本(ben)發明涉(she)及(ji)一種(zhong)穩健的窄帶(dai)主動(dong)(dong)噪聲控(kong)制(zhi)系(xi)統(tong)及(ji)方法,屬于主動(dong)(dong)噪聲控(kong)制(zhi),特(te)別地,本(ben)發明涉(she)及(ji)一種(zhong)基于同步在線(xian)辨識的前饋型窄帶(dai)主動(dong)(dong)噪聲控(kong)制(zhi)系(xi)統(tong)及(ji)方法。。
背景技術:
1、主(zhu)動噪(zao)(zao)聲(sheng)(sheng)控(kong)制(zhi)技術(shu)(shu)(active?noise?control,anc)利用聲(sheng)(sheng)波的(de)(de)相消干涉(she)原理,具(ju)有良好的(de)(de)低頻噪(zao)(zao)聲(sheng)(sheng)抑制(zhi)性能。相對(dui)于傳統的(de)(de)基于隔音、聲(sheng)(sheng)屏障(zhang)、吸音材料等手段(duan)的(de)(de)被(bei)動噪(zao)(zao)聲(sheng)(sheng)控(kong)制(zhi)技術(shu)(shu),該主(zhu)動噪(zao)(zao)聲(sheng)(sheng)控(kong)制(zhi)技術(shu)(shu)具(ju)有體(ti)積小、成本低、易于安裝(zhuang)等優點,適用于控(kong)制(zhi)低頻諧波噪(zao)(zao)聲(sheng)(sheng)和(he)音頻范圍內(nei)的(de)(de)噪(zao)(zao)聲(sheng)(sheng),是(shi)對(dui)傳統的(de)(de)被(bei)動噪(zao)(zao)聲(sheng)(sheng)控(kong)制(zhi)技術(shu)(shu)的(de)(de)有利補(bu)充(s.m.kuo?andd.r.morgan,“active?noise?control:a?tutorial?review,”proc.ieee,vol.87,no.6,pp.943-973,jun.1999.)。
2、實(shi)際環境中存在大量的由(you)于旋轉部件或(huo)裝置產生(sheng)的周期性有害噪(zao)聲(sheng)(如(ru)風機(ji)噪(zao)聲(sheng)、切割機(ji)噪(zao)聲(sheng)、螺旋槳噪(zao)聲(sheng)等),其(qi)(qi)窄(zhai)(zhai)(zhai)帶分(fen)量占主(zhu)導成分(fen)。傳(chuan)(chuan)統(tong)(tong)的前(qian)饋(kui)(kui)型(xing)窄(zhai)(zhai)(zhai)帶anc系統(tong)(tong)采(cai)用非(fei)聲(sheng)學(xue)傳(chuan)(chuan)感(gan)器(如(ru)轉速(su)計等)獲取(qu)參(can)考(kao)(kao)頻率(lv)信(xin)號(hao)時,可(ke)能存在由(you)于非(fei)聲(sheng)學(xue)傳(chuan)(chuan)感(gan)器的老化、溫漂等引起的頻率(lv)失調(diao)問題(ti),將嚴重(zhong)影響控制(zhi)系統(tong)(tong)的降(jiang)噪(zao)性能。前(qian)饋(kui)(kui)型(xing)窄(zhai)(zhai)(zhai)帶anc系統(tong)(tong)若采(cai)用聲(sheng)學(xue)傳(chuan)(chuan)感(gan)器獲取(qu)參(can)考(kao)(kao)信(xin)號(hao),可(ke)有效避(bi)免上述非(fei)聲(sheng)學(xue)傳(chuan)(chuan)感(gan)器可(ke)能存在的頻率(lv)失調(diao)的問題(ti),但(dan)同(tong)時對參(can)考(kao)(kao)信(xin)號(hao)引入聲(sheng)反(fan)饋(kui)(kui)影響,進(jin)而制(zhi)約其(qi)(qi)降(jiang)噪(zao)性能。此外,實(shi)際工(gong)況下次級通道(dao)具有復雜時變性,研究相應(ying)的基于次級通道(dao)在線辨識方(fang)法(fa)的前(qian)饋(kui)(kui)型(xing)anc系統(tong)(tong)具有重(zhong)要(yao)意義。因(yin)此,設計高(gao)性能的同(tong)時含有聲(sheng)反(fan)饋(kui)(kui)和次級通道(dao)在線辨識的前(qian)饋(kui)(kui)型(xing)窄(zhai)(zhai)(zhai)帶anc系統(tong)(tong),具有重(zhong)要(yao)實(shi)際應(ying)用價值。
3、國內外學(xue)者圍繞含聲(sheng)反(fan)(fan)饋(kui)(kui)(kui)(kui)的(de)前饋(kui)(kui)(kui)(kui)型窄帶anc系統優化做(zuo)了大量的(de)研究工作。傳統的(de)聲(sheng)反(fan)(fan)饋(kui)(kui)(kui)(kui)通(tong)道辨識(shi)方法采(cai)用較大長度的(de)線性(xing)預(yu)測濾波(bo)器來降低參考噪(zao)聲(sheng)對聲(sheng)反(fan)(fan)饋(kui)(kui)(kui)(kui)通(tong)道在(zai)線辨識(shi)性(xing)能的(de)影響,且采(cai)用與聲(sheng)反(fan)(fan)饋(kui)(kui)(kui)(kui)通(tong)道估計和預(yu)測濾波(bo)器有(you)關的(de)信號進行輔助噪(zao)聲(sheng)的(de)幅值調整,具有(you)計算(suan)成本高(gao)、局(ju)部最優的(de)缺(que)點(s.ahmed,m.t.akhtar.gain?scheduling?ofauxiliary?noise?and?variable?step-size?for?online?acoustic?feedbackcancellation?in?narrowband?active?noise?control?systems[j].ieee?trans.audio,speech,and?lang.process.,2017,25(2):333-343)。
