2016 Joint International Conference on Service Science, Management and Engineering (SSME 2016) and
International Conference on Information Science and Technology (IST 2016)
ISBN: 978-1-60595-379-3
A Risk Transfer Control Strategy Based on Nodes Configure Constraint
Entropy According to Its Importance
Song WANG
Engineering University of People's Armed Police, Xi’an, China
[email protected]
Keywords: Accident Network, Node Importance, Configuration of Constraint Entropy, Risk Transfer,
Arena Simulation.
Abstract. In order to provide new measure for risk transfer emergency control, reconstructed an
empirical accident network and used the Arena software to simulate the configuration of emergency
constraint entropy in network. Simulation results show that when the emergency constraint entropy is
shared by an important node, system has the best control effect of risk transfer, and the results lay a
theoretical foundation for improving the efficiency of resource allocation in risk control.
Introduction
Emergence is a system's widespread characteristic, through the emergence of control to make the
system emerge out of the desired function has practical significance. Emergence control means based
on the emergence of the mechanism, by changing the individual attributes and the rules of the
organization between each other, and the external environment out of the system and so on, make the
directional emerged expectations macro function and the characteristic of the control method [1]. For
the accident system, to control the emergence of accident, it can change the three aspects of the
characteristics of the node, the associated characteristics and the external environment. But to
implement the emergence of directional control, it must realize the transition from qualitative
description to quantitative analysis. When accident node occurrence risk emergence, corresponding to
the formation of a new risk state. The transfer of risk state caused by the relationship between nodes is
called risk transmit, which involves the transfer mechanism, transfer path, transfer carrier and so on.
The concept of risk transfer is similar to risk conduction, which first appeared in the financial field,
The concept of risk conduction was first put forward in “Fever, Panic, Collapse—Historical Review
of Financial Crisis” by Charles Kindleberger (1996)[2]. Undetwood S (2009) constructed a multi
market price volatility model to study the cross dynamic transmission in the price and the volatility of
the stock and bond market in Europe and the United States. Wang Yongqiao (2011) studied the
problem of financial risk transmission based on time varying Copula function[4]. Li Cunbin
constructed the Markov - Fourier series modified gray prediction model (MFGM) to predict the chain
structure of the project risk element transmission. The model is intended to provide a high precision
prediction model and method for the risk element transmission of chain structure [5]. The above
research shows that the dynamic characteristics of the risk transfer can make the results more
consistent with the actual results. Buzna et al. constructed a universal dynamic model of disaster
spreading, On the basis of this study, Li Zequan et al. studied the influence of network centrality on
the speed and the trend of disaster spreading [7], The above results provide a good reference for the
construction of risk transfer dynamic model of our project, but from model to simulation, and then
through simulation to fit the equation of the research is little. In fact, the simulation analysis is more
suitable to reproduce the accident process or conduct accident behavior in advance, at the same time,
it can provide numerical reference for the construction of transfer model.
In order to prevent the safety accident, we must carry out effective control on all kinds of risk
behaviors of the system. The evolution and transition of system structure under the impact of
interference of the inner and outside world will produce micro internal entropy, To suppress the
collapse of the system requires the introduction of the macro negative entropy to maintain the stability
of the outside world, The process of dissipation of negative entropy is the procession of controlling
the safety accident. Peter Okoh et.al (2013)believed that to implement risk control of technology and
organizational complexity of the growing industrial system, it needs deploy multiple levels of safety
constraints. While the safety constraint integrity protection comes from the timely and reliable
maintenance [8]. Tom Kontogiannis (2012) proposed a regression model based on system theory,
which mainly integrated organizational factors control model and human behavior control model. The
regression model can show a lot of organizational factors, which also pointed out that how should
human beings change their behavior to adapt to environmental change [9]. However, the structure of
complex systems is a kind of potential collapse tendency of brittle structure [10], To ensure the safety
of many nodes and complex association of the operation system, we could implicate and control the
important nodes to inhibit the breakdown of the system. This paper is based on identification on the
network of accident of important node, we used the important degree of node based on level
configuration constraints entropy to achieve the emergency control on risk transfer.
