Table 4 Summary of parameter settings for NCRBMO and competitive algorithms on the CEC2017.
Algorithm | Parameter Name | Value/Range | Parameter Description | Function and Impact |
|---|---|---|---|---|
NCRBMO* | N | 30 (fixed) | Population size | Determines the breadth of algorithm search capability |
T | 500 (fixed) | Maximum iterations | Controls algorithm termination condition | |
Dim | 30 (fixed) | Problem dimension | Number of variables in the optimization problem | |
φ | 0.5 | Switching coefficient | Controls which phase the algorithm enters | |
p | [2, 5] default | Small group size | Number of randomly selected individuals for calculating small group mean | |
q | [10, N] default | Large group size | Number of randomly selected individuals for calculating large group mean | |
F | 0.5 | Local exploitation attack step factor | Controls step size in local exploitation phase | |
γ | 1.5 (fixed) | Lévy flight distribution parameter | Controls heavy-tailed characteristics of Lévy flight steps | |
µ | Normal(0, σ) random selection | Random direction vecto | Introduce random scaling to change the step size. | |
ν | Normal(0, 1) random selection | Random scaling vector | Introduce random scaling to change the step size. | |
m | dim/2 | Boundary dimension threshold | Dimension boundary separating Tent mapping and Chebyshev mapping | |
n | {2,3,4,5} random selection | Chebyshev polynomial order | Controls complexity and diversity of Chebyshev mapping: n = 2(unimodal), n = 3(bimodal), n = 4(multimodal), n = 5(complex distribution) | |
α | 0.499 | Tent mapping parameter | Controls piecewise linear characteristics of Tent mapping, affects ergodicity of chaotic sequences | |
Q | [0,1] uniform distribution | Normal cloud model selection probability | Q ≥ 0.5 uses normal cloud model for position update | |
Ex | BestX | Normal cloud model expectation | Uses current optimal solution as expectation center | |
En | exp(t/T) | Normal cloud model entropy | Increases exponentially with iteration, controlling distribution range | |
He | En/10^(−3) | Normal cloud model superentropy | Controls entropy uncertainty, affects search diversity | |
PO | α_PO | rand(1)/5 | Behavior regulation factor | Controls behavior step size, varies randomly with iteration |
θ_PO | rand(1) * π | Angle parameter | Random angle value affecting search direction | |
St | [1, 4] random integer | Behavior selection parameter | Determines which behavior strategy the parrot adopts | |
H | [0,1] uniform distribution | Communication behavior sub-strategy selection probability | H < 0.5 uses mean-based strategy, H ≥ 0.5 uses exponential decay strategy | |
HO | I1, I2 | [1, 2] random integer | Exploration phase coefficients | Control coefficients in position update formula |
Ip1 | [0, 1] random integer | Inverse correlation coefficient | Increases randomness | |
α_HO | 5 different methods random selection | Coefficient matrix | Contains 5 different random coefficient generation methods, increasing diversity | |
b | [2, 4] uniform distribution | Predator defense parameter b | Controls numerator coefficient in defense formula | |
c | [1, 1.5] uniform distribution | Predator defense parameter c | Controls denominator constant in defense formula | |
d | [2, 3] uniform distribution | Predator defense parameter d | Controls cosine term coefficient in defense formula | |
l | [−2π, 2π] uniform distribution | Angle parameter | Controls angle value of cosine function |
