1.背景
2024年,H Zhang受到黑鹰生物规律行为启发,提出了黑鹰优化算法(Black Eagle Optimizer, BEO)。


2.算法原理
2.1算法思想
BEO结合了黑鹰的生物规律和数学变换来指导个体的搜索行为。
2.2算法过程
围捕行为
X
1
t
+
1
=
X
r
t
−
α
⋅
r
1
⋅
(
t
1
⋅
X
b
e
s
t
t
−
X
k
t
)
X
2
t
+
1
=
u
b
+
l
b
−
X
1
t
+
1
X
t
+
1
=
Minimize fitness
(
X
1
t
+
1
,
X
2
t
+
1
,
X
t
)
(1)
\begin{aligned} &X_{1}^{t+1} =X_{r}^{t}-\alpha\cdot r_{1}\cdot\left(t_{1}\cdot X_{best}^{t}-X_{k}^{t}\right) \\ &X_{2}^{t+1} =ub+lb-X_{1}^{t+1} \\ &X^{t+1} =\text{Minimize fitness}(X_{1}^{t+1},X_{2}^{t+1},X^{t}) \end{aligned}\tag{1}
X1t+1=Xrt−α⋅r1⋅(t1⋅Xbestt−Xkt)X2t+1=ub+lb−X1t+1Xt+1=Minimize fitness(X1t+1,X2t+1,Xt)(1)
其中,Xr为搜索空间中的随机位置,Xk为随机黑鹰的位置,Xbest为当前最佳位置,表示猎物的位置。参数
α
\alpha
α表述为:
α
=
e
−
∥
X
t
−
X
b
e
s
t
t
∥
/
D
(2)
\alpha=e^{-\left\|X^{t}-X_{best}^{t}\right\|/D}\tag{2}
α=e−∥Xt−Xbestt∥/D(2)
靠近行为
X
t
+
1
=
X
b
e
s
t
t
+
m
(
X
t
−
X
b
e
s
t
t
)
(3)
X^{t+1}=X_{best}^t+m(X^t-X_{best}^t)\tag{3}
Xt+1=Xbestt+m(Xt−Xbestt)(3)
其中,m是d维悬停矩阵表述为:
m
=
[
cos
a
−
sin
a
0
⋯
0
sin
a
cos
a
−
sin
a
⋱
⋮
0
sin
a
⋱
⋱
0
⋮
⋱
⋱
cos
a
−
sin
a
0
⋯
0
sin
a
cos
a
]
(4)
\left.m=\left[\begin{array}{ccccc}\cos a&-\sin a&0&\cdots&0\\\sin a&\cos a&-\sin a&\ddots&\vdots\\0&\sin a&\ddots&\ddots&0\\\vdots&\ddots&\ddots&\cos a&-\sin a\\0&\cdots&0&\sin a&\cos a\end{array}\right.\right]\tag{4}
m=
cosasina0⋮0−sinacosasina⋱⋯0−sina⋱⋱0⋯⋱⋱cosasina0⋮0−sinacosa
(4)
狩猎行为
X
∗
=
2
X
t
−
X
b
e
s
t
t
+
s
0
⋅
D
1
X
t
+
1
=
X
∗
+
D
2
⋅
(
X
b
e
s
t
t
−
X
∗
)
(5)
\begin{aligned} &X^{*}=2X^{t}-X_{best}^{t}+s_{0}\cdot D_{1} \\ &X^{t+1}=X^{*}+D_{2}\cdot(X_{best}^{t}-X^{*}) \end{aligned}\tag{5}
X∗=2Xt−Xbestt+s0⋅D1Xt+1=X∗+D2⋅(Xbestt−X∗)(5)
其中,D1,D2为位置调整因子,拟了黑鹰在捕捉猎物时不断调整姿态以确保捕获成功的情况。
抓取行为
X
t
+
1
=
X
b
e
s
t
t
+
e
r
3
π
(
X
b
e
s
t
t
−
X
t
)
(6)
X^{t+1}=X_{best}^t+e^{\frac{r_3}{\pi}}(X_{best}^t-X^t)\tag{6}
Xt+1=Xbestt+eπr3(Xbestt−Xt)(6)
