CAJ | 학술논문

为了更精确地检测出混沌背景下的微弱目标信号，提高预测效果，文中提出了一种混沌混合粒子群优化RBF神经网络(CHPSO-RBFNN)算法。本算法主要采用了基于群体自适应变异和个体退火操作的混沌粒子群优化RBF神经网络，利用群体自适应变异以及个体退火操作优化混沌粒子群，有效地提高了粒子群算法的全局收敛性，优化了RBF神经网络的结构和参数。把该算法用于预测混沌时间序列、检测混沌背景下微弱目标信号，实验结果表明本算法有良好的非线性预测能力，可以有效地检测出混沌背景下的微弱目标信号。
위료경정학지검측출혼돈배경하적미약목표신호，제고예측효과，문중제출료일충혼돈혼합입자군우화RBF신경망락(CHPSO-RBFNN)산법。본산법주요채용료기우군체자괄응변이화개체퇴화조작적혼돈입자군우화RBF신경망락，이용군체자괄응변이이급개체퇴화조작우화혼돈입자군，유효지제고료입자군산법적전국수렴성，우화료RBF신경망락적결구화삼수。파해산법용우예측혼돈시간서렬、검측혼돈배경하미약목표신호，실험결과표명본산법유량호적비선성예측능력，가이유효지검측출혼돈배경하적미약목표신호。
In order to detect the weak target signal accurately in the chaos background, and improve forecast result, a novel algorithm based on RBF Neural Network ( RBFNN) with Chaotic Hybrid Particle Swarm Optimization ( CHPSO) is presented. In this algorithm, the RBF neural network is optimized by chaotic particle swarm optimization with adaptive population mutation and individual annealing operation. In order to improve the global convergence ability of PSO,the colony adaptive mutation and individual annealing operation are used to adjust and optimize PSO. Then the parameters and structures of RBFNN are optimized. This novel algorithm is applied to predict chaotic time sequence and detect weak target signal in the chaos background. Simulation results show that the algorithm has preferable nonlinear prediction ability and can detect weak target signal effectively.