JP4201995B2 - Optical fiber strain measurement method and apparatus - Google Patents

Description

【０００８】
但し、ｎ_ｆは、光ファイバの屈折率、λは、光ファイバに入射する光パルスの波長である。また、ｖ_Ａは、光ファイバ中の音速であり、光ファイバのヤング率とポアソン比と密度とから与えられる。入射光パルスの周波数とブリルアン散乱光の中心周波数との周波数差Δν_Ｂと光ファイバの歪みεとの関係は、
【０００９】
【数２】

【００１０】
で与えられることが、例えば、倉島、堀口、立田、「ブリルアン散乱を応用した分布型光ファイバセンサ」電子情報通信学会論文誌 C-II,Vol.J.74-C-II,No.5,pp.467-476,1991 に記載されている。ここで、Ｋεは、光ファイバの歪みとΔν_Ｂとを関係付ける定数である。式（１）（２）より、入射光パルスの周波数とブリルアン散乱光の中心周波数との周波数差Δν_Ｂを計測することにより、光ファイバの歪みを測定することができる。入射光パルスの周波数は既知であるので、ブリルアン散乱光のパワースペクトルを推定し、その中心周波数を得ることにより、式（１）（２）より光ファイバの歪みを得ることができる。
【００１１】
単一の歪みしか存在しない領域におけるブリルアン散乱光のパワースペクトルの推定手法に関する従来技術について説明する。光ファイバ中で発生するブリルアン散乱光のパワースペクトルは良く知られている様に、ローレンツ型パワースペクトルで良く近似することができる。例えば、C.N.Pannell, J.Dhliwayo and D.J.Webb,“How to estimate the accuracy of a Brillouin distributed temperature sensor, Proc.OFS97（IEEE）”，PP.524-527,New York,1997（以下、文献Pannellという。）または T.Kurashima, T.Horiguchi, H.Izumita, S.Furukawa and Y.Koyamada,“Brillouin Optical-Fiber Time Domain Reflectometry”，IEICE Trans.Comm.,Vol.E76-B,No.4,pp.82-390,1993 に近似方法が記載されている。
【００１２】
ｉ番目の観測周波数ν_ｉにおける観測雑音を含まないブリルアン散乱光のパワーｇ_Ｔｉは、
【００１３】
【数３】

【００１４】
と数式を用いて書くことができる。但し、ν_Ｂは、ブリルアン散乱光の中心周波数、ωは、全半値幅（ＦＷＨＭ）、ｈは、式（１）における最大パワーであり、ピーク値と呼ばれる。通常ブリルアン散乱光のパワー観測時には、観測時に発生する観測雑音が存在する。この観測雑音は、加法的であると仮定すると、観測されるｉ番目の観測周波数における観測値ｇ_ｉは、ｇ_Ｔｉとこの時の観測雑音Δ_ｉを用いて次式の様に表される。
【００１５】
【数４】

【００１６】
ブリルアン散乱光のパワースペクトル分布の推定とは、観測値ｇ_ｉよりｇ_Ｔｉあるいは式（３）のν_Ｂ，ω，ｈを求める問題である。しかし、観測雑音は、一般的には未知量であるため、ｇ_Ｔｉあるいはν_Ｂ，ω，ｈを直接求めることができない。そこで、通常の最小二乗法では、次式に示す二乗誤差Ｊ_０を最小とするパラメータν_Ｂ０，ω_０，ｈ_０を求め、このパラメータより式（３）を用いて最も確からしいと思われるパワースペクトル分布を求めることになる。
【００１７】
【数５】

【００１８】
但し、ｇ_０ｉは、式（３）にν_Ｂ＝ν_Ｂ０，ω＝ω_０，ｈ＝ｈ_０を代入して求めたブリルアン散乱光の推定値であり、Ｎは、観測値の個数である。パラメータν_Ｂ０，ω_０，ｈ_０を解析的に求めるためには、
【００１９】
【数６】

【００２０】
を満たすν_Ｂ０，ω_０，ｈ_０を求めればよい。
【００２１】
しかし、式（１）に示す様に、ローレンツ型パワースペクトルは、推定するパラメータに対して非線形であり、式（６）の解を解析的に求めるのは困難である。この問題を解決するために、下記のＪ_１を最小化するパラメータを求めることにより、式（５）のＪ_０を最小化するパラメータを近似的に求める手法が、文献Pannellに提案されている。
【００２２】
【数７】

【００２３】
式（７）に示す様に、Ｊ_１を最小化する場合には線形化可能であり、通常の最小二乗法により、解であるパラメータν_Ｂ０，ω_０，ｈ_０を求めることができる。上述したように、単一の歪みしか存在しない場合、ブリルアン散乱光のローレンツ型パワースペクトルを推定することが可能である。
【００２４】
図２は、２つの歪みが混在する測定区間を示した図である。歪み１，２は、想定している２つの歪みであり、測定区間Ｂは、この２つの歪みが混在する領域である。測定区間Ａ，Ｃは、測定区間Ｂの両側に存在すると想定される単一の歪みしか存在しない領域である。上述したように、光ファイバ歪み測定器において測定されるブリルアン散乱光は、測定区間内に長さ方向に積分された信号である。
【００２５】
従って、測定区間Ａ，Ｃでは、単一の歪みに対応する単一のローレンツ型パワースペクトルにより近似し得るが、２つの歪みが混在する測定区間Ｂで散乱されるブリルアン散乱光のパワースペクトルは、それぞれの歪みに対応する単一のローレンツ型パワースペクトルの線形和となり、次式の様に近似される。
【００２６】
【数８】

【００２７】
但し、ｙ_Ｃｉは、二つの歪みが混在する測定区間Ｂにおけるｉ番目の観測周波数でのブリルアン散乱光のパワー、ｒ_Ａｉ，ｒ_Ｂｉは、ピーク値を１に規格化した歪み１，２に対応する単一のローレンツ型パワースペクトル、Ｈ_１，Ｈ_２はこれらのパワースペクトルの各々のピーク値である。ｒ_Ａｉ，ｒ_Ｂｉは、１に規格化した単一のローレンツ型パワースペクトルであることから、次式の様に表すことができる。
【００２８】
【数９】

【００２９】
但し、ν_ＡＢ０，ν_ＢＢ０は、歪み１，２の対応するローレンツ型パワースペクトルの中心周波数、ω_Ａ０，ω_Ｂ０は、歪み１，２に対応するローレンツ型パワースペクトルの全半値幅である。
【００３０】
【発明が解決しようとする課題】
しかしながら、２つの歪みが混在する測定区間Ｂにおけるブリルアン散乱光の観測値に、式（７）で示した単一の歪みしか存在しない領域での線形化手法を適用したとすると、下記の二乗誤差Ｊ_２を最小化するようにローレンツ型パワースペクトルのパラメータを決定することになる。
【００３１】
【数１０】

【００３２】
但し、ｇ_Ｃｉは、測定区間Ｂでの観測値である。しかし、ｙ_Ｃｉは、式（８）〜（１０）に示したように、歪み１，２に対応するローレンツ型パワースペクトルの線形和であるから、
【００３３】
【数１１】