黑鹰从一点到另一点曲线飞行时的抢夺行为,其利用点的跳跃运动来模拟黑鹰的抢夺行为。
警告行为
X P t + 1 ( i , j ) = X b e s t t ( i , j ) + ( X d t ( i , j ) − X b e s t t ( i , j ) ) ⋅ ( 1 + P ( j ) ) X t + 1 ( i ) = X P t + 1 ( i ) + β ( X b e s t t ( i ) − X P t + 1 ( i ) ) (7) \begin{aligned}X_P^{t+1}(i,j)&=X_{best}^t(i,j)+(X_d^t(i,j)-X_{best}^t(i,j))\cdot(1+P(j))\\\\X^{t+1}(i)&=X_P^{t+1}(i)+\beta(X_{best}^t(i)-X_P^{t+1}(i))\end{aligned}\tag{7} XPt+1(i,j)Xt+1(i)=Xbestt(i,j)+(Xdt(i,j)−Xbestt(i,j))⋅(1+P(j))=XPt+1(i)+β(Xbestt(i)−XPt+1(i))(7)
迁移行为
X
t
+
1
=
X
b
e
s
t
t
+
z
⋅
s
1
⋅
(
X
t
−
t
2
⋅
X
b
e
s
t
t
)
(8)
X^{t+1}=X_{best}^t+z\cdot s_1\cdot(X^t-t_2\cdot X_{best}^t)\tag{8}
Xt+1=Xbestt+z⋅s1⋅(Xt−t2⋅Xbestt)(8)
迁移机制旨在将适应度较低的个体从当前最优位置移开,以降低陷入局部最优的可能性。
求爱行为
X
j
t
+
1
=
{
X
j
t
+
r
4
⋅
k
⋅
(
X
b
e
s
t
t
−
X
j
t
)
+
r
5
⋅
cos
(
π
j
n
)
⋅
(
X
b
e
s
t
t
−
X
j
t
)
,
j
%
2
=
1
X
j
t
+
r
6
⋅
k
⋅
(
X
b
e
s
t
t
−
X
j
t
)
+
r
7
⋅
sin
(
π
j
n
)
⋅
(
X
b
e
s
t
t
−
X
j
t
)
,
j
%
2
=
0
(9)
X_j^{t+1}=\begin{cases} X_j^t+r_4\cdot k\cdot\left(X_{best}^t-X_j^t\right)+r_5\cdot\cos(\frac{\pi j}{n})\cdot\left(X_{best}^t-X_j^t\right), j\%2=1\\ X_j^t+r_6\cdot k\cdot\left(X_{best}^t-X_j^t\right)+r_7\cdot\sin(\frac{\pi j}{n})\cdot\left(X_{best}^t-X_j^t\right) , j\%2=0\end{cases}\tag{9}
Xjt+1={Xjt+r4⋅k⋅(Xbestt−Xjt)+r5⋅cos(nπj)⋅(Xbestt−Xjt),j%2=1Xjt+r6⋅k⋅(Xbestt−Xjt)+r7⋅sin(nπj)⋅(Xbestt−Xjt),j%2=0(9)
这里类似正弦余弦函数的波动和奇数-偶数的交替模式,模拟了黑鹰求偶期间雌雄黑鹰的互动行为。
孵化行为
X t + 1 = X b e s t t + ( X d t − X b e s t t ) ⋅ ( 1 + 2 R ) (10) X^{t+1}=X_{best}^{t}+\left(X_{d}^{t}-X_{best}^{t}\right)\cdot(1+2R)\tag{10} Xt+1=Xbestt+(Xdt−Xbestt)⋅(1+2R)(10)
流程图

伪代码

3.结果展示


4.参考文献
[1] Zhang H, San H, Chen J, et al. Black eagle optimizer: a metaheuristic optimization method for solving engineering optimization problems[J]. Cluster Computing, 2024: 1-33.

被折叠的 条评论
为什么被折叠?



