JP2004004023A - Infrared flame detector and infrared flame detecting method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an infrared flame detector and an infrared flame detecting method in which a true flame can be detected surely while preventing erroneous detection of an infrared phenomenon. <P>SOLUTION: The infrared flame detector for detecting a flame by observing the intensity of infrared rays radiated from the flame comprises a phase difference detecting means 230 for measuring the time difference of output between infrared detectors having different wavelength sensitivity, and a phase difference randomness judging means 240 for judging the flame based on the output from the phase difference detecting means. <P>COPYRIGHT: (C)2004,JPO

Description

【００９０】
【数２】

【００９１】

【００９２】
さらに、これらの推定値から以下の式によって初期の検出閾値を図１３の閾値設定２３５０においてあらかじめ設定しておく。
【００９３】
【数３】

【００９４】
ここで、ｂ_Ｎ＋１は第Ｎ＋１番目のサンプル時刻に対して用いる検出閾値、定数αは安全係数である。

【００９５】
ここでは、改めて第ｉ番目（ｉはＮ＋１、Ｎ＋２、…）の赤外線波高値ｘ_ｉが得られたとする。このときすでに、検出閾値ｂ_ｉ（ｉはＮ＋１、Ｎ＋２、…）が求まっているから、図１３のレベル判定２３６０において波高値Ｘ_ｉと検出閾値ｂ_ｉの比較２２３０で比較する。
【００９６】
もしも波高値ｘ_ｉが検出閾値ｂ_ｉよりも小さければ、赤外線波高値の平均偏差の更新２２５０に進む。

【００９７】
【数４】

【００９８】
ここで、Ｋ_ｘは平均値を推定するための係数であり１よりも大きな定数を設定する。

【００９９】
【数５】

【０１００】
ここで、Ｋ_ｄは平均偏差を推定するための係数であり、１よりも大きな定数を設定する。

【０１０１】
【数６】

【０１０２】
このように新たな検出域を計算して赤外線波高値の入力２２２０に戻る。

【０１０３】
以上説明した本発明の第３実施例による効果について、図１４及び図１５を用いて説明する。
【０１０４】
図１４において、２４００は背景赤外線変動の波形、２４１０は推定平均値、２４２０は推定平均偏差、２４３０は検出閾値を示す。
【０１０５】
図１５において、図１５（ａ）は赤外線検知窓洗浄直後の背景赤外線検知出力、図１５（ｂ）は赤外線検知窓汚染後の背景赤外線検知出力をそれぞれ示しており、２５００は赤外線検出窓に汚損がない場合に観測される背景赤外線変動の波形、２５１０は赤外線検出窓に汚損がある場合に観測される背景赤外線変動の波形を示す。
【０１０６】
本発明の第３実施例に示したような方法を取れば、火災が起こっていない通常の状態における背景赤外線源２１８０（図１１参照）からの背景赤外線変動の波形２４００から赤外線強度の推定平均値２４１０を得ることができる。またこの赤外線強度の推定平均値２４１０に対する推定平均偏差２４２０も推定することができる。これら赤外線強度の推定平均値２４１０と推定平均偏差２４２０を基づいて災検出の検出閾値２４３０を定めるため、使用環境における背景赤外線の強弱の影響を受けることなく、最適な検出閾値２４３０で炎を検出することができる。
【０１０７】
また、図１５に示したように赤外線検出窓に汚損がない場合には、このときに観測される背景赤外線変動の波形から火災検出の検出閾値を比較的高く定めるのに対し、２５１０のように赤外線検出窓に汚損がある場合には、観測される背景赤外線変動の波形が小さくなるため、火災検出の検出閾値は自動的に低下する。このため、赤外線検出窓に汚損が生じた場合でも最適な検出閾値で火災を検出する。
【０１０８】
次に、本発明の第４実施例について説明する。
【０１０９】
本発明の第４実施例において、炎検出装置の構成は本発明の第３実施例と同様の構成であるため、図１１を用いて説明する。
【０１１０】
図１６は本発明の第４実施例における炎検出装置のフローチャートである。
【０１１１】
この図において、２６００は処理の開始、２６１０は初期値の設定、２６２０は赤外線波高値の入力とバッファリング、２６３０は赤外線波高値の時系列のフーリエ変換、２６４０は赤外線波高値の時系列のスペクトルと検出閾値の比較、２６５０はスペクトルの平均値の更新、２６６０はスペクトル平均偏差の更新、２６７０は炎検出発報処理、２６８０は終了である。
【０１１２】
図１７に本発明の第４実施例におけるＣＰＵ内部の処理のブロック図を示す。
【０１１３】
この図において、２７００は赤外線強度入力、２７１０は波形バッファ処理、２７２０はフーリエ変換、２７３０は各周波数毎の赤外線強度の平均値推定、２７４０は各周波数毎の平均値記憶、２７５０は各周波数毎の平均偏差推定、２７６０は各周波数毎の平均偏差記憶、２７７０は各周波数毎の閾値設定、２７８０は各周波数毎のレベル判定、２７９０は判定出力である。
【０１１４】
次に、本発明の第４実施例における炎検出装置の動作を、図１１を用いて説明する。
【０１１５】
まず、火災が起こっていない通常の状態における背景赤外線源２１８０からはその表面温度や太陽光の反射に応じた赤外線が放射されている。この背景赤外線は保護サファイアガラス２１７０を通過し、赤外線フィルタ２１００によって通過する波長を制限され、４．４μｍ付近の赤外線がサーモパイル素子２１１０に到達する。サーモパイル素子２１１０は到達してきた赤外線の強度に応じた電気信号を生じ、増幅回路２１２０によってこの電気振動を増幅する。ローパスフィルタ２１３０はこの電気信号の通過帯域を制限し、サンプルホールド回路２１４０はこの電気信号をサンプリングし、ＡＤ変換器２１５０によって赤外線強度を数値化する。ＣＰＵ２１６０はこの数値化された赤外線強度に対して炎検出処理を行う。
【０１１６】
次に、本発明の第４実施例において図１１のＣＰＵ２１６０が行う処理について、図１６と図１７を用いて説明する。
【０１１７】
まず、処理の開始２６００で処理を開始する。初期値の設定２６１０では開始時の各周波数平均値記憶２７４０と各周波数平均偏差記憶２７６０に記憶する各種パラメータの初期化を行う。第４実施例では波形バッファ処理２７１０において赤外線強度信号をあらかじめ定めた時間区間長の分析区間毎に区切ってフーリエ変換２７２０において離散フーリエ変換したスペクトル強度を用いた処理を行う。第４実施例においては以下のような式によって各種パラメータの初期化を行う。
【０１１８】
【数７】

【０１１９】
【数８】

【０１２０】

【０１２１】
さらに、これらの推定値から以下の式によって各周波数閾値設定２７７０において初期の検出閾値をあらかじめ設定しておく。
【０１２２】
【数９】

【０１２３】
ここで、Ｂ（ω）_Ｍ＋１は第Ｍ＋１番目の分析区間における変動周波数ωに対して用いる検出閾値、定数αは安全係数である。

【０１２４】
波高値入力とバッファリング２６２０ではあらかじめ定めた分析区間の時間幅に相当するサンプル数の波高値を入力し、ＣＰＵ２１６０の内部バッファに格納する。
【０１２５】
次にフーリエ変換２６３０で図１７のフーリエ変換２７２０において離散フーリエ変換を行う。ＦＦＴアルゴリズムを使うのが好適である。ここでは改めて第ｊ番目（ｊ＝Ｍ＋１，Ｍ＋２，…）の分析区間における変動周波数ωの赤外線変動スペクトルＸ（ω）_ｊが得られたとする。このときすでに検出閾値Ｂ（ω）_ｊ（ｊ＝Ｍ＋１，Ｍ＋２，…）が図１７の各周波数閾値設定２７７０で求まっているから、赤外線変動スペクトルの絶対値｜Ｘ（ω）_ｊ｜と検出閾値Ｂ（ω）_ｊの比較２６４０を図１７の各周波数レベル判定２７８０で比較する。
【０１２６】
もしも赤外線変動スペクトルの絶対値｜Ｘ（ω）_ｊ｜が検出閾値Ｂ（ω）_ｊよりも小さければスペクトル平均値の更新２６５０に進む。