【００３４】
となり、線形化することができない。よって、従来の線形化による最小二乗法は、２つの歪みが混在する測定区間に適用できないという欠点がある。また、上述したように、従来の線形化による最小二乗法は適用してはならないが、もし、誤って２つの歪みが混在する測定区間の観測値に適用された場合、強引に単一のローレンツ型パワースペクトルとして推定するため、誤った中心周波数を推定するのみならず、結果として誤った歪みを推定するという問題があった。
【００３５】
なお、最急降下法を用いて２つのローレンツ型パワースペクトルのパラメータν_ＡＢ０，ω_Ａ０，Ｈ_１，ν_ＢＢ０，ω_Ｂ０，Ｈ_２を求めることも考えられるが、ローレンツ型パワースペクトルは、式（８）〜（１１）に示すように非線形であり、最急降下法の一般的な欠点であるローカルミニマムの問題、繰り返し回数の増大の問題、初期値の決定問題等を免れず、ローレンツ型パワースペクトルのパラメータおよび最終的に得られる歪みの値の信頼性が低いという問題もあった。
【００３６】
本発明は、このような問題に鑑みてなされたもので、その目的とするところは、２つの歪みが混在する光ファイバの測定区間で散乱されるブリルアン散乱光の各々の歪みに対応するローレンツ型パワースペクトルのピーク値を推定することにより、光ファイバに発生している長さ方向の歪み分布を求める光ファイバひずみ計測方法およびその装置を提供することにある。
【００３７】
【課題を解決するための手段】
本発明は、このような目的を達成するために、請求項１に記載の発明は、センサとして用いる光ファイバに光パルスを入射し、該光パルスにより発生した後方散乱光の一つであるブリルアン散乱光を、前記光パルスの時間幅と前記光ファイバ中の光の速度との積で表される測定区間において長さ方向に積分されたブルリアン散乱光として計測し、該計測されたブルリアン散乱光のパワースペクトルのピーク値と、該ピーク値を与える中心周波数と全半値幅とを決定して、前記光ファイバに発生している長さ方向の歪み分布を求める光ファイバ歪み計測方法において、前記測定区間のうち単一の第１の歪みのみ存在する第１の測定区間と、前記第１の歪みとは異なる単一の第２の歪みのみ存在する第２の測定区間と、前記第１の歪みと前記第２の歪みとが混在する第３の測定区間とにおいて散乱されたブリルアン散乱光のパワースペクトルをそれぞれ計測する計測ステップと、該計測ステップで計測された前記第１および第２の測定区間の前記パワースペクトルから、前記第１および第２の測定区間におけるピーク値を１に規格化したローレンツ型パワースペクトルをそれぞれ推定する第１推定ステップと、該第１推定ステップで推定された前記規格化したローレンツ型パワースペクトルと、前記計測ステップで計測された前記第３の測定区間の前記パワースペクトルとを用いて、前記第３の測定区間における前記第１および第２の歪みに対応するローレンツ型パワースペクトルのピーク値を推定する第２推定ステップとを備えることを特徴とする。
より具体的には、前記第２推定ステップは、前記第１推定ステップで推定された前記第１および第２の測定区間の前記規格化したローレンツ型パワースペクトル（ r _Ai , r _Bi ）の線形和（ y _Ci ）と、前記計測ステップで計測された前記第３の測定区間の前記パワースペクトル（ g _Ci ）との二乗誤差 (J ₃ ) が最小となる、前記第３の測定区間における前記第１および第２の歪みに対応するローレンツ型パワースペクトルのピーク値（ H ₁ , H ₂ ）を、最小二乗法を用いて推定する。
【００３８】
この方法によれば、２つの歪みが混在する光ファイバの測定区間で散乱されるブリルアン散乱光の各々の歪みに対応するローレンツ型パワースペクトルのピーク値を推定することにより、光ファイバに発生している長さ方向の歪み分布を求めることができる。
【００３９】
請求項３に記載の発明は、センサとして用いる光ファイバに光パルスを入射するための光パルス生成部と、前記光パルスにより発生した後方散乱光の一つであるブリルアン散乱光を、前記光パルスの時間幅と前記光ファイバ中の光の速度との積で表される測定区間において長さ方向に積分されたブルリアン散乱光として計測する計測部とを備え、前記計測されたブルリアン散乱光のパワースペクトルのピーク値と、該ピーク値を与える中心周波数と全半値幅とを決定して、前記光ファイバに発生している長さ方向の歪み分布を求める光ファイバ歪み計測装置において、前記測定区間のうち単一の第１の歪みのみ存在する第１の測定区間と、前記第１の歪みとは異なる単一の第２の歪みのみ存在する第２の測定区間と、前記第１の歪みと前記第２の歪みとが混在する第３の測定区間とにおいて散乱されたブリルアン散乱光のパワースペクトルをそれぞれ計測する計測手段と、該計測手段で計測された前記第１および第２の測定区間の前記パワースペクトルから、前記第１および第２の測定区間におけるピーク値を１に規格化したローレンツ型パワースペクトルをそれぞれ推定する第１推定手段と、該第１推定手段で推定された前記規格化したローレンツ型パワースペクトルと、前記計測手段で計測された前記第３の測定区間の前記パワースペクトルとを用いて、前記第３の測定区間における前記第１および第２の歪みに対応するローレンツ型パワースペクトルのピーク値を推定する第２推定手段とを備えたことを特徴とする。
より具体的には、前記第２推定手段は、前記第１推定手段で推定された前記第１および第２の測定区間の前記規格化したローレンツ型パワースペクトル（ r _Ai , r _Bi ）の線形和（ y _Ci ）と、前記計測手段で計測された前記第３の測定区間の前記パワースペクトル（ g _Ci ）との二乗誤差 (J ₃ ) が最小となる、前記第３の測定区間における前記第１および第２の歪みに対応するローレンツ型パワースペクトルのピーク値（ H ₁ , H ₂ ）を、最小二乗法を用いて推定する。
【００４０】
請求項５に記載の発明は、センサとして用いる光ファイバに光パルスを入射するための光パルス生成部と、前記光パルスにより発生した後方散乱光の一つであるブリルアン散乱光を、前記光パルスの時間幅と前記光ファイバ中の光の速度との積で表される測定区間において長さ方向に積分されたブルリアン散乱光として計測する計測部とを備え、前記計測されたブルリアン散乱光のパワースペクトルのピーク値と、該ピーク値を与える中心周波数と全半値幅とを決定して、前記光ファイバに発生している長さ方向の歪み分布を求める光ファイバ歪み計測装置を制御するプログラムを記録した記録媒体であって、前記測定区間のうち単一の第１の歪みのみ存在する第１の測定区間と、前記第１の歪みとは異なる単一の第２の歪みのみ存在する第２の測定区間と、前記第１の歪みと前記第２の歪みとが混在する第３の測定区間とにおいて散乱されたブリルアン散乱光のパワースペクトルをそれぞれ計測する計測ステップと、該計測ステップで計測された前記第１および第２の測定区間の前記パワースペクトルから、前記第１および第２の測定区間におけるピーク値を１に規格化したローレンツ型パワースペクトルをそれぞれ推定する第１推定ステップと、該第１推定ステップで推定された前記規格化したローレンツ型パワースペクトルと、前記計測ステップで計測された前記第３の測定区間の前記パワースペクトルとを用いて、前記第３の測定区間における前記第１および第２の歪みに対応するローレンツ型パワースペクトルのピーク値を推定する第２推定ステップとをコンピュータに実行させるためのプログラムを記録したコンピュータ読み取り可能な記録媒体であることを特徴とする。
より具体的には、前記第２推定ステップは、前記第１推定ステップで推定された前記第１および第２の測定区間の前記規格化したローレンツ型パワースペクトル（ r _Ai , r _Bi ）の線形和（ y _Ci ）と、前記計測ステップで計測された前記第３の測定区間の前記パワースペクトル（ g _Ci ）との二乗誤差 (J ₃ ) が最小となる、前記第３の測定区間における前記第１および第２の歪みに対応するローレンツ型パワースペクトルのピーク値（ H ₁ , H ₂ ）を、最小二乗法を用いて推定する。
【００４１】
請求項７に記載の発明は、センサとして用いる光ファイバに光パルスを入射するための光パルス生成部と、前記光パルスにより発生した後方散乱光の一つであるブリルアン散乱光を、前記光パルスの時間幅と前記光ファイバ中の光の速度との積で表される測定区間において長さ方向に積分されたブルリアン散乱光として計測する計測部とを備え、前記計測されたブルリアン散乱光のパワースペクトルのピーク値と、該ピーク値を与える中心周波数と全半値幅とを決定して、前記光ファイバに発生している長さ方向の歪み分布を求める光ファイバ歪み計測装置を制御するプログラムであって、前記測定区間のうち単一の第１の歪みのみ存在する第１の測定区間と、前記第１の歪みとは異なる単一の第２の歪みのみ存在する第２の測定区間と、前記第１の歪みと前記第２の歪みとが混在する第３の測定区間とにおいて散乱されたブリルアン散乱光のパワースペクトルをそれぞれ計測する計測ステップと、該計測ステップで計測された前記第１および第２の測定区間の前記パワースペクトルから、前記第１および第２の測定区間におけるピーク値を１に規格化したローレンツ型パワースペクトルをそれぞれ推定する第１推定ステップと、該第１推定ステップで推定された前記規格化したローレンツ型パワースペクトルと、前記計測ステップで計測された前記第３の測定区間の前記パワースペクトルとを用いて、前記第３の測定区間における前記第１および第２の歪みに対応するローレンツ型パワースペクトルのピーク値を推定する第２推定ステップとをコンピュータに実行させるためのプログラムであることを特徴とする。
より具体的には、前記第２推定ステップは、前記第１推定ステップで推定された前記第１および第２の測定区間の前記規格化したローレンツ型パワースペクトル（ r _Ai , r _Bi ）の線形和（ y _Ci ）と、前記計測ステップで計測された前記第３の測定区間の前記パワースペクトル（ g _Ci ）との二乗誤差 (J ₃ ) が最小となる、前記第３の測定区間における前記第１および第２の歪みに対応するローレンツ型パワースペクトルのピーク値（ H ₁ , H ₂ ）を、最小二乗法を用いて推定する。
【００４２】
【発明の実施の形態】
以下、図面を参照しながら本発明の実施形態について詳細に説明する。
２つの歪みが混在する測定区間での各々の歪みに対応するローレンツ型パワースペクトルの推定方法は、２つの歪みが混在する測定区間の両側に存在すると考えられる単一の歪みしか存在しない測定区間（例えば、図２に示した測定区間Ａと測定区間Ｃ。）でのブリルアン散乱光のパワースペクトルを利用することにより、２つの歪みが混在する測定区間でのパワースペクトルのパラメータν_ＡＢ０，ω_Ａ０，Ｈ_１，ν_ＢＢ０，ω_Ｂ０，Ｈ_２を求める。
【００４３】
図３は、本発明の一実施形態にかかるローレンツ型パワースペクトルの推定方法を示したフローチャートである。ステップＳ１では、２つの歪みが混在する測定区間の両側にある単一の歪みしか存在しない測定区間の観測値を用いてブリルアン散乱光のローレンツ型パワースペクトルを推定する。推定方法は、例えば、山田、成瀬、「重み付き最小二乗法の繰り返しによるＢＯＴＤＲ波形の推定」信学技報、OFT2000-30,pp.55-60,2000 に記載された従来技術を用いることができる。ステップＳ２では、ステップＳ１の推定結果であるローレンツ型パワースペクトルより、ピーク値を１に規格化したローレンツ型パワースペクトルを決定する。
【００４４】
次に、ステップＳ３について説明する。測定区間Ａ，Ｂ，Ｃについて歪み以外の条件は同一だとすると、ステップＳ２の結果であるピーク値を１に規格化したローレンツ型パワースペクトルは、式（８）におけるｒ_Ａｉ，ｒ_Ｂｉと同一である。従って、求める２つの歪みが混在する測定区間のパラメータのうちν_ＡＢ０，ω_Ａ０，ν_ＢＢ０，ω_Ｂ０は、ステップＳ２で求められていることになる。残るピーク値Ｈ_１，Ｈ_２は、最小二乗法を用いて次式の二乗誤差Ｊ_３を最小とする値をＨ_１，Ｈ_２として推定する。
【００４５】
【数１２】