Reconstructed Model of Accident Network
The operation of complex system has characteristics, which are costing a lot and great risks.
Although we can get exact results in real testing, other sides of this system will change a lot.
Meanwhile, it is bad for “accident experiment”. So it is economic and feasible obviously by setting up
logical or mathematical model and describing complex system behaviors now or in the future trough
simulation analyzing. And logical model can be solved by computer program. In this time, we can
predict different system behaviors by changing input parameters in program.
α1
α2
α11
α3
α4
α9
α6
α14
α7
α10
α13
α5
α8
α12
α15
Figure 1. Reconstructed accident network model.
Aviation complex system is a big and complex system covering elements, such as people,
equipment, environment, administration and information, which composes of people-machine- ring,
and has miscellaneous, enormous and interdisciplinary characteristics. Because aviation complex
system stays in service condition of some fields coupling, so it is very complex in evolution laws. This
paper is based on aviation complex system accident net model which is on the basis of literature. By
reconstructing I get new accident net model as diagram 1 shows. In this diagram, relationships of
elements’ gradations and constructions decide delivering directions of risks. Direction is transiting
from natural reasons to beside reasons as a whole. But in fact, the relationship between factors may be
more complex. Such as weak safe culture may affect the cognitive deficits in turn, or uncertainty leads
to nonlinear couple, or brittleness of system structure leads to uncertainty and so on. So in figure 5.4 I
consider risks transfer in turn from node 3 to node13, from node 14 to node 5, form node 10 to node
14. All these mean risks in accident system transfer from downstream node to upstream node. By
Arena simulation on accident net can be used in analyzing impactions which is how characteristics of
net impacts the ability of complex system risk management.
Arena Simulation Results and Analysis
Sort Node Important Degree Parameter
Figure 1 showed accident model, relatively node important degree 2-14 (Node 1 and 15 do consider
temporarily), in accordance with entropy algorithm steps build multidimensional data, obtained as
shown in Table 1 Entropy build multidimensional data obtained results will be sorted.
Table 1. Entropy build multidimensional data sorted results.
Node
Build Result
Sequence
Node
Build Result
Sequence
2
0.357
8
9
0.305
10
3
0.467
5
10
0.422
6
4
0.285
11
11
0.656
3
5
0.334
9
12
0.176
13
6
0.918
1
13
0.845
2
7
0.250
12
14
0.586
4
8
0.372
7
From Table 1, the degree of importance of node 6, followed by the node 13, node 11, node 14 and
node 3. By the early Arena simulation, it shows significant degree evaluation based on entropy
buildup cube node result is the most optimized, by imparting certain properties of the important nodes
sorted under maximum system availability to improve risk management capabilities. Below to
configure certain constraints entropy important nodes of the lower ranking results to verify that the
node network implicated in the control of the accident risk transfer control effect.
Analysis of Emergency Constraint Entropy Configuration When Important Node Exclusive
Enjoy It
When the configuration of the contingency constrained entropy is shared by all nodes, ,although
simulation time can shortened, but the maximum residual risk entropy is no change, further
consideration of the contingency constrained entropy configuration is an important node alone
occupied, and considering the change of system entropy risk, cost and time at this time. Through
respectively monitor node 6,13,11 three important nodes in risk entropy change, and exclusive of the
configuration of the contingency constrained entropy, as Figure 2 shows the important node exclusive
emergency entropy constrained premise system risk entropy variables such as changes of figure, In
simulation design, also based on the important node to be processed risk entropy monitoring, When
the risk entropy is greater than 3, the emergency of constraint entropy is configured, and used only for
risk disposal of the important nodes.
