本文關(guān)鍵詞：正交迭代泛函網(wǎng)絡(luò)在中短期鐘差預(yù)報(bào)中的應(yīng)用　出處：《科技導(dǎo)報(bào)》2014年27期 　論文類型：期刊論文

  更多相關(guān)文章： 泛函網(wǎng)絡(luò) 鐘差預(yù)報(bào) 迭代運(yùn)算 正交序列

【摘要】：在衛(wèi)星鐘源無(wú)法與地面鐘源進(jìn)行實(shí)時(shí)比對(duì)的時(shí)段中,準(zhǔn)確預(yù)報(bào)衛(wèi)星鐘差對(duì)于維持衛(wèi)星的穩(wěn)定運(yùn)行具有重要意義。針對(duì)衛(wèi)星鐘差的中短期預(yù)報(bào)問(wèn)題,選擇多項(xiàng)式模型對(duì)鐘差進(jìn)行建模分析,設(shè)計(jì)了一種基于滑動(dòng)窗模型的正交迭代泛函網(wǎng)絡(luò)算法。利用泛函網(wǎng)絡(luò)的非線性學(xué)習(xí)能力對(duì)鐘差預(yù)報(bào)模型進(jìn)行擬合分析,采用正交函數(shù)作為泛函網(wǎng)絡(luò)的基函數(shù)簇,并引入滑動(dòng)窗思想來(lái)更新輸入層元素進(jìn)行迭代訓(xùn)練,獲得較小的預(yù)報(bào)誤差。分析表明,預(yù)報(bào)時(shí)間小于12 h時(shí),預(yù)報(bào)誤差為0.2~0.5 ns,預(yù)報(bào)精度與IGU P精度相當(dāng);當(dāng)預(yù)報(bào)時(shí)間為24 h時(shí),預(yù)報(bào)誤差總體在1 ns,預(yù)報(bào)精度略次于IGU P精度;當(dāng)預(yù)報(bào)時(shí)間為1個(gè)衛(wèi)星周時(shí),最大誤差達(dá)130 ns,難以滿足衛(wèi)星運(yùn)行對(duì)鐘源的要求。研究表明:該算法適合于短期衛(wèi)星鐘差預(yù)報(bào),不適合中長(zhǎng)期鐘差預(yù)報(bào)。
[Abstract]:In this paper , an orthogonal iterative functional network algorithm based on sliding window model is designed based on the nonlinear learning ability of the functional network . The results show that the prediction error is 0.2 锝，

本文編號(hào)：1364270