【００４６】
式（１３）を最小とするＨ_１，Ｈ_２は、次式を満足する解として与えられる。
【００４７】
【数１３】

【００４８】
このようにして、２つの歪みが混在する測定区間でのパワースペクトルのパラメータν_ＡＢ０，ω_Ａ０，Ｈ_１，ν_ＢＢ０，ω_Ｂ０，Ｈ_２を全て得ることができ、２つの歪みが存在する測定区間での各々の歪みに対応するローレンツ型パワースペクトルを推定することができる。
【００４９】
本発明にかかる２つ歪みが混在する測定区間での各々の歪みに対応するローレンツ型パワースペクトルの推定方法が、実現可能でありかつ有効であることを、数値シミュレーションを用いて説明する。
【００５０】
シミュレーションに用いるピーク値を１に規格化した単一のローレンツ型パワースペクトルｒ_Ａｉ（ｉ＝１〜１００），ｒ_Ｂｉ（ｉ＝１〜１００）を計算機にて用意する。但し、ν_ＡＢ０＝１０．５ＧＨｚ，ω_Ａ０＝８１ＭＨｚ，ν_ＢＢ０＝１０．４ＧＨｚ，ω_Ｂ０＝８２ＭＨｚを用いている。このシミュレーションに用いたｒ_Ａｉ，ｒ_Ｂｉに、１／５０の正規分布で仮定した観測雑音を加算し、シミュレーションに用いた単一の歪みだけの測定区間での観測値を作成する。
【００５１】
図４は、シミュレーションに用いた単一の歪みだけの測定区間でのブリルアン散乱光の観測値を示した図である。但し、実線はｒ_Ａｉより作成した観測値、破線はｒ_Ｂｉより作成した観測値である。縦軸は、ある値で規格化されていることを想定し、相対値で示している。計算機でもとめたものとは別に、ｒ_Ａｉ，ｒ_Ｂｉを基に式（８）に従いｙ_Ｃｉを作成する。このｙ_Ｃｉの値に、１／５０の正規分布で仮定した観測雑音を加算して、シミュレーションに用いた２つの歪みが混在する測定区間での観測値を作成する。
【００５２】
図５は、シミュレーションに用いた２つの歪みが混在する測定区間での観測値を示した図である。ここで、Ｈ_１＝0.７，Ｈ_２＝0.３を用いている。これら計算機により作成した観測値に、図３に示した本発明にかかる推定方法を適用して得られた、２つの歪みが混在する測定区間でのパワースペクトルのパラメータν_ＡＢ０，ω_Ａ０，Ｈ_１，ν_ＢＢ０，ω_Ｂ０，Ｈ_２と、シミュレーションに用いた真の値とを比較して表１に示す。
【００５３】
【表１】