【０１２７】
【数１０】

【０１２８】
ここでＫ_ｘは平均値を推定するための係数であり、１よりも大きな定数を設定する。結果は図１７の各周波数平均値記憶２７４０に記憶する。

【０１２９】
【数１１】

【０１３０】
ここで、Ｋ_ｄは平均偏差を推定するための係数であり、１よりも大きな定数を設定する。結果は図１７の各周波数平均偏差記憶２７６０に記憶する。

【０１３１】
【数１２】

【０１３２】
このように、新たな検出閾値を計算して波高値入力とバッファリング２６２０に戻る。
【０１３３】
一方、図１１において火災が起こった際における赤外線源２１９０が生じた場合には、通常のスペクトルの推定平均値と推定平均偏差で設定された検出閾値Ｂ（ω）_ｊ＋１よりも高いスペクトルが観測される。このような場合で、図１６におけるスペクトルの絶対値｜Ｘ（ω）_ｊ｜と検出閾値Ｂ（ω）_ｊの比較２６４０を図１７の各周波数レベル判定２７８０で行った結果、スペクトルのほうが大きい場合には、炎検出発報処理２６７０にすすみ、図１７の判定出力２７９０により検出した場合の出力を行い終了２６８０となる。
【０１３４】
以上説明した本発明の第４実施例による効果について説明する。
【０１３５】
本発明の第４実施例では赤外線波高値の変化を離散フーリエ変換することにより離散周波数成分に分解している。このため、例えば、回転灯などの周期的な赤外線変動を示すような外乱は定常的なスペクトル強度として観測されるため、スペクトル平均値として閾値にあらかじめ含まれてしまう。
【０１３６】
また、一般に炎の赤外線強度変化は１／ｆ特性を示す不規則な変動周波数成分を持っており、前記外乱による一定な周波数以外の周波数成分を用いることによって周期的な外乱の影響を受けない周波数成分の上昇で炎を検出できる。したがって、第３実施例においては回転灯等の周期的な外乱が常に存在するような場合には、波高値に対する閾値が上昇するため検出が困難になる問題があったのに対し、第４実施例のように周波数成分を用いれば周期的な外乱があっても検出性能に対する影響が少ない。
【０１３７】
また、第３実施例において火炎検出の検出閾値を自動的に定めることによって、使用環境における背景赤外線の強弱の影響を受けることなく、最適な検出閾値で炎を検出するという効果は、第４実施例においても同様に有効である。
【０１３８】
さらに、第３実施例において赤外線検出窓に汚損が生じた場合でも、汚損に応じて検出閾値が低下することにより最適な検出閾値で火炎を検出するという効果は、第４実施例においても同様に有効である。
【０１３９】
本発明の第３実施例では背景赤外線強度の波高値、第４実施例では背景赤外線のスペクトル強度特徴量として用いて炎検出のための閾値を設定したが、本発明では必ずしもこれらの特徴量を使用する必要はなく、例えば波高値の移動平均値、フィルタバンク出力、線形予測係数などの特徴量を使用しても有効である。
【０１４０】
本発明の第３及び第４実施例では、波高値やスペクトルなどの特徴量の平均値とその平均偏差から検出閾値を決定したが、必ずしもこれらの統計量を用いて検出閾値を決定する必要はなく、例えば正規分布、Ｘ−２乗分布等を表現する平均値、標準偏差等の各種統計量を用いて検出閾値を決定することは、容易に類推できる。
【０１４１】
本発明の第３及び第４実施例では波高値やスペクトルなどの特徴量の平均値とその平均偏差を推定するに当たって簡便なローパスフィルタによる平均値、平均偏差の推定を行ったが、必ずしもこのような推定方法を取る必要はなく、ニューラルネットや特異値の判別除去法等による非線形な推定方法をとることは容易に類推できる。
【０１４２】
本発明の第３及び第４実施例では、波高値やスペクトルなどの特徴量と検出閾値の単純な比較によって炎検出の判定を行ったが、必ずしもこのような単純な検出の判定を行う必要はなく、特異であると検出された信号の継続性による判定や、スペクトルに対する重み付けを用いたパラメータを用いるなど、音声区間検出法等で用いられている各種の判定方法を使用することは容易に類推できる。
【０１４３】
本発明の第３及び第４実施例では、炎が輻射する赤外線のうち炎検出に使用する赤外線波長域として４．４μｍ付近の波長域を通過させる赤外線フィルタを用いたが、必ずしもこの波長域を用いる必要はなく、使用目的に応じて他の赤外線波長域や可視光、紫外線等を用いる場合でも本発明は有効である。
【０１４４】
図１８は本発明の第５実施例を示す炎検出方式の構成図であり、上記した第１〜第４実施例を組み合わせて構成される炎検出方式のブロック図である。
【０１４５】
図１８において、２８００は第１の４．４μｍ波長域の赤外線変動波形を互いに重畳した時間（分析）区間に区切って窓データの切り出しを行う処理、２８０５は第２の３．３μｍ波長域の赤外線変動波形を互いに重畳した分析（時間）区間に区切って窓データの切り出しを行う処理、２８１０は第１の波長域の変動波形のフーリエ変換処理、２８１５は第２の波長域の変動波形のフーリエ変換処理、２８２０はそのフーリエ変換処理後の平均値推定処理、２８２５はそのフーリエ変換処理後の平均偏差推定処理、２８３０は平均値記憶処理、２８３５は平均偏差記憶処理、２８４０は平均値と平均偏差とに基づく閾値設定処理、２８５０は閾値に基づくレベル判定処理、２８６０は第１の４．４μｍの波長域の変動スペクトルと第２の波長域の３．３μｍの変動スペクトルとの位相差を検出する処理、２８７０は前記位相差を検出する処理２８６０で検出した位相差を用いて位相差のランダム性を判定する位相差ランダム性判定処理、２８８０は前記位相差ランダム性判定と前記閾値に基づくレベル判定とに基づいて判定出力を得る判定出力処理である。
【０１４６】
あお、本発明は上記実施例に限定されるものではなく、本発明の趣旨に基づいて種々の変形が可能であり、これらを本発明の範囲から排除するものではない。
【０１４７】
【発明の効果】
以上、詳細に説明したように、本発明によれば、以下のような効果を奏することができる。
【０１４８】
（Ａ）炎ではない赤外線現象を炎と誤検出することがなく、真実の炎の検出を的確に行うことができる。
【０１４９】
（Ｂ）まず赤外線の強度あるいは強度変化の上昇を検出するという、通常の赤外線式炎検出方式と同様の検出方式をとる。さらに、このような赤外線現象に対してその原因が炎によるものであるかどうかを、互いに異なる波長域の赤外線強度変化に時間のずれを生ずるという炎特有の特徴を用いて判定し、この処理によって炎以外の赤外線現象と炎による赤外線現象を確実に判定して炎である場合にのみ炎検出出力を行う。
【０１５０】
（Ｃ）まず赤外線の強度あるいは強度変化の上昇を検出するという、通常の赤外線式炎検出方式と同様の検出方式をとる。さらに、このような赤外線現象に対してその原因が炎によるものであるかどうかを、互いに異なる波長域の赤外線強度変化に時間のずれを生ずるという炎特有の特徴を用いて判定し、この処理によって炎以外の線現象と炎による赤外線現象を確実に判定して炎である場合にのみ炎検出出力を行う。
【０１５１】
（Ｄ）使用環境における背景赤外線の強弱や赤外線検出窓の汚損状態の影響を受けることなく、最適な閾値で炎を検出する炎検出装置及び方法を提供することができる。
【図面の簡単な説明】
【図１】本発明の第１実施例を示す炎検出装置の構成図である。
【図２】本発明の第１実施例を示す炎検出方式の構成図である。
【図３】本発明の第１実施例における炎検出処理のフローチャートである。
【図４】本発明の第１実施例による炎検出方式の赤外線強度と分析窓関数の関係についての説明図である。
【図５】本発明の第１実施例による炎検出方式における４．４μｍ波長域の赤外線強度変化の各分析区間における変動スペクトルの説明図である。
【図６】本発明の第１実施例による炎検出方式における４．４μｍ波長域と３．３μｍ波長域の赤外線強度変化時間波形の説明図である。
【図７】実際の赤外線を生ずる外乱と炎の４．４μｍ波長域と３．３μｍ波長域の赤外線強度の位相差についてクロススペクトル（複素数）で表現したデータを示す図である。
【図８】図３に示す処理３３０における４．４μｍ波長と３．３μｍ波長域の赤外線変動周波数成分の間の位相差の集計方法の説明図である。
【図９】本発明の第２実施例による炎検出装置の構成図である。
【図１０】本発明の第２実施例による炎検出装置の各信号のふるまいを示す図である。
【図１１】本発明の第３実施例による炎検出装置の構成図である。
【図１２】本発明の第３実施例による炎検出装置の動作フローチャートである。
【図１３】本発明の第３実施例による炎検出装置のＣＰＵ内部の処理ブロック図である。
【図１４】本発明の第３実施例による炎検出方式における検出閾値算出の説明図である。
【図１５】本発明の第３実施例による炎検出方式における赤外線検知窓汚染時の現象の説明図である。
【図１６】本発明の第４実施例による炎検出装置の動作フローチャートである。
【図１７】本発明の第４実施例による炎検出装置のＣＰＵ内部の処理ブロック図である。
【図１８】本発明の第５実施例を示す炎検出方式の構成図である。
【符号の説明】
１００，１０５，９００，９０５，２１００　　赤外線フィルタ
１１０，１１５，９１０，９１５，２１１０　　サーモパイル素子
１２０，１２５，９２０，９２５，２１２０　　増幅回路（Ａｍｐ）
１３０，１３５，９５５，９５７，９７０，９８０，２１３０　　ローパスフィルタ（ＬＰＦ）
１４０，１４５，２１４０　　サンプルホールド回路（Ｓ／Ｈ）
１５０，１５５，２１５０　　ＡＤ変換器（Ａ−Ｄ）
１６０，２１６０　　中央処理装置（ＣＰＵ）
１７０，９９８，２１７０　　保護サファイアガラス
１８０，２１８０　　通常の状態における背景赤外線源
１９０，９９５，２１９０　　火災が起こった際における赤外線源
２００，２０５，２１０，２１５，２２０，２３０，２４０　　処理（図２）３００，３０５，３１０，３１５，３２０，３２５，３３０，３３５，３４０，３５０，３６０　　処理（図３）
４００　　赤外線強度信号
４１０　　第１の分析区間の窓関数
４２０　　第２の分析区間の窓関数
４３０　　第３の分析区間の窓関数
５００　　赤外線強度の各変動周波数成分
５１０　　傾斜を持った平面
７００　　６０Ｗの白熱電球を赤外線源とした場合クロススペクトルの分布
７１０　　実際の炎を赤外線源として用いた場合のクロススペクトルの分布
９３０，９３２，９３５，９３７　　バンドパスフィルタ（ＢＰＬ）
９４０，９４２，９４５，９４７　　電圧比較器（Ｃｏｍｐ）
９５０，９５２　　排他的論理和回路
９６０　　加算器（Ｍｉｘ）
９６５，９８０　　比較器（Ｃｏｍｐ）
９７０　　ハイパスフィルタ（ＨＰＦ）
９７２　　整流回路
９８５　　論理積回路
９９０　　炎判定出力
９９８　　保護サファイアガラス
１０００，１０１０，１０３０，１０４０　　バンドパスフィルタの出力
１００５，１０１５，１０３５，１０４５　　電圧比較器の出力
１０２０，１０５０　　排他的論理和回路の出力
１０２５，１０５５　　ローパスフィルタの出力
２２００，２２１０，２２２０，２２３０，２２４０，２２５０，２２６０，２２７０　　処理（図１２）
２３００，２３１０，２３２０，２３３０，２３４０，２３５０，２３６０，，２３７０　　処理（図１３）
２４００　　背景赤外線変動の波形
２４１０　　推定平均値
２４２０　　推定平均偏差
２４３０　　検出閾値
２５００　　赤外線検出窓に汚損がない場合に観測される背景赤外線変動の波形
２５１０　　赤外線検出窓に汚損がある場合に観測される背景赤外線変動の波形
２６００，２６１０，２６２０，２６３０，２６４０，２６６０，２６７０，２６８０　　処理（図１６）
２７００，２７１０，２７２０，２７３０，２７４０，２７５０，２７６０，２７７０，２７８０，２７９０　　処理（図１７）
２８００，２８０５，２８１０，２８１５，２８２０，２８２５，２８３０，２８３５，２８４０，２８５０，２８６０，２８７０，２８８０　　処理（図１８）[0001]
TECHNICAL FIELD OF THE INVENTION
The present invention relates to an infrared flame detection device and an infrared flame detection method used for detecting a fire and the like.
[0002]
[Prior art]
Conventionally, a flame detection device has detected a flame by observing the intensity of infrared rays emitted from the flame and the characteristics of the intensity change.
[0003]
[Problems to be solved by the invention]
However, when the intensity of infrared rays radiated from a flame and the characteristic of the intensity change are used as in the past, it is erroneously detected as a flame due to an infrared phenomenon that is not a flame, for example, the passage of a high-temperature object. There was a possibility of false detection.
[0004]
An object of the present invention is to provide a flame detection device and a flame detection method which eliminate the above-described problems and can accurately detect a true flame without erroneously detecting an infrared phenomenon that is not a flame as a flame. Aim.
[0005]
[Means for Solving the Problems]
The present invention, in order to achieve the above object,
[1] In an infrared type flame detecting device for detecting a flame by observing the intensity of infrared rays emitted from the flame, a means for measuring a time lag of an output between infrared detectors having different wavelength sensitivities, and Means for judging a flame based on the output of (1).
[0006]
[2] In the infrared flame detecting apparatus according to [1], a change in infrared intensity around 4.4 μm mainly caused by a carbon dioxide gas resonance phenomenon at the time of combustion and a wavelength around 3.0 μm mainly caused by high-temperature combustion residues. The method is characterized in that a flame and a phenomenon other than the flame are distinguished by measuring a time lag of a change in infrared intensity.
[0007]
[3] In the infrared flame detection device according to the above [1] or [2], when the increase in the infrared intensity is detected, a phase difference of the infrared intensity change is measured.
[0008]
[4] The infrared flame detection device according to [1] or [2], wherein a phase difference of the infrared intensity change is measured when the increase in the infrared intensity change width is detected.
[0009]
[5] The infrared flame detector according to any one of [1] to [4], wherein the infrared intensity change or the phase difference of the infrared intensity change is observed by Fourier transform.
[0010]
[6] In the infrared flame detection method for detecting a flame by observing the intensity of infrared rays emitted from the flame, the flame is determined by measuring a time lag of an output between infrared detectors having different wavelength sensitivities. Is determined.
[0011]
[7] In the infrared flame detection device according to [1], the average intensity of the background radiated infrared rays during a time when no flame is generated is observed, and the flame detection threshold value is determined from the average intensity. It is characterized by.