450
450
(113,388)
(113,388)
(105,371)
360
360
(112,319)
(565,313)
(565,313)
180
(902,135)
90
System Risk Entropy-C
System Risk Entropy-P
Remnants Risk Entropy-C
Remnants Risk Entropy-P
0
0
200
400
Time (h)
600
800
Risk Entropy
Risk Entropy
270
270
(777,235)
180
90
System Risk Entropy-C
System Risk Entropy-P
Remnants Risk Entropy-C
Remnants Risk Entropy-P
0
1000
0
100
200
300
400
500
600
700
800
900
Time (h)
a. Node 6 enjoy alonely
b. Node 13 enjoy alonely
450
350
(104,383)
360
(97,366)
300
(842,309)
(863,293)
Average System Risk Entropy-C
Average System Risk Entropy-P
Average Remnants Risk Entropy-C
AverageRemnants Risk Entropy-P
Risk Entropy
Risk Entropy
250
270
200
180
150
90
System Risk Entropy-C
System Risk Entropy-P
Remnants Risk Entropy-C
Remnants Risk Entropy-P
0
0
200
400
Time (h)
600
800
100
50
Node 6
1000
c. Node 11 enjoy alonely
Node 13
Node 11
d. Average Risk Entropy(100 Times)
480
6500
6000
Average Time (h)
Cost ($)
420
5500
Average Accumulated Cost-C
Average Accumulated Cost-P
5000
4500
360
Average Simulation Time-C
Average Simulation Time-P
300
Node 6
Node 13
Node 11
Node 6
Node 13
Node 11
e. Average Accumulated Cost
f. Average Simulation Time
(C: Resourse is Communal; P: Resourse is Private)
Figure 2. The change of system risk entropy when important node exclusive
enjoy emergency constraint entropy.
Figure 2 a, b, c three figure is based on the results of a simulation proceeds, d, e, f is based on
simulation 100 times the average results, figure c represents the contingency constrained entropy is
shared by all nodes, P represents the contingency constrained entropy is only an important node in the
exclusive. We can see from the chart, the node important degree is high, under the same conditions,
the maximum residual risk entropy is the smallest, and the average system risk entropy and the
average entropy is the smallest residual risk. With decreasing of node important degree, the system
average residual risk entropy and sharing the contingency constrained entropy is close to, which
shows only that node 11 and 13 exclusive contingency constrained entropy does not reduce the
residual risk entropy, node 6 exclusive emergency entropy constrained residual risk entropy decline is
larger, but the residual amount of still more. And from the average cost and the simulation time, the
cost of node 6 is the highest, the longest time, so it is still necessary to further optimize the allocation
of emergency response constraints.
Analysis of Emergency Constraint Entropy Configuration When Important Node Enjoy It
Accordance with the Grade
In order to further improve the effect of emergency constraint entropy allocation on the reduction of
residual risk entropy, According to the order of node 6,13,11, the three nodes jointly occupy the
emergency constraint entropy according to the important degree priority, we can get the curve of
system risk entropy variables of important nodes as shown in Figure 3 by the level of sharing
emergency constraint entropy corresponding to the system without emergency constraint entropy
configuration ); [1] corresponds to the case of node 6, which occupies the emergency constraint
entropy; [2] corresponds to the case of node 13, which occupies the emergency constraint entropy; [3]
corresponds to the case of node 11, which occupies the emergency constraint entropy; [4] corresponds
to three important nodes in the case of sharing the emergency constraint entropy. From the graph,
When three important nodes share the emergency constraint entropy, The maximum residual risk
entropy is effectively controlled, And the maximum system risk entropy is also the lowest. The
average results obtained from the 100 simulation are compared, Three important nodes share the
emergency constraint entropy. There is minimum mean system risk entropy and minimum mean
residual risk entropy. And has the smallest average simulation time. This shows that the allocation
scheme of the key nodes to share the emergency constraint entropy is the most optimized.
450
450
(106,389)
(101,391)
(105,371)
360
Risk Entropy
(1120,315)
(112,319)
270
(778,235)
(97,215)
180
(903,135)
System Risk Entropy-[4]
System Risk Entropy-[3]
System Risk Entropy-[2]
System Risk Entropy-[1]
Remnants Risk Entropy-[4]
Remnants Risk Entropy-[3]
Remnants Risk Entropy-[2]
Remnants Risk Entropy-[1]
90
(362,20)
0
Risk Entropy
360
(1120,313)
270
(97,215)
180
System Risk Entropy-[0]
System Risk Entropy-[4]
Remnants Risk Entropy-[0]
Remnants Risk Entropy-[4]
90
(362,20)
0
-90
0
260
520
780
1040
1300
0
Time (h)
1040
1300
b. Compare to Resource Private Fruition
Average Simulation Time
550
500
Average Time (h)
300
Risk Entropy
780
600
Average System Risk Entropy
Average Remnants Risk Entropy
350
520
Time (h)
a. Compare to Resource Communion
400
260
250
200
150
450
400
350
300
100
250
50
200
0
1
2
Simulation Condition
3
c. Average Risk Entropy
4
0
1
2
Simulation Condition
3
4
d. Average Simulation Time
Figure 3. The change of system risk entropy when important node enjoy emergency
constraint entropy accordance with the grade.