【００５４】
表１に示したように、本発明によりＨ_１，Ｈ_２を含む２つの歪みが混在する測定区間でのパワースペクトルのパラメータが推定できることが分かる。
【００５５】
本実施形態によれば、２つの歪みが混在する測定区間での各々の歪みに対応するローレンツ型パワースペクトルの推定方法が実現可能であり、有効であることが分かる。
【００５６】
上述したように、２つ歪みが混在する測定区間での各々の歪みに対応するローレンツ型パワースペクトルの推定方法を用いることにより、２つ歪みが混在する測定区間での各々の歪みに対応するローレンツ型パワースペクトルのパラメータを推定することが可能となる。ここで、２つの歪みが混在する測定区間のブリルアン散乱光のエネルギーは、それぞれに対応するローレンツ型パワースペクトルより次式で表される。
【００５７】
【数１４】

【００５８】
但し、ｓ_１は、２つの歪みが混在する測定区間での歪み１に起因するブリルアン散乱光のエネルギー、ｓ_２は、歪み２に起因するブリルアン散乱光のエネルギー、ｒ_Ａ，ｒ_Ｂは離散化される前のｒ_Ａｉ，ｒ_Ｂｉである。
【００５９】
歪み１、歪み２に起因するブリルアン散乱光の損失に差が無いとし、各ブリルアン散乱光のエネルギーは、２つの歪みが混在する測定区間内に存在する各歪みの長さに起因するとすると、入射光パルスのエネルギーは一定であるから、式（１５）（１６）のｓ_１とｓ_２との和は一定となる。従って、ω_Ａ０，Ｈ_１，ω_Ｂ０，Ｈ_２を求め、ｓ_１とｓ_２との比を計算することにより、２つの歪みが混在する測定区間内に存在する歪み１と歪み２の長さの比を求めることができる。
【００６０】
【発明の効果】
以上説明したように、本発明によれば、２つの歪みが混在する測定区間での各々の歪みに対応するローレンツ型パワースペクトルのパラメータを推定することができ、この結果を用いて光ファイバに発生している長さ方向の歪み分布を求めることが可能となる。
【図面の簡単な説明】
【図１】従来の光ファイバひずみ計測器を示すブロック図である。
【図２】２つの歪みが混在する測定区間を示した図である。
【図３】本発明の一実施形態にかかるローレンツ型パワースペクトルの推定方法を示したフローチャートである。
【図４】シミュレーションに用いた単一の歪みだけの測定区間でのブリルアン散乱光の観測値を示した図である。
【図５】シミュレーションに用いた２つの歪みが混在する測定区間での観測値を示した図である。
【符号の説明】
１０ 光ファイバひずみ計測器
１１ 光源
１２ 光周波数変換器
１３ 光パルス変換器
１４ コヒーレント光受信器
１５ 信号処理部[0001]
BACKGROUND OF THE INVENTION
The present invention relates to an optical fiber strain measurement method and apparatus, and more particularly, estimates a Lorentz-type power spectrum corresponding to each strain of Brillouin scattered light scattered in a measurement section of an optical fiber in which two strains are mixed. The present invention relates to an optical fiber strain measurement method and an apparatus therefor.
[0002]
[Prior art]
In the strain measurement method using an optical fiber as a sensor, the strain of the optical fiber can be obtained by measuring the frequency shift amount of Brillouin scattered light, which is one of the backscattered light generated in the optical fiber. An optical fiber strain measuring device using such properties is disclosed in, for example, Japanese Patent Laid-Open No. 10-90121. According to the publication, when an optical fiber strain measuring apparatus enters a measurement optical pulse into a sensing optical fiber, backscattered light is generated by receiving Rayleigh scattering or Brillouin scattering in the sensing optical fiber. The optical fiber strain measurement device combines the backscattered light with the reference light, performs coherent detection, and then generates a measurement waveform having characteristics corresponding to the strain generated in the sensing optical fiber based on the detected optical power spectrum. Output.
[0003]
FIG. 1 is a block diagram showing a conventional optical fiber strain measuring instrument. An optical fiber strain measuring instrument 10 is connected to an optical fiber for strain sensing, and a light source 11 that generates signal light and reference light, an optical frequency converter 12 that converts the optical frequency of the signal light, and pulse the signal light. The optical pulse converter 13, the coherent optical receiver 14 for detecting the Brillouin scattered light, and the signal processing unit 15 for outputting a measurement waveform based on the detected optical power.
[0004]
With such a configuration, the continuous light emitted from the light source 11 is branched into signal light and reference light. The optical frequency of the signal light is shifted to a high frequency side by about 11 GHz by the optical frequency converter 12 and converted into an optical pulse by the optical pulse converter 13. The optical fiber strain measuring instrument 10 makes this optical pulse for measurement enter the optical fiber for sensing, and receives the Brillouin scattered light generated thereby. Since Brillouin scattered light is weak, a coherent optical receiver 14 capable of highly sensitive detection is used. The coherent light receiver 14 uses Brillouin scattered light, which is a received signal, and reference light branched from a light source. When light having a wavelength of 1.55 μm is incident on a normal communication optical fiber, the Brillouin scattered light is shifted to the lower frequency side by about 11 GHz than the incident light. Therefore, the optical frequency converter 12 shifts in advance to the high frequency side by a frequency approximately equal to 11 GHz in advance. The coherent light receiver 14 can receive coherent light in which the frequency difference between the Brillouin scattered light and the reference light is reduced. Based on the detected optical power, the signal processing unit 15 outputs a measurement waveform having characteristics corresponding to the strain generated in the sensing optical fiber.
[0005]
Thus, since the optical fiber strain measuring instrument 10 can obtain a continuous signal in the length direction of the optical fiber, it can know the distribution of Brillouin scattered light in the length direction of the optical fiber. That is, the length corresponding to the delay time from the incidence of the light pulse to the measurement is the position where the measured Brillouin scattered light is scattered. However, the incident light pulse has a finite time width. Actually, the incident light pulse is long within a length (hereinafter referred to as a measurement section) represented by the product of the incident light pulse time width and the speed of light in the optical fiber. Brillouin scattered light integrated in the vertical direction is measured.
[0006]
In the measurement section at a certain distance of the optical fiber, the frequency difference Δν between the frequency of the incident light pulse and the center frequency of the Brillouin scattered light_BIs given by:
[0007]
[Expression 1]