4、xiao等人設計了一(yi)(yi)種基于聲(sheng)反(fan)饋(kui)(kui)和次級通(tong)道(dao)(dao)同步在(zai)線(xian)辨(bian)識(shi)的(de)(de)(de)(de)(de)(de)主動噪(zao)(zao)聲(sheng)控制(zhi)系統(tong)(tong),可(ke)有(you)(you)效(xiao)克服聲(sheng)反(fan)饋(kui)(kui)的(de)(de)(de)(de)(de)(de)影響,同時可(ke)應(ying)對次級通(tong)道(dao)(dao)的(de)(de)(de)(de)(de)(de)時變性,降(jiang)(jiang)低(di)了殘(can)余(yu)噪(zao)(zao)聲(sheng)能量(liang)(t.bai,z.wang,y.xiao,y.ma,l.ma,and?k.khorasani.amulti-channel?narrowband?activenoise?control?system?with?simultaneous?online?secondary-and?feedback-pathmodeling[c].ieee?apccas,2019,pp.289-292)。但是,該(gai)系統(tong)(tong)仍(reng)存在(zai)以下問題:1)該(gai)系統(tong)(tong)直接采用(yong)與殘(can)余(yu)噪(zao)(zao)聲(sheng)有(you)(you)關的(de)(de)(de)(de)(de)(de)函數(shu)值來調整(zheng)輔助噪(zao)(zao)聲(sheng)幅值,進而調整(zheng)后的(de)(de)(de)(de)(de)(de)輔助噪(zao)(zao)聲(sheng)用(yong)作聲(sheng)反(fan)饋(kui)(kui)和次級通(tong)道(dao)(dao)同步在(zai)線(xian)辨(bian)識(shi)環(huan)節的(de)(de)(de)(de)(de)(de)參考輸(shu)(shu)(shu)入,導致輔助噪(zao)(zao)聲(sheng)對殘(can)余(yu)噪(zao)(zao)聲(sheng)仍(reng)有(you)(you)較大的(de)(de)(de)(de)(de)(de)貢獻(xian)量(liang),嚴重制(zhi)約系統(tong)(tong)整(zheng)體降(jiang)(jiang)噪(zao)(zao)性能;2)該(gai)系統(tong)(tong)直接采用(yong)殘(can)余(yu)噪(zao)(zao)聲(sheng),一(yi)(yi)方面其(qi)用(yong)作次級通(tong)道(dao)(dao)在(zai)線(xian)辨(bian)識(shi)環(huan)節的(de)(de)(de)(de)(de)(de)期望輸(shu)(shu)(shu)入,由于殘(can)余(yu)噪(zao)(zao)聲(sheng)中(zhong)含有(you)(you)窄帶殘(can)余(yu)噪(zao)(zao)聲(sheng)分量(liang),將影響次級通(tong)道(dao)(dao)在(zai)線(xian)辨(bian)識(shi)環(huan)節的(de)(de)(de)(de)(de)(de)期望輸(shu)(shu)(shu)入;另一(yi)(yi)方面其(qi)用(yong)作窄帶控制(zhi)器的(de)(de)(de)(de)(de)(de)誤(wu)(wu)差(cha)輸(shu)(shu)(shu)出(chu),由于殘(can)余(yu)噪(zao)(zao)聲(sheng)中(zhong)寬帶殘(can)余(yu)噪(zao)(zao)聲(sheng)分量(liang),將影響上述誤(wu)(wu)差(cha)輸(shu)(shu)(shu)出(chu);以上兩(liang)方面均會導致控制(zhi)器和次級通(tong)道(dao)(dao)在(zai)線(xian)辨(bian)識(shi)環(huan)節之(zhi)間的(de)(de)(de)(de)(de)(de)獨立性較差(cha),影響系統(tong)(tong)的(de)(de)(de)(de)(de)(de)收斂性能和降(jiang)(jiang)噪(zao)(zao)速度。
5、ma&xiao等(deng)人開發(fa)了(le)基于(yu)次(ci)級通道在線辨識技術(shu)的(de)前饋(kui)型寬窄(zhai)帶混合anc系(xi)統(tong),引入了(le)基于(yu)iir并(bing)型濾(lv)波(bo)器結(jie)構的(de)殘(can)(can)余誤(wu)差分(fen)離子系(xi)統(tong),可應(ying)對(dui)次(ci)級通道的(de)復雜時變性,改善了(le)控制器和(he)次(ci)級通道在線辨識環節之間的(de)獨立性,同時進(jin)一(yi)步降低了(le)殘(can)(can)余噪聲(y.ma,y.xiao,l.ma,and?k.khorasani.arobust?feedforward?hybrid?active?noise?controlsystem?with?online?secondary-path?modelling[j].iet?signal?process.,2023,17(1):e12183)。然而,該系(xi)統(tong)一(yi)方(fang)面未考(kao)慮聲反饋(kui)的(de)影(ying)響,另一(yi)方(fang)面其采用(yong)的(de)殘(can)(can)余誤(wu)差分(fen)離子系(xi)統(tong),其中的(de)iir并(bing)型濾(lv)波(bo)結(jie)構需要窄(zhai)帶目標噪聲的(de)頻率(lv)信(xin)息,且收(shou)斂(lian)半徑的(de)取值將(jiang)影(ying)響iir陷波(bo)器的(de)動態性能(neng)和(he)頻率(lv)追蹤(zong)性能(neng)。
技術實現思路
1、為了解決目前的(de)(de)前饋型窄(zhai)(zhai)帶(dai)anc系統(tong)(tong)存(cun)在著聲反饋、次級通道的(de)(de)復(fu)雜時變性(xing)等,嚴重制約前饋型窄(zhai)(zhai)帶(dai)anc系統(tong)(tong)的(de)(de)收斂性(xing)和(he)穩定性(xing),進而降低整體(ti)系統(tong)(tong)的(de)(de)窄(zhai)(zhai)帶(dai)噪聲抑制性(xing)能的(de)(de)問題,本發明(ming)提(ti)供(gong)了一種基(ji)于同步在線辨識的(de)(de)前饋型窄(zhai)(zhai)帶(dai)主(zhu)動噪聲控制系統(tong)(tong)及方(fang)法,技術方(fang)案如下所述(shu)。
