楊延超 李英順
摘 要
為解決武器觀瞄系統(tǒng)受多種不確定因素的影響難以測(cè)得大量數(shù)據(jù),文中提出一種新的狀態(tài)評(píng)估方法。使用局部聚類算法對(duì)已采集到的數(shù)據(jù)進(jìn)行聚類,得到各個(gè)等級(jí)聚類中心及其分類,再通過(guò)小子樣統(tǒng)計(jì)方法將小樣本轉(zhuǎn)換成大樣本,解決了試驗(yàn)樣本隨機(jī)性和樣本不足性對(duì)評(píng)估模型的影響。實(shí)例分析表明,經(jīng)該方法建立的模型得到的結(jié)論與基于先驗(yàn)知識(shí)的判斷一致,驗(yàn)證了所提方法的有效性。
關(guān)鍵詞
觀瞄系統(tǒng);狀態(tài)評(píng)估;監(jiān)督信息;密度聚類;Bootstrap小子樣統(tǒng)計(jì)
中圖分類號(hào): E923.1 ? ? ? ? ? ? 文獻(xiàn)標(biāo)識(shí)碼: A
DOI:10.19694/j.cnki.issn2095-2457.2020.09.066
Abstract
In order to solve the problem that the weapon sighting system is difficult to measure a large amount of data due to various uncertain factors,a new state assessment method is proposed.Use the local clustering algorithm to cluster the collected data to obtain the cluster centers of various levels and their classification.Then,the small sample is converted into a large sample by the small sample statistical method,which solves the impact of randomness and insufficient samples on the evaluation model.The example analysis shows that the conclusions obtained by the model established by this method are consistent with the judgment based on prior knowledge,which verifies the effectiveness of the proposed method.
Key Words
Viewing system;Status assessment;Density clustering;Bootstrap small sample statistics;State Assessment
0 引言
觀瞄系統(tǒng)是一個(gè)邏輯十分復(fù)雜的系統(tǒng),不光有瞄準(zhǔn)鏡控制盒的電氣系統(tǒng)還有瞄準(zhǔn)鏡精密的機(jī)械系統(tǒng),在作戰(zhàn)過(guò)程中極易發(fā)生故障,從而在很大程度上影響武器的戰(zhàn)斗力。
本文使用半監(jiān)督局部密度聚類算法(Semi-Supervised Local Density Clustering)[1-2],該算法降低算法聚類過(guò)程中搜索類的盲目性,引導(dǎo)算法快速實(shí)現(xiàn)數(shù)據(jù)聚類;同時(shí),簇心點(diǎn)的選擇方式也改進(jìn)為由機(jī)器自動(dòng)識(shí)別完成,使樣本的聚類過(guò)程具有客觀性,避免了由于對(duì)簇心點(diǎn)選擇的錯(cuò)誤而降低了算法的準(zhǔn)確性??紤]到武器實(shí)驗(yàn)屬于消耗性實(shí)驗(yàn),試驗(yàn)成本較高,大量試驗(yàn)品不能重復(fù)使用,因此本文將聚類算法與統(tǒng)計(jì)方法相結(jié)合,充分運(yùn)用Bootstrap小子樣自助統(tǒng)計(jì)方法,利用小子樣能夠?qū)⑿颖巨D(zhuǎn)換成大樣本的特點(diǎn),彌補(bǔ)試驗(yàn)樣本的隨機(jī)性和樣本不足性的缺點(diǎn),增強(qiáng)瞄準(zhǔn)鏡控制盒狀態(tài)評(píng)估模型的穩(wěn)定性。
1 半監(jiān)督局部密度聚類算法
1.1 局部密度聚類算法基本思想
4 結(jié)語(yǔ)
武器觀瞄系統(tǒng)的狀態(tài)評(píng)估是一項(xiàng)復(fù)雜細(xì)致的工作,本文中對(duì)局部密度聚類算法進(jìn)行改進(jìn),而后通過(guò)與Bootstrap小子樣自助統(tǒng)計(jì)方法結(jié)合建立了基于半監(jiān)督局部密度聚類算法的瞄準(zhǔn)鏡控制盒狀態(tài)評(píng)估模型。該模型有效地解決了樣本隨機(jī)性和樣本不足性對(duì)狀態(tài)評(píng)估結(jié)果的影響,并且不需要通過(guò)大量試驗(yàn)去獲取實(shí)驗(yàn)數(shù)據(jù),節(jié)省了高額成本產(chǎn)生了巨大的經(jīng)濟(jì)效益;通過(guò)實(shí)車狀態(tài)下的信號(hào)檢測(cè),能夠較為全面地反映實(shí)際技術(shù)狀況,既保證了系統(tǒng)狀態(tài)評(píng)估的結(jié)果也提高了效率,體現(xiàn)了人工智能狀態(tài)評(píng)估在軍事中的重要作用。
參考文獻(xiàn)
[1]周宇飛.基于偶對(duì)約束的半監(jiān)督模糊聚類算法研究[D].河北大學(xué),2014.
[2]李軼然,張春娜.基于成對(duì)約束的主動(dòng)半監(jiān)督聚類算法[J].計(jì)算機(jī)工程與設(shè)計(jì),2013,34(08):2897-2902.
[3]葉佐.局部密度聚類算法研究[D].浙江工業(yè)大學(xué),2017.
[4]羅偉明,吳帆,黃業(yè)廣,吳杰康,覃煒梅,龔杰,金尚婷.基于局部密度聚類算法的變壓器故障狀態(tài)評(píng)估[J].廣東電力,2018,31(08):81-90.
[5]Xuanxia Yao,Jiafei Wang,Mengyu Shen,Huafeng Kong,Huansheng Ning.An improved clustering algorithm and its application in IoT data analysis[J].Computer Networks,2019,159.
[6]Ahmed Fahim.A Clustering Algorithm based on Local Density of Points[J]. International Journal of Modern Education and Computer Science (IJMECS),2017,9(12).
[7]汶小妮.攻擊型無(wú)人機(jī)小子樣攻擊精度評(píng)估方法研究[D].西北工業(yè)大學(xué),2005.