetics – Fundamental and Applied Aspects” 1(4/49),
46–50 (2011). (In Russian).
8. Chabanenko, P.: Investigation of the safety of the functioning of the "human-technical"
systems by ergonomic nets. Publishing House of the P.S. Nahimov Academy of the Navy,
Sevastopol, (2012). (In Russian).
9. Lavrov, E., Pasko, N., Krivodub, A., Tolbatov, A.: Mathematical models for the distribu-
tion of functions between the operators of the computer-integrated flexible manufacturing
systems. In: Proceedings of the XIII-th international conference TCSET-2016 “Modern
problems of radio engineering, telecommunications, and computer science”, pp. 72–76.
Lviv-Slavsko, Ukraine, february 23 – 26, (2016).
10. Lavrov, E., Pasko, N., Tolbatov, A., Tolbatov, V.: Cybersecurity of Distributed Infor-
mation Systems. The Minimization of Damage Caused by Errors of Operators During
Group Activity. In: Proceedings of 2nd International Conference on Advanced Information and
Communication Technologies-2017 (AICT-2017), pp. 83–87. Lviv, Ukraine, July 4 – 7, (2017).