2、本發明(ming)的(de)第(di)一個目的(de)在(zai)于提供(gong)(gong)一種基(ji)于同步在(zai)線(xian)辨識的(de)前饋(kui)(kui)型窄帶主動噪聲(sheng)(sheng)(sheng)(sheng)(sheng)控制(zhi)系(xi)(xi)統(tong)(tong),所述主動噪聲(sheng)(sheng)(sheng)(sheng)(sheng)控制(zhi)系(xi)(xi)統(tong)(tong)分別利用(yong)參考傳(chuan)聲(sheng)(sheng)(sheng)(sheng)(sheng)器(qi)(qi)(qi)采集(ji)參考信(xin)號、利用(yong)誤(wu)差(cha)傳(chuan)聲(sheng)(sheng)(sheng)(sheng)(sheng)器(qi)(qi)(qi)采集(ji)殘余噪聲(sheng)(sheng)(sheng)(sheng)(sheng)、利用(yong)次級(ji)(ji)(ji)(ji)(ji)(ji)(ji)揚聲(sheng)(sheng)(sheng)(sheng)(sheng)器(qi)(qi)(qi)提供(gong)(gong)次級(ji)(ji)(ji)(ji)(ji)(ji)(ji)聲(sheng)(sheng)(sheng)(sheng)(sheng)源;實(shi)際初級(ji)(ji)(ji)(ji)(ji)(ji)(ji)通道(dao)(dao)為參考信(xin)號傳(chuan)播到誤(wu)差(cha)傳(chuan)聲(sheng)(sheng)(sheng)(sheng)(sheng)器(qi)(qi)(qi)的(de)通道(dao)(dao)模型;實(shi)際聲(sheng)(sheng)(sheng)(sheng)(sheng)反饋(kui)(kui)通道(dao)(dao)為次級(ji)(ji)(ji)(ji)(ji)(ji)(ji)揚聲(sheng)(sheng)(sheng)(sheng)(sheng)器(qi)(qi)(qi)提供(gong)(gong)的(de)次級(ji)(ji)(ji)(ji)(ji)(ji)(ji)聲(sheng)(sheng)(sheng)(sheng)(sheng)源傳(chuan)播到參考傳(chuan)聲(sheng)(sheng)(sheng)(sheng)(sheng)器(qi)(qi)(qi)的(de)通道(dao)(dao)模型;實(shi)際次級(ji)(ji)(ji)(ji)(ji)(ji)(ji)通道(dao)(dao)為次級(ji)(ji)(ji)(ji)(ji)(ji)(ji)揚聲(sheng)(sheng)(sheng)(sheng)(sheng)器(qi)(qi)(qi)提供(gong)(gong)的(de)次級(ji)(ji)(ji)(ji)(ji)(ji)(ji)聲(sheng)(sheng)(sheng)(sheng)(sheng)源傳(chuan)播到誤(wu)差(cha)傳(chuan)聲(sheng)(sheng)(sheng)(sheng)(sheng)器(qi)(qi)(qi)的(de)通道(dao)(dao)模型;所述主動噪聲(sheng)(sheng)(sheng)(sheng)(sheng)控制(zhi)系(xi)(xi)統(tong)(tong)包括:第(di)一線(xian)性預測子(zi)(zi)系(xi)(xi)統(tong)(tong)1、第(di)二(er)線(xian)性預測子(zi)(zi)系(xi)(xi)統(tong)(tong)2、窄帶次級(ji)(ji)(ji)(ji)(ji)(ji)(ji)聲(sheng)(sheng)(sheng)(sheng)(sheng)源合(he)成子(zi)(zi)系(xi)(xi)統(tong)(tong)3、聲(sheng)(sheng)(sheng)(sheng)(sheng)反饋(kui)(kui)通道(dao)(dao)在(zai)線(xian)辨識子(zi)(zi)系(xi)(xi)統(tong)(tong)4和次級(ji)(ji)(ji)(ji)(ji)(ji)(ji)通道(dao)(dao)在(zai)線(xian)辨識子(zi)(zi)系(xi)(xi)統(tong)(tong)5;
3、所(suo)(suo)(suo)述(shu)(shu)(shu)第(di)一線(xian)性(xing)(xing)預(yu)測(ce)(ce)子(zi)系(xi)統(tong)(tong)(tong)(tong)(tong)1分別(bie)與(yu)所(suo)(suo)(suo)述(shu)(shu)(shu)窄(zhai)帶(dai)次(ci)(ci)級(ji)聲源合(he)成(cheng)子(zi)系(xi)統(tong)(tong)(tong)(tong)(tong)3、所(suo)(suo)(suo)述(shu)(shu)(shu)聲反(fan)(fan)饋通(tong)(tong)道(dao)(dao)在(zai)線(xian)辨(bian)(bian)識(shi)(shi)子(zi)系(xi)統(tong)(tong)(tong)(tong)(tong)4連(lian)(lian)接(jie)(jie);所(suo)(suo)(suo)述(shu)(shu)(shu)第(di)二線(xian)性(xing)(xing)預(yu)測(ce)(ce)子(zi)系(xi)統(tong)(tong)(tong)(tong)(tong)2分別(bie)與(yu)所(suo)(suo)(suo)述(shu)(shu)(shu)窄(zhai)帶(dai)次(ci)(ci)級(ji)聲源合(he)成(cheng)子(zi)系(xi)統(tong)(tong)(tong)(tong)(tong)3、所(suo)(suo)(suo)述(shu)(shu)(shu)次(ci)(ci)級(ji)通(tong)(tong)道(dao)(dao)在(zai)線(xian)辨(bian)(bian)識(shi)(shi)子(zi)系(xi)統(tong)(tong)(tong)(tong)(tong)5連(lian)(lian)接(jie)(jie);所(suo)(suo)(suo)述(shu)(shu)(shu)窄(zhai)帶(dai)次(ci)(ci)級(ji)聲源合(he)成(cheng)子(zi)系(xi)統(tong)(tong)(tong)(tong)