RVEAa: Reference vector-guided evolution algorithm embedded with the reference vector regeneration strategy
Reference: Cheng R, Jin Y, Olhofer M, et al. A reference vector guided evolutionary algorithm for many-objective optimization[J]. IEEE Transactions on Evolutionary Computation, 2016, 20(5): 773-791.
RVEAa is an improved version of RVEA which can solve many-objective optimization problems (MaOPs) with irregular Pareto front.
Variables | Meaning |
---|---|
npop | Population size |
iter | Iteration number |
lb | Lower bound |
ub | Upper bound |
nobj | The dimension of objective space (default = 3) |
eta_c | Spread factor distribution index (default = 30) |
eta_m | Perturbance factor distribution index (default = 20) |
alpha | The parameter to control the change rate of APD (default = 2) |
fr | Reference vector adaption parameter (default = 0.1) |
nvar | The dimension of decision space |
pop | Population |
objs | Objectives |
V0 | Original reference vectors |
V | Reference vectors |
gamma | The smallest angle value of each reference vector to the others |
APD | Angle-penalized distance |
dom | Domination matrix |
pf | Pareto front |
if __name__ == '__main__':
main(105, 500, np.array([0] * 12), np.array([1] * 12))
It can be seen that RVEAa performs betters on MaOPs with irregular Pareto fronts than the original RVEA.