[0008]
However, n_fIs the refractive index of the optical fiber, and λ is the wavelength of the optical pulse incident on the optical fiber. And v_AIs the speed of sound in the optical fiber, and is given by the Young's modulus, Poisson's ratio, and density of the optical fiber. Frequency difference Δν between the frequency of the incident light pulse and the center frequency of the Brillouin scattered light_BAnd the strain ε of the optical fiber is
[0009]
[Expression 2]

[0010]
For example, Kurashima, Horiguchi, Tachida, “Distributed optical fiber sensor using Brillouin scattering”, IEICE Transactions C-II, Vol. J. 74-C-II, No. 5, pp.467-476,1991. Here, Kε is the strain of the optical fiber and Δν_BIs a constant that relates From equations (1) and (2), the frequency difference Δν between the frequency of the incident light pulse and the center frequency of the Brillouin scattered light._BBy measuring this, the strain of the optical fiber can be measured. Since the frequency of the incident light pulse is known, the distortion of the optical fiber can be obtained from the equations (1) and (2) by estimating the power spectrum of the Brillouin scattered light and obtaining its center frequency.
[0011]
The prior art relating to a method for estimating the power spectrum of Brillouin scattered light in a region where only a single strain exists will be described. As is well known, the power spectrum of Brillouin scattered light generated in an optical fiber can be well approximated by a Lorentz power spectrum. For example, CNPannell, J. Dhliwayo and DJ Webb, “How to estimate the accuracy of a Brillouin distributed temperature sensor, Proc. OFS97 (IEEE)”, PP.524-527, New York, 1997 (hereinafter referred to as document Pannell. ) Or T.Kurashima, T.Horiguchi, H.Izumita, S.Furukawa and Y.Koyamada, “Brillouin Optical-Fiber Time Domain Reflectometry”, IEICE Trans.Comm., Vol.E76-B, No.4, pp. 82-390,1993 describes the approximation method.
[0012]
i-th observation frequency ν_iPower of Brillouin scattered light without observation noise_TiIs
[0013]
[Equation 3]

[0014]
And using mathematical formulas. Where ν_BIs the center frequency of the Brillouin scattered light, ω is the full width at half maximum (FWHM), h is the maximum power in equation (1), and is called the peak value. Usually, when observing the power of Brillouin scattered light, there is an observation noise generated during the observation. Assuming that this observation noise is additive, the observed value g at the i-th observed frequency observed_iIs g_TiAnd observation noise Δ at this time_iIs expressed as follows.
[0015]
[Expression 4]

[0016]
Estimating the power spectrum distribution of Brillouin scattered light is the observed value g_iFrom g_TiOr ν in equation (3)_B, Ω, h. However, since the observation noise is generally an unknown quantity, g_TiOr ν_B, Ω, h cannot be obtained directly. Therefore, in the normal least square method, the square error J₀Parameter ν that minimizes_B0, Ω₀, H₀From this parameter, the most probable power spectrum distribution is obtained using equation (3).
[0017]
[Equation 5]

[0018]
Where g_0iIs ν in equation (3)_B= Ν_B0, Ω = ω₀, H = h₀Is an estimated value of Brillouin scattered light obtained by substituting, and N is the number of observed values. Parameter ν_B0, Ω₀, H₀To find it analytically,
[0019]
[Formula 6]

[0020]
Satisfy ν_B0, Ω₀, H₀You can ask for.
[0021]
However, as shown in Expression (1), the Lorentz power spectrum is nonlinear with respect to the parameter to be estimated, and it is difficult to analytically obtain the solution of Expression (6). To solve this problem, the following J₁By obtaining a parameter that minimizes₀A method for approximately obtaining a parameter for minimizing the above is proposed in the document Pannell.
[0022]
[Expression 7]

[0023]
As shown in equation (7), J₁Can be linearized, and the parameter ν, which is the solution, can be obtained by ordinary least squares method._B0, Ω₀, H₀Can be requested. As described above, when only a single strain exists, it is possible to estimate the Lorentz power spectrum of the Brillouin scattered light.
[0024]
FIG. 2 is a diagram showing a measurement section in which two distortions are mixed. The distortions 1 and 2 are two assumed distortions, and the measurement section B is an area where the two distortions are mixed. The measurement sections A and C are areas in which only a single distortion assumed to exist on both sides of the measurement section B exists. As described above, the Brillouin scattered light measured by the optical fiber strain measuring instrument is a signal integrated in the length direction in the measurement section.
[0025]
Therefore, in the measurement sections A and C, the power spectrum of the Brillouin scattered light scattered in the measurement section B in which two distortions are mixed can be approximated by a single Lorentz-type power spectrum corresponding to a single distortion. It becomes a linear sum of a single Lorentzian power spectrum corresponding to each distortion, and is approximated by the following equation.
[0026]
[Equation 8]

[0027]
However, y_CiIs the power of the Brillouin scattered light at the i-th observation frequency in the measurement section B where two distortions are mixed, r_Ai, R_BiIs a single Lorentz-type power spectrum corresponding to distortions 1 and 2 with a peak value normalized to 1, H₁, H₂Is the peak value of each of these power spectra. r_Ai, R_BiIs a single Lorentz-type power spectrum normalized to 1, and can be expressed as the following equation.
[0028]
[Equation 9]

[0029]
Where ν_AB0, Ν_BB0Is the center frequency of the corresponding Lorentz-type power spectrum of distortions 1 and 2, ω_A0, Ω_B0Is the full width at half maximum of the Lorentz power spectrum corresponding to the strains 1 and 2.
[0030]
[Problems to be solved by the invention]
However, if the linearization method in the region where only a single strain exists as shown in Expression (7) is applied to the observed value of Brillouin scattered light in the measurement section B where two strains are mixed, the following square error J₂The parameter of the Lorentz type power spectrum is determined so as to minimize.
[0031]
[Expression 10]

[0032]
Where g_CiIs an observed value in the measurement section B. But y_CiIs a linear sum of Lorentz-type power spectra corresponding to distortions 1 and 2, as shown in equations (8) to (10).
[0033]
[Expression 11]

