本文選題：參數(shù)辨識(shí)　切入點(diǎn)：期望最大化　出處：《西北工業(yè)大學(xué)學(xué)報(bào)》2016年04期

【摘要】：針對(duì)初值和噪聲統(tǒng)計(jì)特性未知情形下的飛行器系統(tǒng)辨識(shí)的問(wèn)題,提出了基于期望最大化(expectation maximization,EM)和求容積卡爾曼平滑器(cubature Kalman smoother,CKS)的辨識(shí)算法。該算法用期望最大化算法對(duì)初值和噪聲的統(tǒng)計(jì)特性進(jìn)行估計(jì);用求容積卡爾曼平滑器估計(jì)狀態(tài)向量和未知參數(shù)。在期望最大化算法的求期望步驟中,所求的期望值通過(guò)求容積規(guī)則獲得,用較少的采樣點(diǎn)保證了估計(jì)精度;在期望最大化算法的最大化步驟中,未知量的最優(yōu)值以解析解形式給出,減小了計(jì)算量。仿真結(jié)果說(shuō)明,該算法在飛行器氣動(dòng)參數(shù)辨識(shí)問(wèn)題中,能給出較好的辨識(shí)結(jié)果。與其他方法的對(duì)比驗(yàn)證說(shuō)明新算法具有辨識(shí)精度高、收斂速度快等優(yōu)點(diǎn)。
[Abstract]:Aircraft system identification problem for statistical characteristics of initial and unknown noise case, is proposed based on the expectation maximization (expectation maximization EM) and Calman (cubature Kalman smoothers volume smoother, CKS) identification algorithm. The algorithm uses expectation maximization algorithm to the initial statistical characteristics and noise estimation; for the volume of Calman smoother estimation of state vector and unknown parameters. The expectation maximization algorithm for the expectation step, the expected value obtained by volume, with fewer sampling points to ensure the estimation accuracy; in the maximization step the expectation maximization algorithm, the analytic solution of the optimal value of unknown form that reduces the amount of computation. The simulation results show that the algorithm in aerodynamic parameter identification problem, can give a good identification results. Compared with other methods to prove that the new algorithm The method has the advantages of high recognition precision and fast convergence speed.

【作者單位】： 西北工業(yè)大學(xué)航空學(xué)院;中國(guó)航空工業(yè)第一飛機(jī)設(shè)計(jì)研究院;陜西省實(shí)驗(yàn)飛機(jī)設(shè)計(jì)與試驗(yàn)技術(shù)工程試驗(yàn)室;
【基金】：國(guó)家自然科學(xué)基金(11472222、61473227) 陜西省自然科學(xué)基金(2015JM6304) 航空科學(xué)基金(20151353018) 航天技術(shù)支撐基金(2014-HT-XGD)資助
【分類號(hào)】：V211

【相似文獻(xiàn)】

相關(guān)期刊論文 前1條

1 盧伯英,陳國(guó)清,陶漢銘;卡爾曼平滑在動(dòng)態(tài)推力測(cè)量中的應(yīng)用[J];北京航空航天大學(xué)學(xué)報(bào);1990年01期

，

本文編號(hào)：1681904