[0012]
[8] In the infrared flame detection device according to the above [1], observing the average fluctuation of the background radiated infrared ray during the time when no flame is generated, and determining the flame detection threshold from the average fluctuation. It is characterized by.
[0013]
[9] In the infrared flame detection device according to the above [7], a flame detection threshold value is determined from the average intensity according to a state of fouling of a window for transmitting infrared light for detecting infrared light intensity. It is characterized by.
[0014]
[10] The infrared flame detection device according to the above [8], wherein a flame detection threshold value is determined from the average fluctuation in accordance with the state of fouling of a window that transmits infrared light for detecting infrared light intensity. And
[0015]
BEST MODE FOR CARRYING OUT THE INVENTION
Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.
[0016]
First, a first embodiment of the present invention will be described.
[0017]
FIG. 1 is a configuration diagram of a flame detection device showing a first embodiment of the present invention, and FIG. 2 is a configuration diagram of the flame detection system.
[0018]
In FIG. 1, reference numeral 100 denotes an infrared filter that passes a wavelength range around 4.4 μm as a first infrared wavelength range used for flame detection among the infrared rays radiated by the flame, and 105 denotes a flame detection among the infrared rays radiated by the flame. An infrared filter that passes a wavelength range around 3.3 μm as a second infrared wavelength range to be used, 110 is a thermopile element that converts infrared intensity in the first wavelength range to voltage, and 115 is infrared intensity in the second wavelength range Is a thermopile element, which converts a signal into a voltage; 120, an amplifier circuit (Amp) for amplifying the infrared intensity voltage in the first wavelength range; 125, an amplifier circuit (Amp) for amplifying the infrared intensity voltage in the second wavelength range; A low-pass filter (LPF) 135 for limiting the frequency component of the change in the infrared intensity in the first wavelength range limits the frequency component of the change in the infrared intensity in the second wavelength range. A low-pass filter (LPF) 140, a sample and hold circuit (S / H) 140 for holding the infrared intensity voltage in the first wavelength range, and 145 a sample and hold circuit (S / H) for holding the infrared intensity voltage in the second wavelength range. H), 150 are AD converters (AD) for digitizing the infrared intensity in the first wavelength range, 155 are AD converters (AD) for digitizing the infrared intensity in the second wavelength range, 160 Denotes a central processing unit (CPU) for detecting and judging a flame based on a change in the intensity of infrared rays in the first and second wavelength ranges, and 170 denotes an infrared filter 100 in the first wavelength range and a second wavelength range. Protective sapphire glass that protects the infrared filter 105 from contamination and impact from the outside and transmits infrared light, 180 is a background infrared light source in a normal state where no fire has occurred, and 190 is when a fire has occurred. Show a definitive infrared source.
[0019]
In FIG. 2, reference numeral 200 denotes a process of cutting out window data by dividing into a time (analysis) section in which infrared fluctuation waveforms of the first 4.4 μm wavelength region are superimposed on each other, and 205 denotes an infrared ray of the second 3.3 μm wavelength region A process for cutting out window data by dividing the variation waveform into analysis (time) sections superimposed on each other, 210: Fourier transform processing of the variation waveform in the first wavelength range, 215: Fourier transform of the variation waveform in the second wavelength range Processing, 220 is a process of detecting an abnormal signal using a fluctuation waveform in the 4.4 μm wavelength region and its Fourier transform, and 230 is a fluctuation spectrum of the first 4.4 μm wavelength region and the second wavelength region. A process 240 for detecting a phase difference from the fluctuation spectrum of 3 μm, 240 is a process of detecting the abnormal signal detected in the process 220 for detecting the abnormal signal and the phase difference detected in the process 230 for detecting the phase difference. This is a process for judging and outputting whether or not a flame has occurred (the randomness of the phase difference).
[0020]
FIG. 3 shows a flowchart of the flame detection processing in the first embodiment of the present invention.
[0021]
As shown in this figure, 300 is the start of processing, 305 is initialization, and 310 is one-quarter of the Fourier transform analysis section of the infrared intensity waveform in the two wavelength ranges of 4.4 μm and 3.3 μm. A process for capturing a sample, 315 is a multiplication process of a window function for one Fourier transform analysis section of infrared intensity waveforms in two wavelength ranges of 4.4 μm and 3.3 μm, respectively, and 320 is two wavelengths of 4.4 μm and 3.3 μm. Fast Fourier transform processing for each Fourier transform analysis section of the infrared intensity waveform in the region, 325 is a process for determining whether or not the fluctuation component in the wavelength region of 4.4 μm exceeds the detection threshold for the intensity level set in advance, 330 is The process 335 collects and digitizes the phase difference between the infrared intensity fluctuation frequency components in the two wavelength ranges of 4.4 μm and 3.3 μm, and 335 is the process of 4.4 μm and 3.3 μm. It is determined whether there are many components having a large phase difference between the infrared intensity fluctuation frequency components in the wavelength range of m, 340 is output processing for determining that a flame has occurred, and the result is output, 350 is end processing, and 360 is 4 This is a process of discarding the oldest quarter data in the analysis section of the infrared intensity waveforms in the two wavelength ranges of 0.4 μm and 3.3 μm.
[0022]
Hereinafter, the operation of the flame detecting device according to the first embodiment of the present invention will be described with reference to FIGS. 1, 2 and 3.
[0023]
First, infrared rays corresponding to the surface temperature and the reflection of sunlight are emitted from the background infrared ray source 180 in a state where no fire has occurred. This background infrared light is transmitted through the protective sapphire glass 170 and is limited by the wavelengths passed by the infrared filters 100 and 105, respectively. The infrared light near 4.4 μm reaches the thermopile element 110 and the infrared light near 3.3 μm reaches the thermopile element 115. I do. The thermopile elements 110 and 115 generate electric signals corresponding to the intensity of the arriving infrared rays in the respective wavelength ranges, and the electric circuits are amplified by the amplifier circuits 120 and 125, respectively. The low-pass filters 130 and 135 limit the pass band of the respective electric signals, the sample-hold circuits 140 and 145 sample the respective electric signals, and the AD converters 150 and 155 calculate the infrared intensity in the respective wavelength ranges numerically. Become The CPU 160 performs a flame detection process on the quantified infrared intensity in these two wavelength ranges.
[0024]
Next, in the first embodiment of the present invention, the configuration of the flame detection system performed by the CPU 160 shown in FIG. 1 will be described with reference to the block diagram of FIG.
[0025]