[0034]
And cannot be linearized. Therefore, the conventional method of least squares by linearization has a drawback that it cannot be applied to a measurement section in which two distortions are mixed. In addition, as described above, the conventional least square method by linearization should not be applied. However, if it is applied to the observation value of the measurement section in which two distortions are mixed by mistake, the single Lorentz forcefully is forcibly applied. Therefore, there is a problem that not only an erroneous center frequency is estimated but also an erroneous distortion is estimated as a result.
[0035]
Note that the parameter ν of the two Lorentz-type power spectra using the steepest descent method_AB0, Ω_A0, H₁, Ν_BB0, Ω_B0, H₂However, the Lorentz-type power spectrum is nonlinear as shown in the equations (8) to (11), and it is a problem of local minimum, which is a general drawback of the steepest descent method. There is also a problem that the reliability of the Lorentz-type power spectrum parameters and the finally obtained distortion value is low, without avoiding the problem and the problem of determining the initial value.
[0036]
The present invention has been made in view of such problems, and its object is to provide a Lorentz type corresponding to each distortion of Brillouin scattered light scattered in a measurement section of an optical fiber in which two distortions are mixed. An object of the present invention is to provide an optical fiber strain measurement method and apparatus for obtaining a lengthwise strain distribution generated in an optical fiber by estimating a peak value of a power spectrum.
[0037]
[Means for Solving the Problems]
  In order to achieve such an object, the invention according to claim 1 is directed to a Brillouin which is one of backscattered light generated by an optical pulse incident on an optical fiber used as a sensor and generated by the optical pulse. Scattered lightAs the Brillouin scattered light integrated in the length direction in the measurement section represented by the product of the time width of the optical pulse and the speed of light in the optical fiber.MeasureOf measured bullian lightIn an optical fiber strain measurement method for determining a peak distribution of a power spectrum, a center frequency giving the peak value and a full width at half maximum, and obtaining a strain distribution in a length direction generated in the optical fiber,Of the measurement intervals, a first measurement interval in which only a single first distortion exists, a second measurement interval in which only a single second distortion different from the first distortion exists, and the first The third distortion in which the second distortion and the second distortion are mixedThe power spectrum of the Brillouin scattered light scattered in the measurement intervalRespectivelyA measurement step for measuring, and the measurement step measured in the measurement stepFirst and secondFrom the power spectrum of the measurement interval,First and secondLorentz-type power spectrum with the peak value in the measurement section normalized to 1.RespectivelyA first estimating step to estimate, the normalized Lorentz-type power spectrum estimated in the first estimating step, and the measured in the measuring stepThirdUsing the power spectrum of the measurement section,ThirdSaid in the measurement intervalFirst and secondAnd a second estimation step for estimating a peak value of a Lorentz type power spectrum corresponding to the distortion of.
  More specifically, in the second estimation step, the normalized Lorentz power spectrum (in the first and second measurement sections estimated in the first estimation step) r _Ai , r _Bi ) Linear sum ( y _Ci ), And the power spectrum of the third measurement section measured in the measurement step ( g _Ci Squared error with (J _Three ) Of the Lorentz power spectrum corresponding to the first and second distortions in the third measurement interval ( H ₁ , H ₂ ) Is estimated using the least squares method.
[0038]
According to this method, the peak value of the Lorentz-type power spectrum corresponding to each distortion of the Brillouin scattered light scattered in the measurement section of the optical fiber in which two distortions are mixed is estimated to be generated in the optical fiber. The strain distribution in the length direction can be obtained.
[0039]
  Claim3The invention described in 1 is directed to an optical pulse generator for making an optical pulse incident on an optical fiber used as a sensor, and Brillouin scattered light that is one of backscattered light generated by the optical pulse.As the Brillouin scattered light integrated in the length direction in the measurement section represented by the product of the time width of the optical pulse and the speed of light in the optical fiber.A measuring unit for measuring,Of measured bullian lightIn the optical fiber strain measuring device for determining the peak value of the power spectrum, the center frequency giving the peak value and the full width at half maximum, and obtaining the strain distribution in the length direction generated in the optical fiber,Of the measurement intervals, a first measurement interval in which only a single first distortion exists, a second measurement interval in which only a single second distortion different from the first distortion exists, and the first The third distortion in which the second distortion and the second distortion are mixedThe power spectrum of the Brillouin scattered light scattered in the measurement intervalRespectivelyA measuring means for measuring, and the measuring means measured by the measuring meansFirst and secondFrom the power spectrum of the measurement interval,First and secondLorentz-type power spectrum with the peak value in the measurement section normalized to 1.RespectivelyA first estimating means for estimating; the normalized Lorentz power spectrum estimated by the first estimating means; and the measuring means measured by the measuring means.ThirdUsing the power spectrum of the measurement section,ThirdSaid in the measurement intervalFirst and secondAnd a second estimating means for estimating a peak value of a Lorentz type power spectrum corresponding to the distortion.
  More specifically, the second estimating means is the normalized Lorentz-type power spectrum of the first and second measurement sections estimated by the first estimating means ( r _Ai , r _Bi ) Linear sum ( y _Ci ) And the power spectrum of the third measurement section measured by the measuring means ( g _Ci Squared error with (J _Three ) Of the Lorentz power spectrum corresponding to the first and second distortions in the third measurement interval ( H ₁ , H ₂ ) Is estimated using the least squares method.
[0040]
  Claim5The invention described in 1 is directed to an optical pulse generator for making an optical pulse incident on an optical fiber used as a sensor, and Brillouin scattered light that is one of backscattered light generated by the optical pulse.As the Brillouin scattered light integrated in the length direction in the measurement section represented by the product of the time width of the optical pulse and the speed of light in the optical fiber.A measuring unit for measuring,Of measured bullian lightA program for controlling an optical fiber strain measuring device that determines a peak value of a power spectrum, a center frequency that gives the peak value, and a full width at half maximum, and calculates a lengthwise strain distribution generated in the optical fiber. A recorded recording medium,Of the measurement intervals, a first measurement interval in which only a single first distortion exists, a second measurement interval in which only a single second distortion different from the first distortion exists, and the first The third distortion in which the second distortion and the second distortion are mixedThe power spectrum of the Brillouin scattered light scattered in the measurement intervalRespectivelyA measurement step for measuring, and the measurement step measured in the measurement stepFirst and secondFrom the power spectrum of the measurement interval,First and secondLorentz-type power spectrum with the peak value in the measurement section normalized to 1.RespectivelyA first estimating step to estimate, the normalized Lorentz-type power spectrum estimated in the first estimating step, and the measured in the measuring stepThirdUsing the power spectrum of the measurement section,ThirdSaid in the measurement intervalFirst and secondIt is a computer-readable recording medium which records the program for making a computer perform the 2nd estimation step which estimates the peak value of a Lorentz type power spectrum corresponding to distortion of this.
  More specifically, in the second estimation step, the normalized Lorentz power spectrum (in the first and second measurement sections estimated in the first estimation step) r _Ai , r _Bi ) Linear sum ( y _Ci ), And the power spectrum of the third measurement section measured in the measurement step ( g _Ci Squared error with (J _Three ) Of the Lorentz power spectrum corresponding to the first and second distortions in the third measurement interval ( H ₁ , H ₂ ) Is estimated using the least squares method.
[0041]
  Claim7The invention described in 1 is directed to an optical pulse generator for making an optical pulse incident on an optical fiber used as a sensor, and Brillouin scattered light that is one of backscattered light generated by the optical pulse.As the Brillouin scattered light integrated in the length direction in the measurement section represented by the product of the time width of the optical pulse and the speed of light in the optical fiber.A measuring unit for measuring,Of measured bullian lightA program for controlling an optical fiber strain measuring device that determines a peak value of a power spectrum, a center frequency that gives the peak value, and a full width at half maximum, and obtains a strain distribution in the length direction generated in the optical fiber. There,Of the measurement intervals, a first measurement interval in which only a single first distortion exists, a second measurement interval in which only a single second distortion different from the first distortion exists, and the first The third distortion in which the second distortion and the second distortion are mixedThe power spectrum of the Brillouin scattered light scattered in the measurement intervalRespectivelyA measurement step for measuring, and the measurement step measured in the measurement stepFirst and secondFrom the power spectrum of the measurement interval,First and secondLorentz-type power spectrum with the peak value in the measurement section normalized to 1.RespectivelyA first estimating step to estimate, the normalized Lorentz-type power spectrum estimated in the first estimating step, and the measured in the measuring stepThirdUsing the power spectrum of the measurement section,ThirdSaid in the measurement intervalFirst and secondAnd a second estimation step for estimating the peak value of the Lorentz power spectrum corresponding to the distortion of the computer.
  More specifically, in the second estimation step, the normalized Lorentz power spectrum (in the first and second measurement sections estimated in the first estimation step) r _Ai , r _Bi ) Linear sum ( y _Ci ), And the power spectrum of the third measurement section measured in the measurement step ( g _Ci Squared error with (J _Three ) Of the Lorentz power spectrum corresponding to the first and second distortions in the third measurement interval ( H ₁ , H ₂ ) Is estimated using the least squares method.
[0042]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.
The Lorentz-type power spectrum estimation method corresponding to each distortion in a measurement section in which two distortions coexist is a measurement section in which there is only a single distortion considered to exist on both sides of a measurement section in which two distortions are mixed ( For example, by using the power spectrum of the Brillouin scattered light in the measurement section A and the measurement section C shown in FIG. 2, the parameter ν of the power spectrum in the measurement section in which two distortions are mixed is used._AB0, Ω_A0, H₁, Ν_BB0, Ω_B0, H₂Ask for.
[0043]
FIG. 3 is a flowchart showing a Lorentz power spectrum estimation method according to an embodiment of the present invention. In step S1, a Lorentz-type power spectrum of Brillouin scattered light is estimated using observation values in a measurement section where only a single strain exists on both sides of a measurement section in which two strains are mixed. As the estimation method, for example, Yamada, Naruse, “Estimation of BOTDR waveform by iterative weighted least-squares method”, IEICE Technical Report, OFT 2000-30, pp.55-60, 2000, use the conventional technique it can. In step S2, a Lorentz power spectrum with a peak value normalized to 1 is determined from the Lorentz power spectrum that is the estimation result of step S1.
[0044]
Next, step S3 will be described. Assuming that the conditions other than the distortion are the same for the measurement sections A, B, and C, the Lorentz power spectrum obtained by normalizing the peak value as a result of step S2 to 1 is represented by r_Ai, R_BiIs the same. Therefore, of the parameters of the measurement section where the two distortions to be found are mixed, ν_AB0, Ω_A0, Ν_BB0, Ω_B0Is obtained in step S2. Remaining peak value H₁, H₂Is the square error J₃The value that minimizes H₁, H₂Estimate as
[0045]
[Expression 12]