(tong)3分別(bie)與(yu)所(suo)(suo)(suo)述(shu)(shu)(shu)第(di)一線(xian)性(xing)(xing)預(yu)測(ce)(ce)子(zi)系(xi)統(tong)(tong)(tong)(tong)(tong)1、所(suo)(suo)(suo)述(shu)(shu)(shu)第(di)二線(xian)性(xing)(xing)預(yu)測(ce)(ce)子(zi)系(xi)統(tong)(tong)(tong)(tong)(tong)2連(lian)(lian)接(jie)(jie);所(suo)(suo)(suo)述(shu)(shu)(shu)聲反(fan)(fan)饋通(tong)(tong)道(dao)(dao)在(zai)線(xian)辨(bian)(bian)識(shi)(shi)子(zi)系(xi)統(tong)(tong)(tong)(tong)(tong)4分別(bie)與(yu)所(suo)(suo)(suo)述(shu)(shu)(shu)第(di)一線(xian)性(xing)(xing)預(yu)測(ce)(ce)子(zi)系(xi)統(tong)(tong)(tong)(tong)(tong)1、所(suo)(suo)(suo)述(shu)(shu)(shu)次(ci)(ci)級(ji)通(tong)(tong)道(dao)(dao)在(zai)線(xian)辨(bian)(bian)識(shi)(shi)子(zi)系(xi)統(tong)(tong)(tong)(tong)(tong)5連(lian)(lian)接(jie)(jie);所(suo)(suo)(suo)述(shu)(shu)(shu)次(ci)(ci)級(ji)通(tong)(tong)道(dao)(dao)在(zai)線(xian)辨(bian)(bian)識(shi)(shi)子(zi)系(xi)統(tong)(tong)(tong)(tong)(tong)5分別(bie)與(yu)所(suo)(suo)(suo)述(shu)(shu)(shu)第(di)二線(xian)性(xing)(xing)預(yu)測(ce)(ce)子(zi)系(xi)統(tong)(tong)(tong)(tong)(tong)2、所(suo)(suo)(suo)述(shu)(shu)(shu)聲反(fan)(fan)饋通(tong)(tong)道(dao)(dao)在(zai)線(xian)辨(bian)(bian)識(shi)(shi)子(zi)系(xi)統(tong)(tong)(tong)(tong)(tong)4連(lian)(lian)接(jie)(jie);
4、所述第(di)一線性預測子(zi)系統(tong)1包括自(zi)適(shi)應線性預測濾波器(qi),用于合成所述窄(zhai)帶(dai)次(ci)級(ji)聲(sheng)(sheng)源合成子(zi)系統(tong)3的參考(kao)輸入(ru)和所述聲(sheng)(sheng)反(fan)饋通道在線辨識子(zi)系統(tong)4的誤差輸出;
5、所述(shu)(shu)第二線性(xing)預測子系統(tong)2包(bao)括自適應線性(xing)預測濾波器(qi),用(yong)(yong)于從殘(can)(can)余(yu)噪(zao)聲(sheng)(sheng)中(zhong)分離(li)出寬(kuan)帶(dai)殘(can)(can)余(yu)噪(zao)聲(sheng)(sheng)分量和窄帶(dai)殘(can)(can)余(yu)噪(zao)聲(sheng)(sheng)分量,把(ba)(ba)所述(shu)(shu)寬(kuan)帶(dai)殘(can)(can)余(yu)噪(zao)聲(sheng)(sheng)分量用(yong)(yong)作次級通(tong)道(dao)在線辨(bian)識(shi)模塊(kuai)51的(de)期(qi)望輸入,把(ba)(ba)所述(shu)(shu)窄帶(dai)殘(can)(can)余(yu)噪(zao)聲(sheng)(sheng)分量分別(bie)用(yong)(yong)作輔(fu)助(zhu)噪(zao)聲(sheng)(sheng)幅值(zhi)調整模塊(kuai)52的(de)輸入和帶(dai)一階延遲的(de)濾波-x最小均(jun)方算法模塊(kuai)32的(de)誤差輸出;
6、所(suo)(suo)述(shu)(shu)(shu)窄帶次級聲(sheng)源合成(cheng)子(zi)系統(tong)3包括控制器31和所(suo)(suo)述(shu)(shu)(shu)帶一(yi)階延遲的濾波-x最小(xiao)均方(fang)算法模塊(kuai)32,用于合成(cheng)窄帶次級聲(sheng)源,其與所(suo)(suo)述(shu)(shu)(shu)輔助噪(zao)聲(sheng)幅(fu)值調整模塊(kuai)52的輸出進行疊加,合成(cheng)次級聲(sheng)源;
7、所(suo)(suo)述聲(sheng)反(fan)(fan)(fan)饋通道(dao)在線辨識子系統4包括聲(sheng)反(fan)(fan)(fan)饋通道(dao)估計模(mo)(mo)型(xing)41和最小(xiao)均方(fang)算法(fa)模(mo)(mo)塊42,利用所(suo)(suo)述最小(xiao)均方(fang)算法(fa)模(mo)(mo)塊42以在線方(fang)式(shi)進行(xing)對實(shi)際(ji)聲(sheng)反(fan)(fan)(fan)饋通道(dao)進行(xing)建(jian)模(mo)(mo),獲得所(suo)(suo)述聲(sheng)反(fan)(fan)(fan)饋通道(dao)估計模(mo)(mo)型(xing)41,進而用于聲(sheng)反(fan)(fan)(fan)饋補償;