First, in processes 200 and 205, the analysis waveforms having a predetermined number of samples and overlapping each other are analyzed with respect to the time waveforms of the infrared intensity in the 4.4 μm wavelength range and the 3.3 μm wavelength range. Data cutout processing is performed according to the window. Further, in steps 210 and 215, a Fourier transform process for transforming each time waveform into a complex spectrum is performed. In the first embodiment of the present invention, it is preferable to use a fast Fourier transform algorithm. Next, in processing 220, the intensity of the intensity change spectrum in the 4.4 μm wavelength region is compared with a predetermined threshold to detect whether an abnormal signal has occurred.
[0026]
On the other hand, in process 230, the phase difference between the complex spectra between the wavelength ranges of 4.4 μm and 3.3 μm is detected using the cross spectrum. In process 240, when an abnormal signal is detected by the abnormal signal detection 220 in the 4.4 μm wavelength region, the randomness of the phase difference is determined using the result of the phase difference detection 230 between 4.4 μm and 3.3 μm, and the position is determined. If the phase difference shows a wide distribution other than 0 degree, it is determined to be a flame and output.
[0027]
Next, a procedure in which the CPU 160 in FIG. 1 executes the processing blocks described in FIG. 2 will be described with reference to the flowchart in FIG. In FIG. 3, the process starts in a process 300. In the process 305, initialization such as setting of initial values of various variables is performed. In the process 310, a sample for a quarter time section of a predetermined analysis section is fetched from among the numerically-quantized infrared intensities in the respective wavelength ranges of 4.4 μm and 3.3 μm. In the process 315, the window function is multiplied over the length of a predetermined analysis section of the quantified infrared intensity in the respective wavelength ranges of 4.4 μm and 3.3 μm.
[0028]
The relationship between the analysis time interval and the window function will be described with reference to FIG.
[0029]
FIG. 4 is an explanatory diagram showing the relationship between the infrared intensity and the analysis window function of the flame detection system according to the first embodiment of the present invention, wherein the horizontal axis represents time and the vertical axis represents infrared intensity.
[0030]
In this figure, 400 is an infrared intensity signal of, for example, 4.4 μm, 410 is the window function of the first analysis section, 420 is the window function of the second analysis section, and 430 is the window function of the third analysis section. Each is shown. As described above, data is effectively used by superimposing and analyzing the analysis sections, and fluctuation phenomena can be completely analyzed. Here, the infrared fluctuation waveform in the 4.4 μm wavelength range is exemplified, but the analysis section and the window function at the same time synchronized with the 4.4 μm wavelength region are similarly used for the back ultraviolet fluctuation waveform in the 3.3 μm wavelength region. .
[0031]
The description returns to the processing 320 shown in FIG. In the process 320, fast Fourier transform is performed on each of the analysis sections synchronized with each other in the wavelength ranges of 4.4 μm and 3.3 μm as described above. In processing 325, the spectrum intensity in the 4.4 μm wavelength region is compared with the detection threshold, and if there is a component larger than the detection threshold (YES), the flow proceeds to processing 330.
[0032]
The comparison with the detection threshold in FIG. 3 will be described with reference to FIG.
[0033]
FIG. 5 is an explanatory diagram of a fluctuation spectrum of the infrared intensity change in the 4.4 μm wavelength range in each analysis section in the flame detection method according to the first embodiment of the present invention, in which the horizontal axis indicates the fluctuation frequency of the infrared intensity and the vertical axis indicates the fluctuation. The spectrum intensity and the depth direction indicate time (analysis section). In addition, each variable frequency component (plurality) of the infrared intensity in the 4.4 μm wavelength range in each analysis section is indicated by 500. Further, the detection threshold value can be represented by a plane having a slope such as 510 determined for each variation frequency.
[0034]
In the example shown in FIG. 3, in the first and second analysis sections, the state of the background infrared source 180 in which no fire has occurred in FIG. 1 and all the fluctuation frequency components are smaller than the detection threshold, If the result of determination in step 325 is NO, the process proceeds to step 360. In the third and fourth analysis sections, the state of the infrared light source 190 when a fire occurs in FIG. 1 and the fluctuation frequency component exceeds the threshold value is shown in FIG. If the result of determination in step 325 is YES, processing proceeds to step 330.
[0035]
The description returns to the processing 330 shown in FIG. In the process 330, the phase difference between the infrared fluctuation frequency components in the 4.4 μm wavelength region and the 3.3 μm wavelength region is totaled.
[0036]
Since the feature of the present invention lies in the flame judgment processing based on this phase difference, a detailed description will be given including a phenomenon inherent to the flame.
[0037]
First, FIG. 6 shows a change pattern of the infrared intensity waveform in the wavelength region of 4.4 μm and the wavelength region of 3.3 μm when a flame such as the infrared light source 190 occurs when the fire shown in FIG. 1 occurs.
[0038]
FIG. 6 is an explanatory diagram of the infrared intensity change time waveforms in the 4.4 μm wavelength region and the 3.3 μm wavelength region in the flame detection system according to the first embodiment of the present invention. The horizontal axis represents time, and the vertical axis represents each wavelength region. Infrared intensity.
[0039]
Intensity changes in the 4.4 μm and 3.3 μm wavelength regions until the ignition time show almost the same change. This is because the infrared source generally emits a wide range of infrared wavelengths mainly from sunlight or a heating element, so that changes in the infrared ray such as the intensity of the light beam and the rise and fall of the temperature of the heating element are in the 4.4 μm wavelength range. In most cases, the phase changes in the 3.3 μm wavelength region.
[0040]
On the other hand, the changes in the infrared intensity waveforms at 4.4 μm and 3.3 μm after the firing time are different from those before the firing, and have a property that their phases are shifted. This is because the infrared radiation in the 4.4 μm wavelength region is mainly due to the resonance phenomenon of carbon dioxide gas generated by combustion, whereas the infrared radiation in the 3.3 μm wavelength region is mainly due to heat radiation from combustion residues generated by incomplete combustion. Because it is due to. The intensity of the infrared radiation intensity of these infrared sources is caused by different phenomena, and is affected by the gas generation situation, the burning speed, the temperature of the gas and the combustion residue in different ways, etc. Generally, there is a time lag between the rise and fall of the infrared intensity.
[0041]
The present invention uses the characteristics of such a combustion phenomenon to determine whether the cause of the infrared rays is a flame after capturing the phenomenon of generating the infrared rays, by using the phase difference shift of the infrared intensity in different wavelength ranges. It is characterized by the following.
[0042]
FIG. 7 shows data expressed by a cross spectrum (complex number) about the phase difference between the infrared intensity of the 4.4 μm wavelength region and the 3.3 μm wavelength region of a disturbance and a flame that actually generate infrared rays.
[0043]