[0046]
H that minimizes Equation (13)₁, H₂Is given as a solution satisfying the following equation.
[0047]
[Formula 13]

[0048]
In this way, the parameter ν of the power spectrum in the measurement section where two distortions coexist._AB0, Ω_A0, H₁, Ν_BB0, Ω_B0, H₂Can be obtained, and the Lorentz-type power spectrum corresponding to each distortion in the measurement section where two distortions exist can be estimated.
[0049]
A numerical simulation is used to explain that the Lorentz-type power spectrum estimation method corresponding to each distortion in a measurement section in which two distortions are mixed according to the present invention is feasible and effective.
[0050]
A single Lorentz-type power spectrum with the peak value used for simulation normalized to 1._Ai(I = 1 to 100), r_Bi(I = 1 to 100) is prepared by a computer. Where ν_AB0= 10.5 GHz, ω_A0= 81 MHz, ν_BB0= 10.4 GHz, ω_B0= 82 MHz is used. R used in this simulation_Ai, R_BiIs added with the observation noise assumed in the normal distribution of 1/50, and the observation value in the measurement section of only a single distortion used in the simulation is created.
[0051]
FIG. 4 is a diagram showing observed values of Brillouin scattered light in a measurement section with only a single distortion used in the simulation. However, the solid line is r_AiObservation values created from the graph, dashed line is r_BiIt is the observation value made from. The vertical axis indicates relative values assuming that the values are normalized with a certain value. Apart from what I stopped with the computer, r_Ai, R_BiY according to formula (8)_CiCreate This y_CiIs added to the observed noise assumed in the normal distribution of 1/50, and the observed value in the measurement section where the two distortions used in the simulation are mixed is created.
[0052]
FIG. 5 is a diagram showing observed values in a measurement section in which two distortions used in the simulation are mixed. Where H₁= 0.7, H₂= 0.3 is used. The parameter ν of the power spectrum in the measurement section in which two distortions are mixed, obtained by applying the estimation method according to the present invention shown in FIG. 3 to the observation values created by these computers._AB0, Ω_A0, H₁, Ν_BB0, Ω_B0, H₂And the true value used in the simulation are shown in Table 1.
[0053]
[Table 1]

[0054]
As shown in Table 1, according to the present invention, H₁, H₂It can be seen that the parameters of the power spectrum can be estimated in the measurement section in which two distortions including are mixed.
[0055]
According to the present embodiment, it can be seen that a Lorentz power spectrum estimation method corresponding to each distortion in a measurement section in which two distortions coexist can be realized and is effective.
[0056]
As described above, by using the Lorentz power spectrum estimation method corresponding to each distortion in the measurement section where two distortions are mixed, the Lorentz corresponding to each distortion in the measurement section where two distortions are mixed. It is possible to estimate the parameters of the type power spectrum. Here, the energy of the Brillouin scattered light in the measurement section in which two strains are mixed is expressed by the following equation from the corresponding Lorentz power spectrum.
[0057]
[Expression 14]

[0058]
However, s₁Is the energy of Brillouin scattered light caused by strain 1 in the measurement section where two strains coexist, s₂Is the energy of the Brillouin scattered light due to strain 2, r_A, R_BIs r before being discretized_Ai, R_BiIt is.
[0059]
Assuming that there is no difference in loss of Brillouin scattered light due to strain 1 and strain 2, and that the energy of each Brillouin scattered light is caused by the length of each strain existing in the measurement section where two strains are mixed, Since the energy of the light pulse is constant, s in equations (15) and (16)₁And s₂The sum of and is constant. Therefore, ω_A0, H₁, Ω_B0, H₂S₁And s₂The ratio of the lengths of strain 1 and strain 2 existing in the measurement section where two strains are mixed can be obtained.
[0060]
【The invention's effect】
As described above, according to the present invention, it is possible to estimate the Lorentz power spectrum parameter corresponding to each distortion in the measurement section where two distortions coexist, and the result is generated in the optical fiber. Thus, it is possible to obtain the strain distribution in the length direction.
[Brief description of the drawings]
FIG. 1 is a block diagram showing a conventional optical fiber strain measuring instrument.
FIG. 2 is a diagram showing a measurement section in which two distortions are mixed.
FIG. 3 is a flowchart showing a Lorentz power spectrum estimation method according to an embodiment of the present invention.
FIG. 4 is a diagram showing observed values of Brillouin scattered light in a measurement section with only a single strain used in the simulation.
FIG. 5 is a diagram showing observed values in a measurement section in which two distortions used in a simulation are mixed.
[Explanation of symbols]
10 Optical fiber strain measuring instrument
11 Light source
12 Optical frequency converter
13 Optical pulse converter
14 Coherent optical receiver
15 Signal processor