8、所(suo)述(shu)(shu)(shu)次(ci)(ci)級(ji)(ji)(ji)(ji)通道(dao)在線(xian)辨(bian)識子(zi)系(xi)統5包括所(suo)述(shu)(shu)(shu)次(ci)(ci)級(ji)(ji)(ji)(ji)通道(dao)在線(xian)辨(bian)識模(mo)塊(kuai)51和所(suo)述(shu)(shu)(shu)輔助噪聲(sheng)幅值(zhi)調整模(mo)塊(kuai)52;所(suo)述(shu)(shu)(shu)次(ci)(ci)級(ji)(ji)(ji)(ji)通道(dao)在線(xian)辨(bian)識子(zi)系(xi)統5對實際次(ci)(ci)級(ji)(ji)(ji)(ji)通道(dao)進行實時地在線(xian)估計(ji),得到(dao)所(suo)述(shu)(shu)(shu)次(ci)(ci)級(ji)(ji)(ji)(ji)通道(dao)在線(xian)辨(bian)識模(mo)塊(kuai)51中的(de)(de)次(ci)(ci)級(ji)(ji)(ji)(ji)通道(dao)估計(ji)模(mo)型,并將得到(dao)的(de)(de)次(ci)(ci)級(ji)(ji)(ji)(ji)通道(dao)估計(ji)模(mo)型用作所(suo)述(shu)(shu)(shu)窄帶次(ci)(ci)級(ji)(ji)(ji)(ji)聲(sheng)源合成(cheng)子(zi)系(xi)統3中的(de)(de)所(suo)述(shu)(shu)(shu)帶一(yi)階延遲的(de)(de)濾波-x最(zui)小均方算法模(mo)塊(kuai)32的(de)(de)濾波環節(jie)。
9、可選的(de)(de),所(suo)(suo)述第(di)一(yi)線性預測子系(xi)統(tong)1把參考信號估計(ji)用作輸入(ru)(ru),分(fen)離(li)出(chu)的(de)(de)窄帶(dai)分(fen)量xf(n)和寬(kuan)帶(dai)分(fen)量xw(n),并(bing)分(fen)別(bie)用作所(suo)(suo)述窄帶(dai)次級(ji)聲(sheng)源合成(cheng)子系(xi)統(tong)3的(de)(de)參考輸入(ru)(ru)和所(suo)(suo)述聲(sheng)反饋通(tong)道在(zai)線辨(bian)識子系(xi)統(tong)4的(de)(de)誤(wu)差輸出(chu),提升所(suo)(suo)述窄帶(dai)次級(ji)聲(sheng)源合成(cheng)子系(xi)統(tong)3和所(suo)(suo)述聲(sheng)反饋通(tong)道在(zai)線辨(bian)識子系(xi)統(tong)4之間的(de)(de)獨立性;
10、所述(shu)第(di)(di)一線性(xing)預(yu)測(ce)子系(xi)統(tong)1包括第(di)(di)一延(yan)遲環節(jie)11和第(di)(di)一線性(xing)預(yu)測(ce)濾(lv)波(bo)器12,所述(shu)第(di)(di)一延(yan)遲環節(jie)11和第(di)(di)一線性(xing)預(yu)測(ce)濾(lv)波(bo)器12級(ji)聯,所述(shu)第(di)(di)一延(yan)遲環節(jie)11的階數(shu)為(wei)(wei)d1;所述(shu)第(di)(di)一線性(xing)預(yu)測(ce)濾(lv)波(bo)器12為(wei)(wei)有限長沖激響應濾(lv)波(bo)器模型,其(qi)系(xi)數(shu)和長度分別(bie)為(wei)(wei)和l1,其(qi)系(xi)數(shu)利(li)用(yong)最小(xiao)均方算法進行(xing)更(geng)新,更(geng)新公(gong)式(shi)為(wei)(wei):
11、
12、其中(zhong),μ1為所述第一線性預測濾波器(12)的更新步長,取值為正值;n是時刻(ke),
13、n≥0;
14、所述第一線性預測子系統(tong)(1)合成得到的(de)窄帶(dai)分量和寬(kuan)帶(dai)分量分別(bie)為:
15、
16、可(ke)選的,所(suo)述第(di)二(er)線(xian)性預測子系統(tong)2用于從殘余噪(zao)聲e(n)中分離出所(suo)述寬(kuan)帶殘余噪(zao)聲分量ew(n)和(he)所(suo)述窄帶殘余噪(zao)聲分量ef(n);
17、所(suo)述第(di)(di)二(er)線性(xing)預(yu)(yu)測(ce)(ce)子系(xi)統2包括(kuo)第(di)(di)二(er)延遲環節21和第(di)(di)二(er)線性(xing)預(yu)(yu)測(ce)(ce)濾波(bo)器(qi)22,所(suo)述第(di)(di)二(er)延遲環節21和第(di)(di)二(er)線性(xing)預(yu)(yu)測(ce)(ce)濾波(bo)器(qi)22級聯,所(suo)述第(di)(di)二(er)延遲環節21的(de)階數(shu)(shu)為(wei)d2;所(suo)述第(di)(di)二(er)線性(xing)預(yu)(yu)測(ce)(ce)濾波(bo)器(qi)22為(wei)有限長(chang)沖激(ji)響應濾波(bo)器(qi)模型(xing),其系(xi)數(shu)(shu)和長(chang)度分別為(wei)和l2,其系(xi)數(shu)(shu)利用最小均方(fang)算法進行更(geng)新(xin),更(geng)新(xin)公式為(wei):
18、h2,j(n+1)=h2,j(n)+μ2ew(n)e(n-d2-j)
19、其中,μ2為所(suo)述第二線性預測濾波(bo)器22的更(geng)新步長(chang),取值為正值;
20、所(suo)述第二線性預測(ce)子系統2合成(cheng)得(de)到的所(suo)述寬帶殘余噪聲分量ew(n)和(he)所(suo)述窄帶殘余噪聲分量ef(n)分別為:
21、ew(n)=e(n)-ef(n)
22、
23、可選的(de),所(suo)述(shu)(shu)(shu)(shu)(shu)第(di)二線性預測子(zi)系(xi)統(tong)(tong)2,將所(suo)述(shu)(shu)(shu)(shu)(shu)寬帶(dai)(dai)殘(can)余噪聲(sheng)(sheng)分量(liang)ew(n)用作所(suo)述(shu)(shu)(shu)(shu)(shu)次級通道在(zai)線辨(bian)識(shi)模塊51的(de)期望輸入,將所(suo)述(shu)(shu)(shu)(shu)(shu)窄帶(dai)(dai)殘(can)余噪聲(sheng)(sheng)分量(liang)ef(n)用作所(suo)述(shu)(shu)(shu)(shu)(shu)窄帶(dai)(dai)次級聲(sheng)(sheng)源合成子(zi)系(xi)統(tong)(tong)3的(de)所(suo)述(shu)(shu)(shu)(shu)(shu)帶(dai)(dai)一階延遲的(de)濾波-x最(zui)小(xiao)均方(fang)算法模塊32的(de)誤差輸出,提升所(suo)述(shu)(shu)(shu)(shu)(shu)窄帶(dai)(dai)次級聲(sheng)(sheng)源合成子(zi)系(xi)統(tong)(tong)3和所(suo)述(shu)(shu)(shu)(shu)(shu)次級通道在(zai)線辨(bian)識(shi)子(zi)系(xi)統(tong)(tong)4之間的(de)獨(du)立性。