In FIG. 7, the angle is a phase angle of the cross spectrum, and the radial direction is a logarithmic scale display of the absolute value of the cross spectrum. 700 is the distribution of the cross spectrum of the 4.4 μm wavelength range and the 3.3 μm wavelength range when the incandescent lamp of 60 W is used as the infrared source instead of the flame (plotted without dividing the fluctuation frequency), and 710 is the actual flame ( FIG. 6 is a cross-spectrum distribution of the 4.4 μm wavelength range and the 3.3 μm wavelength range when the gasoline is burned in a fire plate as an infrared source (plotted without distinguishing the fluctuating frequency). As is clear from 700, in the case of a simple heat source such as an incandescent lamp, the fluctuations in the 4.4 μm wavelength region and the 3.3 μm wavelength region are almost synchronized, so that the phase of the cross spectrum is close to 0 regardless of the fluctuation frequency. .
[0044]
On the other hand, as shown in 710, in the case of a flame, the fluctuations in the 4.4 μm wavelength region and the 3.3 μm wavelength are often out of timing with each other, and this deviation is observed as the phase angle of the cross spectrum, and is independent of the fluctuation frequency. A component having a phase angle increases. It is a feature of the present invention that the flame and the disturbance infrared source other than the flame are distinguished by observing the timing shift of the variation of the infrared wavelength region of the flame different from each other.
[0045]
As described above, in the first embodiment, the flame and the disturbance infrared source other than the flame are distinguished by observing the timing shift of the fluctuation of the infrared wavelength region in the form of the difference in the phase angle of the cross spectrum.
[0046]
FIG. 8 illustrates a method of calculating the phase difference between the infrared fluctuation frequency components in the 4.4 μm wavelength region and the 3.3 μm wavelength region in the process 330 illustrated in FIG. In FIG. 8, the horizontal axis represents the phase component of the cross spectrum between the infrared intensity change in the 4.4 μm wavelength region and the infrared intensity change in the 3.3 μm wavelength region, and the vertical axis represents the frequency of the spectrum. 800 is a frequency distribution when the disturbance other than the flame is an infrared source, and 810 is a frequency distribution when the flame is an infrared source. In the counting method of the phase difference in the first embodiment, the ratio of the frequency of the component having a large phase difference as indicated by hatching is used as the counting result of the phase difference. That is, assuming that the phase difference evaluation amount is β,
β = N ₀ / N _all … (1)
Where N _all Is the total number of discrete frequency components in the time interval of interest, N ₀ Indicates the number of discrete frequency components having a phase difference of the cross spectrum greater than a predetermined specific value.
[0047]
The phase difference evaluation amount β will be described with reference to FIG. 8, and is a numeral indicating a ratio of an area of a hatched portion where the absolute value of the phase angle is larger than a certain angle θ in the entire area of the frequency distribution. β ranges from 0 to 1, and the larger the value, the greater the proportion of components having a large phase difference, and it can be said that the value is more flame-like.
[0048]
The description returns to the processing of the processing 330 shown in FIG. In the process 330, the phase difference evaluation amount β is totaled. In the process 335, it is determined whether or not the component having a large phase difference is large based on whether or not the phase difference evaluation amount β is larger than a predetermined threshold value. In the process 360, the oldest 1/4 fluctuation waveform data in the time section is discarded, and the process returns to the process 310 to acquire a new sample. On the other hand, in the process 340, an output process in the case where it is determined that the flame has occurred is performed, and the process ends in 350.
[0049]
The effect of the first embodiment of the present invention described above will be described.
[0050]
In the first embodiment of the present invention, a detection method similar to a normal infrared flame detection method is employed, which first detects an increase in the intensity or change in intensity of infrared light. Further, it is determined whether or not the cause of such an infrared phenomenon is caused by a flame, using a characteristic feature of the flame that a time lag occurs in changes in infrared intensity in different wavelength ranges. Infrared phenomena other than flame and infrared phenomena due to flame are reliably determined, and a flame detection output is performed only in the case of a flame.
[0051]
Therefore, in the case of detecting a flame with the configuration of the first embodiment, an infrared phenomenon that is not a flame as in the prior art, for example, a fluctuation input of an infrared ray due to the passage of a high-temperature object, etc. This eliminates the possibility of erroneous detection, such as detecting that there is, and provides a flame detection device having a feature of detecting only flame.
[0052]
Next, a second embodiment of the present invention will be described.
[0053]
FIG. 9 is a block diagram of a flame detecting device according to a second embodiment of the present invention, and FIG. 10 is a diagram showing the behavior of each signal of the flame detecting device according to the second embodiment of the present invention.
[0054]
In FIG. 9, reference numeral 900 denotes an infrared filter that passes a wavelength range around 4.4 μm as a first infrared wavelength range used for flame detection among infrared rays radiated by a flame, and reference numeral 905 denotes a flame detection among infrared rays radiated by a flame. An infrared filter that passes a wavelength range around 3.3 μm as a second infrared wavelength to be used, 910 is a thermopile element that converts infrared intensity in the first wavelength range into voltage, and 915 is infrared ray intensity in the second wavelength range. A thermopile element for converting into a voltage; 920, an amplifier circuit (Amp) for amplifying the infrared intensity voltage in the first wavelength range; 925, an amplifier circuit (Amp) for amplifying the infrared intensity voltage in the second wavelength range; A band-pass filter (BPF) 932 that passes a component of a predetermined first fluctuating frequency band among the changes of the infrared intensity in the first wavelength band is a first wavelength 932. Among the changes in the infrared intensity, a band-pass filter (BPF) 935 that passes a component of a second variable frequency band that is predetermined is a first one of the changes in the infrared intensity in the second wavelength band. A band-pass filter (BPF) 937 that passes a component of a variable frequency band, 937 is a band-pass filter (BPF) that passes a component of a predetermined second variable frequency band among the changes in the infrared intensity in the second wavelength range. , 940 is a voltage comparator (Comp) for creating a zero-crossing waveform of the first variation frequency band of the infrared intensity change in the first wavelength range, and 942 is a second comparator of the infrared intensity change in the first wavelength range. A voltage comparator (Comp) for generating a zero-crossing waveform in a variable frequency band, and a voltage ratio for generating a zero-crossing waveform in a first variable frequency band of the infrared intensity change in the second wavelength range; A comparator (Comp) 947 is a voltage comparator (Comp) for creating a zero-crossing waveform of a second fluctuation frequency band of the infrared intensity change in the second wavelength range, and 950 is an infrared intensity change in the first wavelength range. An exclusive OR circuit 952 for detecting a phase difference signal between the first fluctuation frequency band of the first variation frequency band and the first variation frequency band of the infrared intensity change in the second wavelength band. An exclusive OR circuit 955 for detecting a phase difference signal between a second variable frequency band of the