Summary
The allocation of emergency constraint entropy corresponds to the emergency handling of the
accident. In general, the emergency constraint entropy is limited, so it should be configured in the
most urgent needs of the node, so as to ensure that the limited resources to obtain the maximum risk
management benefits. Simulation and analysis of the contingency constrained entropy is shared by all
nodes, the first three important node exclusive and three important nodes according to the level of
sharing three allocation strategy, and analyzes the effect of emergency entropy constrained the
number on the performance of the system. The results show that:
(1)When only monitor the first three important nodes, and by monitoring the node exclusive
contingency constrained entropy, with the increase of node important degree, residual risk entropy
decreases gradually. When the monitoring node 6 risk entropy is accumulated, and let its exclusive
contingency constrained entropy, although the residual risk entropy is the smallest, but average cost
and the simulation time, the longest, the allocation strategy still need to be improved.
(2)The contingency constrained entropy before being three important nodes according to the level of
sharing (namely according to node important degree given the level of the contingency constrained
entropy), maximum system risk entropy and residual risk entropy have been effectively controlled,
and simulation time is the shortest. This shows that the corresponding emergency constraint entropy is
given according to the importance of a few key nodes, which can significantly improve the system's
risk management ability.
Acknowledgment
This study was partially supported by the National Natural Science Foundation of China with the
Grant number of 71401179 and Basic research fund of Engineering University of People's Armed
Police with the Grant number of WJY 201608 and WJY 201410.
References
[1] Tianyun Huang, Xue-bo Chen,Wei Wang, et al. Thinking in Emergence Control for Complex
System[A].Proceedings of the 8th World Congress on Intelligent Control and
Automation[C].June,Taipei,Taiwan,2011:21-25.
[2] Charles P. Kindleberger. Manias, Panics and Crashes: A History Financial Crisis[M]. Macmilian
Press Ltd,1996.
[3] Undetwood S. The cross-market information content of stock and bond order flow[J]. Journal of
Financial Markets, 2009, 13(2):268-289.
[4] WANG Yongqiao, Liu Shiwen. Financial market openness and risk contagion: A time-varying
Copula approach [J].Systems Engineering-Theory & Practice,2011,31(4):778-784.
[5] LI Cun-bin, Li Peng, Lu Gong-shu. Research of Prediction of Enterprise Projects Chain Risk
Element
Transmission
Based
on
Markov-fourier
Modified
Grey
Prediction
Model(MFGM)[J].Operations Research and Management Science,2013,22(4):241-247.
[6] Buzna L, Peters K, Helbing D. Modeling the dynamics of disaster spreading in
networks[J].Physica A,2006(363):132-140.
[7] Li Ze-Quan, Zhang Rui-Xin, Yang Zhao, et al. Influence complex network centrality on disaster
spreading[J].Acta Physica Sinica,2012,61(23):238902-l-7.
[8] Peter Okoh, Stein Haugen. Maintenance-related major accidents: Classification of causes and
case study[J].Journal of Loss Prevention in the Process Industries,2013,26(6):1060-1070.
[9] Tom Kontogiannisa, Stathis Malakisb. Recursive modeling of loss of control in human and
organizational processes: A systemic model for accident analysis[J].Accident Analysis and
Prevention,2012(48):303-316.
[10] WANG Song, WANG Ying. Causation Analysis of Complex System Safety Accident Based on
Brittle Structure Collapse Theory[J].Procedia Engineering,2011(15):365-369.