基于Bayesian-MCMC估计的隐身飞机RCS模型优化

对隐身飞机的雷达散射截面(RCS)统计建模时,传统方法通过直接计算RCS样本的统计特征估计模型参数,可能会产生较大的拟合误差。本文提出采用贝叶斯-蒙特卡罗(Bayesian-MCMC )方法提高起伏模型的参数估计精度,从而减小模型的拟合误差。首先将卡方分布模型和对数正态分布模型进行贝叶斯推导,得到其特征参数的后验估计表达式。然后采用MCMC算法构造后验分布的马尔可夫链,从而计算特征参数的估计值。最后通过比较2种方法的拟合曲线及其误差可知,本文方法适用于2种起伏模型,模型参数的估计误差比收敛误差门限值低1~2个数量级,2种分布模型的拟合精度均提高50%以上。 

Optimization on stealth aircraft RCS models using Bayesian-MCMC estimation

When statistically modeling a stealth aircraft, conventional methods estimate characteristic parameters via calculating the statistics of radar cross section (RCS) directly, which may lead to relatively large fitting errors. Therefore,we introduce the Bayesian-Markov chain Monte Carlo (Bayesian-MCMC) method to improve the parameter accuracy so as to reduce the fitting errors. The posterior parameter estimators of the Chi-square and lognormal models are derived in the Bayesian framework. Then the MCMC sampling algorithm is adopted to calculate the parameter estimates by constructing Markov chains. Numerical results show that the estimation errors of the proposed method are 1-2 orders of magnitude lower than the error convergence threshold. Besides, the proposed method is suitable for both target fluctuation models and improves the curve fitting accuracy by more than 50%.