Claims (8)

A light pulse is incident on an optical fiber used as a sensor, and Brillouin scattered light , which is one of backscattered light generated by the light pulse, is obtained by multiplying the time width of the light pulse by the speed of light in the optical fiber. Measured as bullian scattered light integrated in the length direction in the measurement section represented, and determined the peak value of the power spectrum of the measured bullian scattered light, the center frequency giving the peak value, and the full width at half maximum In the optical fiber strain measurement method for obtaining the strain distribution in the length direction generated in the optical fiber,
Of the measurement intervals, a first measurement interval in which only a single first distortion exists, a second measurement interval in which only a single second distortion different from the first distortion exists, and the first a measurement step of measuring respective power spectrum of the Brillouin scattered light distortion and said second distortion is scattered at a third measuring section to be mixed,
From the power spectrum of the measured said measured in step the first and second measurement interval, the estimated respective Lorentz type power spectrum obtained by normalizing the peak values of the first and second measurement period to 1 1 An estimation step;
Using the normalized Lorentz-type power spectrum estimated in the first estimation step and the power spectrum of the third measurement interval measured in the measurement step, the first in the third measurement interval . And a second estimating step for estimating a peak value of a Lorentz type power spectrum corresponding to the first and second distortions. In the second estimating step, the normalized Lorentz power spectrum of the first and second measurement sections estimated in the first estimating step ( r _Ai , r _Bi ) Linear sum ( y _Ci ), And the power spectrum of the third measurement section measured in the measurement step ( g _Ci Squared error with (J _Three ) Of the Lorentz-type power spectrum corresponding to the first and second distortions in the third measurement interval ( H ₁ , H ₂ ) Is estimated using a least squares method, the optical fiber strain measurement method according to claim 1. An optical pulse generator for making an optical pulse incident on an optical fiber used as a sensor, and Brillouin scattered light that is one of backscattered light generated by the optical pulse, the time width of the optical pulse and the optical fiber in the optical fiber A measurement unit that measures as a Brillouin scattered light integrated in the length direction in a measurement section represented by the product of the speed of light, and a peak value of the power spectrum of the measured Brillouin scattered light , and the peak value In the optical fiber strain measuring device for determining the center frequency and the full width at half maximum to give the strain distribution in the length direction generated in the optical fiber,
Of the measurement intervals, a first measurement interval in which only a single first distortion exists, a second measurement interval in which only a single second distortion different from the first distortion exists, and the first Measuring means for respectively measuring the power spectrum of the Brillouin scattered light scattered in the third measurement section in which the distortion of the second and the second distortion are mixed ,
A first Lorentz-type power spectrum in which the peak values in the first and second measurement sections are normalized to 1 are respectively estimated from the power spectra in the first and second measurement sections measured by the measuring means. An estimation means;
Using a Lorentz type power spectrum in which the normalized estimated by the first estimating means and the power spectrum of the third measurement interval measured by said measuring means, said in the third measurement interval the And a second estimating means for estimating a peak value of a Lorentz type power spectrum corresponding to the first and second distortions. The second estimating means is the normalized Lorentz-type power spectrum of the first and second measurement sections estimated by the first estimating means ( r _Ai , r _Bi ) Linear sum ( y _Ci ) And the power spectrum of the third measurement section measured by the measuring means ( g _Ci Squared error with (J _Three ) Of the Lorentz power spectrum corresponding to the first and second distortions in the third measurement interval ( H ₁ , H ₂ ) Is estimated using the least squares method, The optical fiber distortion measuring apparatus according to claim 3, wherein: An optical pulse generator for making an optical pulse incident on an optical fiber used as a sensor, and Brillouin scattered light that is one of backscattered light generated by the optical pulse, the time width of the optical pulse and the optical fiber in the optical fiber A measurement unit that measures as a Brillouin scattered light integrated in the length direction in a measurement section represented by the product of the speed of light, and a peak value of the power spectrum of the measured Brillouin scattered light , and the peak value A recording medium on which a program for controlling an optical fiber strain measuring device for determining a distribution of longitudinal strain generated in the optical fiber is determined by determining a center frequency and a full width at half maximum,
Of the measurement intervals, a first measurement interval in which only a single first distortion exists, a second measurement interval in which only a single second distortion different from the first distortion exists, and the first a measurement step of measuring respective power spectrum of the Brillouin scattered light distortion and said second distortion is scattered at a third measuring section to be mixed,
From the power spectrum of the measured said measured in step the first and second measurement interval, the estimated respective Lorentz type power spectrum obtained by normalizing the peak values of the first and second measurement period to 1 1 An estimation step;
Using the normalized Lorentz-type power spectrum estimated in the first estimation step and the power spectrum of the third measurement interval measured in the measurement step, the first in the third measurement interval . A computer-readable recording medium storing a program for causing a computer to execute a second estimation step of estimating a peak value of a Lorentz type power spectrum corresponding to the first and second distortions. In the second estimating step, the normalized Lorentz power spectrum of the first and second measurement sections estimated in the first estimating step ( r _Ai , r _Bi ) Linear sum ( y _Ci ), And the power spectrum of the third measurement section measured in the measurement step ( g _Ci Squared error with (J _Three ) Of the Lorentz-type power spectrum corresponding to the first and second distortions in the third measurement interval ( H ₁ , H ₂ The computer-readable recording medium according to claim 5, wherein the method is estimated using a least square method. An optical pulse generator for making an optical pulse incident on an optical fiber used as a sensor, and Brillouin scattered light that is one of backscattered light generated by the optical pulse, the time width of the optical pulse and the optical fiber A measurement unit that measures as a Brillouin scattered light integrated in the length direction in a measurement section represented by the product of the speed of light, and a peak value of the power spectrum of the measured Brillouin scattered light , and the peak value Is a program for controlling an optical fiber strain measuring device to determine a center frequency and a full width at half maximum to obtain a strain distribution in the length direction generated in the optical fiber,
Of the measurement intervals, a first measurement interval in which only a single first distortion exists, a second measurement interval in which only a single second distortion different from the first distortion exists, and the first a measurement step of measuring respective power spectrum of the Brillouin scattered light distortion and said second distortion is scattered at a third measuring section to be mixed,
From the power spectrum of the measured said measured in step the first and second measurement interval, the estimated respective Lorentz type power spectrum obtained by normalizing the peak values of the first and second measurement period to 1 1 An estimation step;
Using the normalized Lorentz-type power spectrum estimated in the first estimation step and the power spectrum of the third measurement interval measured in the measurement step, the first in the third measurement interval . And a second estimating step for estimating a peak value of a Lorentz type power spectrum corresponding to the first and second distortions. In the second estimating step, the normalized Lorentz power spectrum of the first and second measurement sections estimated in the first estimating step ( r _Ai , r _Bi ) Linear sum ( y _Ci ), And the power spectrum of the third measurement section measured in the measurement step ( g _Ci Squared error with (J _Three ) In the third measurement interval in which the first And the Lorentz power spectrum peak value corresponding to the second distortion ( H ₁ , H ₂ ) Is estimated using the method of least squares.

Images