infrared intensity change and a second variable frequency band of the infrared intensity change in the second wavelength range; A low-pass filter (LPF) for smoothing the phase difference signal in the variable frequency band, 957 is a low-pass filter (LPF) for smoothing the phase difference signal in the second variable frequency band, and 960 is each variable frequency Band phase difference signal Threshold voltage V for the calculation adder for (Mix), 965 is a predetermined phase difference signal the sum of the phase difference signal for each fluctuation frequency band ₂ 970 is a low-pass filter (LPF) for extracting the entire variable frequency component around 4.4 μm which is the first infrared wavelength range, and 972 is a low-pass filter for the first infrared wavelength range. A rectifier circuit for obtaining a power signal of the entire variable frequency component, 975 is a low-pass filter (LPF) for smoothing a power signal of the entire variable frequency component of the first infrared wavelength range, and 980 is a first infrared wavelength range. And the predetermined threshold voltage V ₁ A comparator (Comp) 985 for comparing the first infrared wavelength range and the second back ultraviolet wavelength with the result of the abnormality detection of the infrared variation based on the power signal of the entire variable frequency component in the first infrared wavelength range. AND circuit for determining whether the flame is determined to be a flame based on the phase difference of the fluctuation component with the area, 990 is a flame determination output, 995 is an infrared source when a fire has occurred, and 998 is an infrared source. A protective sapphire glass that protects the infrared filter 900 in the first wavelength band and the infrared filter 905 in the second wavelength band from contamination and impact from the outside and transmits infrared rays is shown.
[0055]
Next, the operation of the flame detection device according to the second embodiment of the present invention will be described with reference to FIGS.
[0056]
First, infrared light from outside the flame detector passes through the protective sapphire glass 998, and the wavelengths of light passing therethrough are limited by the infrared filters 900 and 905, and infrared light near 4.4 μm is transmitted to the thermopile element 910 at around 3.3 μm. Infrared rays reach thermopile element 915. The thermopile elements 910 and 915 generate electric signals corresponding to the intensity of the infrared rays that arrive in the respective wavelength ranges, and amplify the electric signals by the amplifier circuits 920 and 925, respectively. Bandpass filters 930, 932 and 935, 937 limit the fluctuating frequency of each of these electrical signals. Note that the bandpass filters 930 and 935 and 932 and 937 corresponding to the respective wavelength ranges use filters having a common pass band and a phase characteristic with respect to the variable frequency. The voltage comparators 940, 942, 945, and 947 detect the zero-crossing points of the signals whose fluctuation frequencies are restricted, respectively. Further, the shift of the zero crossing point between the wavelength ranges at the respective variable frequencies is detected by the exclusive OR of the exclusive OR circuits 950 and 952, and smoothed by the low-pass filters 955 and 957.
[0057]
Detection of such a phase shift at a fluctuation frequency whose band is limited between wavelength ranges will be described with reference to FIG.
[0058]
FIG. 10 is a diagram for explaining the behavior of each signal in a case other than a flame and a case of a flame in the block operation of the band-pass filter 930 to the low-pass filter 957 in FIG. 9, and FIG. FIG. 10B is a diagram showing the behavior, and FIG. 10B is a diagram showing the behavior of each signal with respect to the flame.
[0059]
In these figures, reference numeral 1000 denotes an intensity change signal in which the fluctuation frequency of the infrared light source other than the flame in the 4.4 μm wavelength band is band-limited, for example, the output of the band-pass filter 930 shown in FIG. For example, the output 1010 of the voltage comparator 940 shown in FIG. 9 is an intensity change signal in which the fluctuation frequency of the infrared light source other than the flame in the 3.3 μm wavelength range is band-limited, for example, the bandpass filter 935 shown in FIG. The output 1015 is a signal binarized by the zero-crossing, for example, the output of the voltage comparator 945 shown in FIG. 9, and the reference numeral 1020 is an exclusive OR circuit between the zero-crossings of the band-limited signals in the two wavelength ranges. The output of the exclusive OR circuit 950 shown in FIG. 10 is a smoothed signal, and the output of the low-pass filter 955 shown in FIG. For example, the output of the bandpass filter 930 shown in FIG. 9 is an intensity change signal in which the fluctuation frequency of the 4.4 μm wavelength range is band-limited, and the signal 1035 is a signal binarized by the zero crossing. An output 940, 1040 is an intensity change signal of the infrared source other than the flame in which the fluctuation frequency in the 3.3 μm wavelength range is band-limited. For example, an output of the band-pass filter 935 shown in FIG. For example, the output of the voltage comparator 945 shown in FIG. 9 is an exclusive OR circuit between the zero crossings of the band-limited signals in the two wavelength ranges, and the output of the exclusive OR circuit 950 shown in FIG. Reference numeral 1055 denotes an output of the low-pass filter 955 shown in FIG.
[0060]
First, as shown in FIG. 10A, in the case of an infrared light source that is not a flame, the fluctuation components in the 4.4 μm wavelength region and the 3.3 wavelength region generally show similar fluctuation waveforms. In such a case, the signals in the corresponding variable frequency bands of each wavelength range fluctuate almost synchronously (see 1000 and 1005 in FIG. 10). Therefore, the outputs (eg, 1010 and 1015 of FIG. 10) of the voltage comparators 940 and 945 shown in FIG. 9, which are the zero-crossing binarized signals, are also substantially synchronized.
[0061]
Therefore, the output of the exclusive OR circuit 950 in FIG. 9, which is the exclusive OR signal, is often zero, and there are many pulses with a small time width (1020 in FIG. 10). The output obtained by smoothing this with, for example, a low-pass filter 955 shown in FIG. 9 changes at a low level (1025 in FIG. 10).
[0062]
On the other hand, as shown in FIG. 10B, when the flame is an infrared light source, the fluctuation components of the 4.4 μm wavelength region and the 3.3 μm wavelength region generally have temporally shifted fluctuation waveforms. In such a case, the signal in the corresponding variable frequency band of each wavelength range fluctuates back and forth in time (1030 and 1035 in FIG. 10).
[0063]
Therefore, the outputs of the voltage comparators 940 and 945 (1040 and 1045 in FIG. 10) shown in FIG.
[0064]
Therefore, in the output of the exclusive OR circuit 950 shown in FIG. 9 which is the exclusive OR signal, there are many pulses having a certain time width (1050 in FIG. 10).
[0065]
The output obtained by smoothing this with, for example, a low-pass filter 955 shown in FIG. 9 changes at a relatively high level (1055 in FIG. 10).
[0066]
In the second embodiment of the present invention, the detection of the time lag of the signal as described above is performed in a plurality of bands having different variable frequency bands. Further, the outputs indicating the phase shifts of the plurality of variable frequency bands are added by the adder 960. The output of the adder 960 is a threshold voltage V ₂ By comparing with, it is determined whether or not it is flame-like.
[0067]
On the other hand, in parallel with this, an infrared source is detected by a change in infrared intensity using the fluctuation component in the 4.4 μm wavelength range. This processing detects a normal abnormal infrared light source and is not essential in the present invention, so that a simple configuration is used. A high-pass filter 970 shown in FIG. 9 extracts only a fluctuation component of the infrared intensity, and a rectification circuit 972 rectifies this. Further, the low-pass filter 975 smoothes the rectified waveform. Further, in the comparator 980 shown in FIG. ₁ Compare with
