International Core Journal of Engineering 2020-26 | Page 24
Mass(m) function, and satisfies Equation (1) and (2).
m 0
¦ m A
m A
(1)
(2)
1
A 4
¦ m B , A 4 澳澳澳澳澳澳澳澳澳澳澳澳澳澳澳澳(3)
The plausible function refers to the degree that the
proposition A is not considered to false, expressed as
Equation (4).
¦ m B , A 4 澳澳澳澳澳澳澳澳澳澳澳澳澳澳澳(4)
In the reliability function Bel , if m A ! 0 , it is called a
focal element. For any of the focal elements A in the
recognition framework, if there are two evidences to support
it, that is, there are two reliability functions Bel 1 and Bel 2
on the same identification frame 4 , the corresponding
BPAs are m 1 and m 2 , and the focal elements are X and Y
respectively. The synthesis rules for the two reliability
functions are shown in Equation (5) and (6).
m A
K
1 K
¦
A n z
m 1 X m 2 Y , A z , A 4
(7)
, A z , A 4
m 1 A 1
m n A n
Find the average support
level through Murphy and
determine the main focus
A
¦
1
A n A
N
(5)
Is the primary focal
element unique
or its
value ≤ the uncertainty ?
X Y A
Y
¦
X Y
(8)
Therefore, the conflict evidence needs to be processed in
the multi-evidence fusion. The Murphy method can be used
to find the average support level of each evidence, and the
focal point with the highest average support degree is
selected to obtain the final support and become the main
focus. It is judged whether there is pseudo evidence in the
evidence source by calculating the support of other
evidences on the proposition of the main focus position, and
processing the detected pseudo evidence. The multi-
evidence fusion method considering the conflict evidence
processing is shown in Fig. 1.
B Az
A z
0,
°
® 1
° 1 K
¯
K
m n A n
B. Processing Method to Conflict Evidence
It can be found that the conflict coefficient K plays an
important role in the fusion of multiple evidences in the
practical application of evidence theory. If there are
evidences in the multiple evidences that provide the wrong
decision information, that is, the pseudo evidences, the
conflict coefficient K will increase. The accuracy of the
BPA calculated by Equation (7) will be significantly
reduced. Although there is a large amount of valid evidence
to make the evidence fusion can get the correct conclusion,
it will lead to slow convergence and low efficiency in the
process of evidence processing.
B A
Pls A
¦
m 1 A 1
A 1
Also called m function is the Basic Probability
Assignment (BPA) on the recognition framework 4 . In
addition, the reliability function and the plausible function
are important concepts in evidence theory [11]. The
reliability function Bel(A) is the sum of the reliability of the
evidence against A and its premises, expressed as Equation
(3).
Bel A
A
0,
°
®
° A 1
¯
Calculate the support of each
evidence for the main focus
proposition
m 1 X m 2 Y 澳澳澳澳澳澳澳澳澳澳澳澳澳澳澳澳澳澳澳澳澳澳(6)
where X Y indicates the conflict in the process of
fusion, the value K is the reliability lost after the fusion of
evidence, its size indicates the degree of evidence conflict:
K 0 means the evidence is completely consistent;
K 1 means the evidence is completely conflicted;
0 K 1 means the evidence is partially consistent.
Is there any false
evidence?
N
Y
Determine the weight of each
evidence using the support
for the primary coke
Similarly, when there is multiple evidences on the same
identification framework, the corresponding BPA is
m 1 , m 2 , , m n and the focal elements are A 1 , A 2 ,
,
A n respectively and the corresponding reliability function
synthesis rules are shown in Equation (7) and (8).
Evidence fusion using
evidence theory
Fig. 1. Multi-evidence fusion method considering conflict evidence
2