24、可選的(de),將所述窄(zhai)帶殘余噪聲分量ef(n)用作(zuo)所述輔助噪聲幅值(zhi)調整模塊52的(de)輸(shu)入,所述輔助噪聲幅值(zhi)調整模塊52輸(shu)出的(de)有(you)色噪聲v(n)為:
25、v(n)=v0(n)gs(n)
26、gs(n)=(1-λ)gs(n)+|ef(n-1)γ
27、其中,v0(n)為(wei)均值為(wei)零、方(fang)差(cha)為(wei)的(de)(de)高斯白噪聲;gs(n)為(wei)所述輔助噪聲幅值調整模(mo)塊52的(de)(de)增益調整因子;λ為(wei)遺忘因子,取(qu)小于1的(de)(de)正值;γ取(qu)1或2;
28、所(suo)述輔(fu)助噪(zao)聲(sheng)(sheng)幅值調整模塊52輸(shu)出(chu)的有色噪(zao)聲(sheng)(sheng)v(n)分別同步地輸(shu)入(ru)到所(suo)述次級(ji)通道(dao)在線(xian)辨識模塊51和所(suo)述聲(sheng)(sheng)反饋通道(dao)估計模型41中(zhong),實(shi)現次級(ji)通道(dao)和聲(sheng)(sheng)反饋通道(dao)同步在線(xian)辨識,同時(shi)降低(di)所(suo)述有色噪(zao)聲(sheng)(sheng)v(n)對殘余噪(zao)聲(sheng)(sheng)的貢獻(xian)量。
29、可選的,所(suo)述(shu)窄(zhai)帶次級聲源合成子系統3中,所(suo)述(shu)控制器31采用有限(xian)長(chang)沖激響應濾波器模型,其系數和長(chang)度分別為和l3;利用所(suo)述(shu)帶一(yi)階延遲的濾波-x最小(xiao)均(jun)方算法(fa)模塊32,對所(suo)述(shu)控制器31進行更新,即(ji):
30、
31、其(qi)中,μ3為所述控制(zhi)器31的更新步長,取(qu)值(zhi)為正值(zhi);為所述第一線(xian)性(xing)預測(ce)子(zi)系(xi)統1合成得到的窄(zhai)帶分(fen)量經所述次級通道(dao)在線(xian)辨識模(mo)塊51中次級通道(dao)估計模(mo)型后(hou)的輸出。
32、可選的,所(suo)述(shu)(shu)(shu)聲反饋(kui)(kui)通(tong)道(dao)(dao)在線辨識子系統4,利用(yong)所(suo)述(shu)(shu)(shu)最(zui)小均方算法模(mo)塊(kuai)42對所(suo)述(shu)(shu)(shu)聲反饋(kui)(kui)通(tong)道(dao)(dao)估(gu)計模(mo)型(xing)41進(jin)行(xing)更(geng)新(xin),所(suo)述(shu)(shu)(shu)聲反饋(kui)(kui)通(tong)道(dao)(dao)估(gu)計模(mo)型(xing)41為(wei)有限長(chang)沖激(ji)響應(ying)濾波(bo)器模(mo)型(xing),其系數和長(chang)度分別(bie)為(wei)和即:
33、
34、其中,μf為(wei)所述聲反饋通(tong)(tong)道估計(ji)模(mo)型(xing)41的更新步長,取(qu)值為(wei)正值;所述聲反饋通(tong)(tong)道估計(ji)模(mo)型(xing)41的輸出抵消實(shi)(shi)際聲學空間中聲反饋yf(n),實(shi)(shi)現(xian)聲反饋補償。
35、可選的,所述(shu)次(ci)級(ji)通(tong)道(dao)(dao)在線辨(bian)識模塊(kuai)51包(bao)括次(ci)級(ji)通(tong)道(dao)(dao)估計(ji)模型其采用有限長(chang)沖激響應濾波器模型,其系數和(he)長(chang)度(du)分別為(wei)和(he)所述(shu)次(ci)級(ji)通(tong)道(dao)(dao)在線辨(bian)識模塊(kuai)51以(yi)所述(shu)寬帶殘(can)余噪聲分量(liang)ew(n)為(wei)期(qi)望輸(shu)入、以(yi)所述(shu)輔助(zhu)噪聲幅(fu)值調整模塊(kuai)52的輸(shu)出v(n)為(wei)參(can)考輸(shu)入,利用最(zui)小均方算法實(shi)時地估計(ji)時變的次(ci)級(ji)通(tong)道(dao)(dao);
36、所述次級(ji)通(tong)道在線辨識模塊51的(de)更新公式為(wei):
37、
38、其中,μs為(wei)次級通道估(gu)計模(mo)(mo)型的更(geng)新步長(chang),取值為(wei)正值;es(n)為(wei)所述次級通道在線辨識模(mo)(mo)塊51的誤差輸(shu)出。
39、可選的(de),所(suo)述控(kong)制器(qi)31合(he)成窄帶次級(ji)(ji)聲源y0(n),其與所(suo)述輔(fu)助噪(zao)聲幅值調整模塊52的(de)輸出v(n)進行疊加,合(he)成次級(ji)(ji)聲源y(n),即:
40、y(n)=y0(n-1)+v(n)
41、在聲(sheng)學空間內(nei)y(n)送給次級(ji)揚聲(sheng)器并經過實(shi)際次級(ji)通道后,與目標噪聲(sheng)在局(ju)部區(qu)域內(nei)進行干(gan)涉相(xiang)消。
42、本發明的(de)第二(er)個目的(de)在(zai)于提供一種基于同步在(zai)線辨識的(de)前饋(kui)型(xing)(xing)窄(zhai)帶主動噪聲(sheng)控(kong)制方(fang)法(fa)(fa),其特征(zheng)在(zai)于,所(suo)述(shu)方(fang)法(fa)(fa)采用如上任一項所(suo)述(shu)的(de)前饋(kui)型(xing)(xing)窄(zhai)帶主動噪聲(sheng)控(kong)制系統,所(suo)述(shu)方(fang)法(fa)(fa)包括:
43、步驟一:系統初始化:
44、分(fen)別(bie)設(she)置(zhi)第一線(xian)性(xing)預測(ce)濾(lv)波(bo)器12、第二(er)(er)(er)線(xian)性(xing)預測(ce)濾(lv)波(bo)器22、控制器31、聲(sheng)反饋(kui)通(tong)(tong)(tong)道(dao)估(gu)計(ji)模(mo)型41、次(ci)級(ji)(ji)通(tong)(tong)(tong)道(dao)估(gu)計(ji)模(mo)型的(de)(de)長(chang)度及(ji)更(geng)新步長(chang);分(fen)別(bie)設(she)置(zhi)第一延遲環(huan)(huan)節11、第二(er)(er)(er)延遲環(huan)(huan)節21的(de)(de)階(jie)數(shu);設(she)置(zhi)輔(fu)助(zhu)噪(zao)聲(sheng)幅(fu)值(zhi)調整(zheng)模(mo)塊52的(de)(de)遺忘因子(zi);分(fen)別(bie)設(she)置(zhi)所(suo)述(shu)(shu)第一線(xian)性(xing)預測(ce)濾(lv)波(bo)器12、所(suo)述(shu)(shu)第二(er)(er)(er)線(xian)性(xing)預測(ce)濾(lv)波(bo)器42、所(suo)述(shu)(shu)控制器31、所(suo)述(shu)(shu)聲(sheng)反饋(kui)通(tong)(tong)(tong)道(dao)估(gu)計(ji)模(mo)型41、次(ci)級(ji)(ji)通(tong)(tong)(tong)道(dao)估(gu)計(ji)模(mo)型以及(ji)輔(fu)助(zhu)噪(zao)聲(sheng)幅(fu)值(zhi)調整(zheng)模(mo)塊52的(de)(de)增益調整(zheng)因子(zi)gs(n)的(de)(de)初(chu)始值(zhi)均為零(ling);設(she)置(zhi)輔(fu)助(zhu)噪(zao)聲(sheng)v0(n);
45、步驟二(er):在n時(shi)刻的系統狀態(tai):
46、在(zai)n時刻,窄帶(dai)次(ci)級聲(sheng)源合(he)成子(zi)系統3合(he)成窄帶(dai)次(ci)級聲(sheng)源,并與輔助(zhu)噪聲(sheng)幅值調整模塊52產生的輔助(zhu)噪聲(sheng)v(n)進行疊加,合(he)成得到次(ci)級聲(sheng)源y(n);
47、步驟三:在n時刻(ke),所(suo)述(shu)次(ci)(ci)級(ji)聲(sheng)(sheng)(sheng)源y(n)經實(shi)(shi)(shi)(shi)際(ji)聲(sheng)(sheng)(sheng)反(fan)饋(kui)通道(dao)產(chan)生(sheng)實(shi)(shi)(shi)(shi)際(ji)聲(sheng)(sheng)(sheng)反(fan)饋(kui)信號(hao)yf(n);實(shi)(shi)(shi)(shi)際(ji)參考(kao)(kao)信號(hao)xs(n)與(yu)所(suo)述(shu)實(shi)(shi)(shi)(shi)際(ji)聲(sheng)(sheng)(sheng)反(fan)饋(kui)信號(hao)yf(n)疊加得(de)(de)到參考(kao)(kao)傳聲(sheng)(sheng)(sheng)器采集的參考(kao)(kao)信號(hao)xr(n);所(suo)述(shu)參考(kao)(kao)信號(hao)xr(n)減(jian)去(qu)聲(sheng)(sheng)(sheng)反(fan)饋(kui)信號(hao)后得(de)(de)到的信號(hao)用作第一(yi)線(xian)性(xing)預測子系統1的輸入;所(suo)述(shu)第一(yi)線(xian)性(xing)預測子系統1分別合成窄帶分量和寬帶分量;與(yu)所(suo)述(shu)輔助噪(zao)(zao)聲(sheng)(sheng)(sheng)幅值(zhi)調整(zheng)模塊52產(chan)生(sheng)的輔助噪(zao)(zao)聲(sheng)(sheng)(sheng)v(n)用作次(ci)(ci)級(ji)通道(dao)在線(xian)辨(bian)識模塊51的參考(kao)(kao)輸入;所(suo)述(shu)次(ci)(ci)級(ji)聲(sheng)(sheng)(sheng)源y(n)經實(shi)(shi)(shi)(shi)際(ji)次(ci)(ci)級(ji)通道(dao)產(chan)生(sheng)反(fan)噪(zao)(zao)聲(sheng)(sheng)(sheng)yp(n);目標噪(zao)(zao)聲(sheng)(sheng)(sheng)p(n)減(jian)去(qu)所(suo)述(shu)反(fan)噪(zao)(zao)聲(sheng)(sheng)(sheng)yp(n)得(de)(de)到殘余噪(zao)(zao)聲(sheng)(sheng)(sheng)e(n);
48、步驟四:在(zai)n時(shi)刻(ke),所(suo)述殘(can)余(yu)(yu)噪(zao)聲(sheng)e(n)經過(guo)第(di)二線性預測子系統(tong)2分離出寬帶(dai)殘(can)余(yu)(yu)噪(zao)聲(sheng)分量(liang)(liang)(liang)ew(n)和(he)窄(zhai)帶(dai)殘(can)余(yu)(yu)噪(zao)聲(sheng)分量(liang)(liang)(liang)ef(n);同時(shi),所(suo)述寬帶(dai)殘(can)余(yu)(yu)噪(zao)聲(sheng)分量(liang)(liang)(liang)ew(n)用作(zuo)所(suo)述次(ci)級通道(dao)在(zai)線辨識模(mo)塊(kuai)51的(de)期望輸入(ru)(ru);同時(shi)所(suo)述窄(zhai)帶(dai)殘(can)余(yu)(yu)噪(zao)聲(sheng)分量(liang)(liang)(liang)ef(n)分別用作(zuo)所(suo)述窄(zhai)帶(dai)次(ci)級聲(sheng)源合(he)成子系統(tong)3的(de)帶(dai)一階延(yan)遲的(de)濾(lv)波-x最小均方算法模(mo)塊(kuai)32的(de)誤差輸出和(he)所(suo)述輔助(zhu)噪(zao)聲(sheng)幅(fu)值調整模(mo)塊(kuai)52的(de)輸入(ru)(ru);
49、步驟五(wu):系(xi)統(tong)狀(zhuang)態更新:
50、分別更(geng)新(xin)所(suo)(suo)述(shu)第一(yi)線(xian)性預(yu)測(ce)子系(xi)統(tong)1和所(suo)(suo)述(shu)第二(er)線(xian)性預(yu)測(ce)子系(xi)統(tong)2中(zhong)(zhong)線(xian)性預(yu)測(ce)濾波(bo)器的系(xi)數(shu);同時,利(li)用(yong)(yong)所(suo)(suo)述(shu)帶一(yi)階(jie)延遲的濾波(bo)-x最(zui)小(xiao)均(jun)方(fang)(fang)算(suan)法模(mo)塊32,更(geng)新(xin)所(suo)(suo)述(shu)窄帶次(ci)級(ji)聲(sheng)源(yuan)合成子系(xi)統(tong)3中(zhong)(zhong)所(suo)(suo)述(shu)控制(zhi)器31的系(xi)數(shu);同時,利(li)用(yong)(yong)最(zui)小(xiao)均(jun)方(fang)(fang)算(suan)法更(geng)新(xin)聲(sheng)反饋通道在線(xian)辨識(shi)子系(xi)統(tong)4中(zhong)(zhong)所(suo)(suo)述(shu)聲(sheng)反饋通道估(gu)計模(mo)型(xing)41的系(xi)數(shu);同時,利(li)用(yong)(yong)最(zui)小(xiao)均(jun)方(fang)(fang)算(suan)法更(geng)新(xin)次(ci)級(ji)通道在線(xian)辨識(shi)子系(xi)統(tong)5中(zhong)(zhong)所(suo)(suo)述(shu)次(ci)級(ji)通道估(gu)計模(mo)型(xing)51的系(xi)數(shu);
51、步(bu)(bu)驟(zou)(zou)六:返回到(dao)步(bu)(bu)驟(zou)(zou)二,重復上述步(bu)(bu)驟(zou)(zou)二到(dao)步(bu)(bu)驟(zou)(zou)五,直至系(xi)統收斂(lian)并(bing)達(da)到(dao)穩態(tai)。
52、本發明有益效果是(shi):
53、一(yi)、本(ben)(ben)發(fa)明無(wu)需安裝(zhuang)非聲學傳感器,降低(di)了(le)對按照(zhao)空間的要求(qiu)和系(xi)統硬件(jian)成本(ben)(ben),同(tong)時還(huan)避免了(le)頻率失調(diao)。
54、二、本(ben)發明利用第(di)一(yi)線(xian)性(xing)預測子(zi)系(xi)統1實現參(can)考信號(hao)中的(de)(de)寬帶分(fen)量(liang)和(he)窄帶分(fen)量(liang)的(de)(de)分(fen)離,分(fen)離的(de)(de)出寬帶分(fen)量(liang)用作聲反饋(kui)通(tong)道(dao)(dao)在線(xian)辨識(shi)子(zi)系(xi)統的(de)(de)誤差輸出,同時分(fen)離出的(de)(de)窄帶分(fen)量(liang)為窄帶次(ci)級(ji)聲源合成(cheng)子(zi)系(xi)統提(ti)供高質量(liang)輸入,提(ti)升了(le)窄帶次(ci)級(ji)聲源合成(cheng)子(zi)系(xi)統3和(he)聲反饋(kui)通(tong)道(dao)(dao)在線(xian)辨識(shi)子(zi)系(xi)統4之間的(de)(de)獨立性(xing),提(ti)升了(le)整體系(xi)統的(de)(de)動(dong)態(tai)性(xing)能。
55、三、本發明(ming)利用(yong)第二(er)(er)線性預測子(zi)系(xi)(xi)(xi)統2分離(li)出(chu)(chu)的(de)(de)(de)寬帶殘余噪聲分量作為次級(ji)通道在線辨識模(mo)塊(kuai)51的(de)(de)(de)期望輸入,同(tong)時利用(yong)第二(er)(er)線性預測子(zi)系(xi)(xi)(xi)統2分離(li)出(chu)(chu)的(de)(de)(de)窄帶分量,用(yong)作帶一階延(yan)遲的(de)(de)(de)濾波-x最小均(jun)方算法(fa)模(mo)塊(kuai)32的(de)(de)(de)誤(wu)差輸出(chu)(chu),提升(sheng)了窄帶次級(ji)聲源(yuan)合(he)成子(zi)系(xi)(xi)(xi)統3和(he)次級(ji)通道在線辨識子(zi)系(xi)(xi)(xi)統4之間的(de)(de)(de)獨(du)立性,提升(sheng)了整體(ti)系(xi)(xi)(xi)統的(de)(de)(de)收斂性和(he)應對強非(fei)平穩噪聲的(de)(de)(de)性能。
56、四、本發明利用二線性預測子(zi)系統(tong)2分離出的窄帶殘(can)余噪聲分量,用作輔(fu)助噪聲幅(fu)值調整(zheng)模塊52的輸入(ru),產生(sheng)的有色噪聲v(n)用作聲反饋(kui)(kui)通道在線辨(bian)識子(zi)系統(tong)4的參(can)考輸入(ru),有效提升聲反饋(kui)(kui)通道在線辨(bian)識子(zi)系統(tong)4的性能,繼而實(shi)現聲反饋(kui)(kui)補償。
57、五(wu)、本(ben)發明利用二(er)線性預測(ce)子(zi)系統2分(fen)離出的窄帶(dai)殘余(yu)噪(zao)(zao)聲(sheng)分(fen)量(liang),用作輔(fu)助噪(zao)(zao)聲(sheng)幅值(zhi)調(diao)整(zheng)(zheng)模(mo)塊52的輸(shu)入,產(chan)生的有色噪(zao)(zao)聲(sheng)v(n)為輔(fu)助噪(zao)(zao)聲(sheng),進而用作次級(ji)通道在(zai)線辨識模(mo)塊51的參(can)考輸(shu)入,有效降低引入的輔(fu)助噪(zao)(zao)聲(sheng)v(n)對殘余(yu)噪(zao)(zao)聲(sheng)的貢(gong)獻(xian)量(liang),使系統整(zheng)(zheng)體降噪(zao)(zao)性能趨(qu)于(yu)理(li)想(xiang)水平。
58、六、本發(fa)明(ming)采用(yong)聲反(fan)饋通(tong)道和次級通(tong)道同步在線辨(bian)識技(ji)術,可(ke)同時(shi)應對(dui)實際工況下聲反(fan)饋通(tong)道和次級通(tong)道的時(shi)變性,有(you)利于控制系統的實際應用(yong)。