[0068]
The result of detection of the infrared fluctuation due to the intensity change is output from the comparator 980, and the result of the deviation of the fluctuation between the 4.4 μm wavelength area, the 3.3 μm wavelength area, and the 3.3 μm wavelength area is calculated by the comparator 965. Is output from the AND circuit 985, it is determined that the flame has been detected, and the flame detection result is output to the flame detection output 990.
[0069]
The effects of the second embodiment of the present invention described above will be described.
[0070]
In the second embodiment of the present invention, a detection method similar to a normal infrared flame detection method, in which an increase in the intensity or a change in the intensity of infrared light is first detected. Furthermore, it is determined whether or not the cause of such an infrared phenomenon is caused by a flame, using a flame-specific feature that causes a time lag in changes in infrared intensity in different wavelength ranges. The line phenomenon other than the flame and the infrared phenomenon due to the flame are reliably determined, and the flame detection output is performed only when the flame is detected.
[0071]
Therefore, in the case of detecting a flame with the configuration of the second embodiment, the flame is excessively applied to an infrared phenomenon that is not a flame as in the related art, for example, a fluctuation input of infrared due to passage of a high-temperature object. This eliminates the possibility of erroneous detection of detecting a flame, and provides a flame detection device having a feature of detecting only a flame.
[0072]
In the second embodiment of the present invention, since the phase difference of the infrared intensity change in two different wavelength ranges is realized by an analog circuit, the circuit scale is smaller than that of the first embodiment using a CPU. It has the advantage of lower costs.
[0073]
Further, a usage mode of the present invention will be described.
[0074]
In the first and second embodiments of the present invention, the detection of the occurrence of an abnormal infrared light source based on the infrared intensity is performed, but any method may be used for this detection. The present invention is to determine with high accuracy whether or not such an abnormal infrared light source is caused by a flame.
[0075]
In the embodiment of the present invention, the combination of wavelengths is not limited to the combination of 4.4 μm and 3.3 μm, but may be any combination. In the present invention, since a property in which a chemical or physical phenomenon in a flame occurs with a time difference is used, a set of wavelengths for observing this time difference is also a combination that can observe the time difference between these chemical or physical phenomena. If any, it is optional. For example, it can be easily analogized to use visible light or ultraviolet light other than infrared light depending on the application.
[0076]
In the embodiment of the present invention, a combination of two wavelength ranges is used. However, it can be easily analogized that the accuracy of determination of a flame and other phenomena is improved by using the third and fourth wavelength ranges. For example, the accuracy of the determination can be improved by using the visible light and ultraviolet wavelength ranges in addition to the two wavelength ranges used in the embodiment of the present invention.
[0077]
In the embodiment of the present invention, the infrared intensity of the entire flame is used, but the flame is captured as an image in a different wavelength range, and the flame is determined using the present invention for a part or all of the image of the flame. It can be easily analogized. For example, a method of using a silicon CCD image sensor having sensitivity to infrared rays as a light receiving element and determining whether or not there is a flame in each image capturing area can be easily analogized.
[0078]
Next, a third embodiment of the present invention will be described.
[0079]
FIG. 11 is a configuration diagram of a flame detection device according to a third embodiment of the present invention. FIG. 12 is an operation flowchart of the detection device according to the third embodiment of the present invention, and FIG. 13 is a processing block diagram of the CPU.
[0080]
In FIG. 11, reference numeral 2100 denotes an infrared filter which passes a wavelength range around 4.4 μm as an infrared wavelength range used for flame detection among infrared rays radiated by a flame, 2110 denotes a thermopile element which converts infrared intensity into a voltage, and 2120 denotes infrared rays. An amplification circuit (Amp) for amplifying the intensity voltage, a low-pass filter (LPF) 2130 for limiting a frequency component of a change in infrared intensity, a sample-and-hold circuit (S / H) 2140 for holding an infrared intensity voltage, and a 2150 infrared intensity Converter 2160 is a central processing unit (CPU) for detecting and judging a flame based on a change in the intensity of the infrared rays, and 2170 protects the infrared filter 2100 from contamination and impact from the outside, and Protective sapphire glass, 2180, is in a normal fire-free state. Background infrared source that, 2190 shows the infrared source at the time of a fire has occurred.
[0081]
FIG. 12 is a flowchart of the flame detection processing according to the third embodiment of the present invention.
[0082]
In this figure, 2200 is the start of the process, 2210 is the setting of the initial value, 2220 is the input of the infrared peak value, 2230 is the comparison of the peak value with the detection threshold, 2240 is the update of the average value of the infrared peak value, and 2250 is the average deviation , 2260 is the flame detection alerting process, and 2270 is the end.
[0083]
FIG. 13 is a block diagram of processing inside the CPU according to the third embodiment of the present invention.
[0084]
In this figure, 2300 is an infrared intensity input, 2310 is an average estimation of infrared intensity, 2320 is an average value storage, 2330 is an average deviation estimation, 2340 is an average deviation storage, 2350 is a threshold setting, 2360 is a level judgment, and 2370 is a judgment. Output.
[0085]
Next, the operation of the flame detector according to the third embodiment of the present invention will be described with reference to FIG.
[0086]
First, infrared rays corresponding to the surface temperature and the reflection of sunlight are emitted from the background infrared ray source 2180 in a normal state where no fire has occurred. The background infrared light passes through the protective sapphire glass 2170, and the wavelength of the light passing therethrough is restricted by the infrared filter 2100. The infrared light near 4.4 μm reaches the thermopile element 2110. The thermopile element 2110 generates an electric signal corresponding to the intensity of the arriving infrared light, and amplifies the electric signal by the amplifier circuit 2120. The low-pass filter 2130 limits the pass band of the electric signal, and the sample-and-hold circuit 2140 samples the electric signal and digitizes the infrared intensity by the AD converter 2150. The CPU 2160 performs a flame detection process on the digitized infrared intensity.
[0087]
Next, in the third embodiment of the present invention, processing performed by the CPU 2160 in FIG. 11 will be described with reference to the flowchart in FIG. 12 and FIG.
[0088]
First, the process is started at the start 2200 of the process. In the initial value setting 2210, various parameters at the start of the average value storage 2320 and the average deviation storage 2340 in FIG. 13 are initialized. In the third embodiment, various parameters are initialized by the following equations.
[0089]
(Equation 1)

