论文部分内容阅读
现有时变自回归(TVAR)模型参数谱估计容易导致谱峰漂移。针对该问题,提出一种基于组合目标函数和遗传算法的TVAR参数估计方法,并将之应用于飞行器结构响应序列的建模及谱估计。通过U-C算法获得TVAR模型参数的初始估计;依据现代谱估计理论结合连续函数极值存在的必要条件,推导模型参数的频域约束条件并构造组合目标函数;采用遗传算法对模型参数初始估计值进行优化。应用结果证明了该方法的有效性。
Existing time-varying autoregressive (TVAR) model parameter spectrum estimation easily lead to spectral peak drift. Aimed at this problem, a TVAR parameter estimation method based on combined objective function and genetic algorithm is proposed and applied to modeling and spectrum estimation of aircraft structural response sequence. The initial estimation of TVAR model parameters is obtained by UC algorithm. Based on the modern spectrum estimation theory and the necessary conditions of continuous function extremum, the frequency domain constraints of model parameters are deduced and the combined objective function is constructed. The initial estimation of model parameters is carried out by genetic algorithm optimization. The application results prove the effectiveness of this method.