クロススペクトル法による音源寄与率の測定
スポンサーリンク
概要
- 論文の詳細を見る
The measurement of noise source contributions is to determine the amount of acoustic power contributed to the field at a given observation point by each of several noise sources. K. W. Goff presented an approach to this measurement problem that, by the use of a correlation technique, yields the desired results without requiring any control of the sources involved. However, the practical application of this method is limited by the severe restricting conditions imposed. This paper proposes a new approach to the problem using a cross-spectral density function of the acoustic signals. This method of analysis in the frequency domain eliminates those restrictions inherent in the correlation technique. Referring to Fig. 1, x(t) and y(t) are the acoustic signals at the source in question and the observation point, respectively. G(f) is assumed to be the frequency transfer function between these two points. Not only pure delays dut any linear transfer characteristic can be assumed for G(f). The signal z(t) is the component of y(t) due to the source in question. The acoustic signals origination from the other sources are represented together by n(t). Although the signal z(t) is not directly observable, it is possible to estimate its power spectrum using the spectra of the other signals as shown by Eqs. (2), (3) and (4), in which Ф indicates the spectra of the signals designated by the respective subscript indices. The average power of z(t) is obtained by integrating the estimated power spectrum Ф_zz(f) over the frequency range of interest. Therefore, the power contribution desired is given by the ratio of this integrated value to that measured power spectrum Ф_yy(f) as in Eq. (5). Fig. 4 is an experimental result obtained with the arrangement shown in Fig. 3. The solid line in Fig. 4(b) is the power spectrum of the total acoustic signal at the observation point P and the dotted line represents the estimated power spectrum of the component due to the speaker A. The correlation technique fails to give the correct result in this case because the two peaks of the cross-correlation function corresponding to the direct and reflected acoustic signals overlap as shown in Fig. 4(a). However, the cross-spectral method is applied without any difficulty, yielding the measured contribution of 45. 3%. This agrees well with the directly measured value of 45. 0% obtained by silencing the speaker B. In order to verify the applicability of the cross-spectral method to highly periodic noises, an attempt was made to separate the exhaust noise of an automobile from the engine noise. A microphone for collecting the total acoustic signal was placed 5 meters from one side of the car and a second microphone, for picking up the sound at the source, was placed next to the exhaust outlet just out of the gas stream. In Fig. 5, the large peak of Ф_yy(f) in the frequency range below 50 Hz is the power of the wind noise produced at the structure of the microphone. The peaks at 65Hz and 130 Hz are mainly due to the exhaust noise, while the power of the engine noise, except for the fan noise component of 160 Hz, disperses widely to higher frequencies.
- 社団法人日本音響学会の論文
- 1973-10-01
著者
関連論文
- 空気ジェットを用いた姿勢制御系の適応性をもった最短時間制御
- 走行中の鉄道車輪径の自動測定(技術談話室)
- 音響的方法による航空機の位置標定
- 同期積分の手法による騒音伝搬測定
- 信号検出の手法を用いた騒音伝播測定
- ミニコンピュータによる航空機騒音の予測コンターの計算と作成
- 自己相関関数の差異に基づく雑音中の信号検出法とそのノック検出への応用
- ディジタル型相関器の試作
- ディジタルフィルタリングによる雑音中の音声の抽出
- クロススペクトル法による音源寄与率の測定
- 振動ピトー管 : 質量流量の新しい測定法
- クロススペクトル分析計
- 振動ピト-管--質量流量の新しい測定法
- パイロット信号による同期相関法
- 相関技術(III)
- 相関技術(II)
- 相関技術(I)
- 相関技術の応用
- 対数型ディジタル電圧計 : 測定技術