[0090]
(Equation 2)

[0091]

[0092]
Further, an initial detection threshold value is set in advance in the threshold value setting 2350 in FIG.
[0093]
[Equation 3]

[0094]
Where b _{N + 1} Is a detection threshold used for the (N + 1) th sample time, and the constant α is a safety coefficient.

[0095]
Here, the i-th (i is N + 1, N + 2,...) Infrared peak value x _i Is obtained. At this time, the detection threshold b _i (I is N + 1, N + 2,...), The peak value X in the level determination 2360 in FIG. _i And detection threshold b _i The comparison 2230 of FIG.
[0096]
If the peak value x _i Is the detection threshold b _i If it is smaller, the process proceeds to the update 2250 of the average deviation of the infrared peak values.

[0097]
(Equation 4)

[0098]
Where K _x Is a coefficient for estimating the average value, and a constant larger than 1 is set.

[0099]
(Equation 5)

[0100]
Where K _d Is a coefficient for estimating the average deviation, and a constant larger than 1 is set.

[0101]
(Equation 6)

[0102]
In this way, a new detection range is calculated, and the process returns to the input 2220 of the infrared peak value.

[0103]
The effects of the third embodiment of the present invention described above will be described with reference to FIGS.
[0104]
In FIG. 14, reference numeral 2400 denotes a waveform of background infrared fluctuation, 2410 denotes an estimated average value, 2420 denotes an estimated average deviation, and 2430 denotes a detection threshold.
[0105]
15A shows the background infrared detection output immediately after cleaning the infrared detection window, and FIG. 15B shows the background infrared detection output after the contamination of the infrared detection window. 2510 shows a waveform of the background infrared fluctuation observed when there is no image, and 2510 shows a waveform of the background infrared fluctuation observed when the infrared detection window is contaminated.
[0106]
According to the method shown in the third embodiment of the present invention, the estimated average value of the infrared intensity is obtained from the waveform 2400 of the background infrared fluctuation from the background infrared source 2180 (see FIG. 11) in a normal state where no fire occurs. 2410 can be obtained. Also, an estimated average deviation 2420 of the infrared intensity with respect to the estimated average value 2410 can be estimated. Since the detection threshold value 2430 for disaster detection is determined based on the estimated average value 2410 and the estimated average deviation 2420 of the infrared intensity, the flame is detected with the optimum detection threshold value 2430 without being affected by the intensity of the background infrared ray in the use environment. be able to.
[0107]
In addition, when there is no contamination in the infrared detection window as shown in FIG. 15, the detection threshold value for fire detection is set relatively high from the waveform of the background infrared fluctuation observed at this time. If the infrared detection window is contaminated, the waveform of the observed background infrared fluctuation becomes smaller, so that the detection threshold for fire detection automatically lowers. For this reason, even if the infrared detection window becomes contaminated, a fire is detected with an optimum detection threshold.
[0108]
Next, a fourth embodiment of the present invention will be described.
[0109]
In the fourth embodiment of the present invention, the configuration of the flame detection device is the same as that of the third embodiment of the present invention, and therefore will be described with reference to FIG.
[0110]
FIG. 16 is a flowchart of the flame detecting apparatus according to the fourth embodiment of the present invention.
[0111]
In this figure, 2600 is the start of the processing, 2610 is the setting of the initial value, 2620 is the input and buffering of the infrared peak value, 2630 is the time series Fourier transform of the infrared peak value, and 2640 is the time series spectrum of the infrared peak value 2650 is the update of the average value of the spectrum, 2660 is the update of the average deviation of the spectrum, 2670 is the fire detection alerting process, and 2680 is the end.
[0112]
FIG. 17 is a block diagram showing processing inside the CPU according to the fourth embodiment of the present invention.
[0113]
In this figure, 2700 is an infrared intensity input, 2710 is a waveform buffer process, 2720 is a Fourier transform, 2730 is an average estimation of the infrared intensity for each frequency, 2740 is an average value storage for each frequency, and 2750 is an average value storage for each frequency. Average deviation estimation, 2760 is an average deviation storage for each frequency, 2770 is a threshold setting for each frequency, 2780 is a level determination for each frequency, and 2790 is a determination output.
[0114]
Next, the operation of the flame detector according to the fourth embodiment of the present invention will be described with reference to FIG.
[0115]
First, infrared rays corresponding to the surface temperature and the reflection of sunlight are emitted from the background infrared ray source 2180 in a normal state where no fire has occurred. The background infrared light passes through the protective sapphire glass 2170, and the wavelength of the light passing therethrough is restricted by the infrared filter 2100. The infrared light near 4.4 μm reaches the thermopile element 2110. The thermopile element 2110 generates an electric signal according to the intensity of the arriving infrared light, and amplifies the electric vibration by the amplifier circuit 2120. The low-pass filter 2130 limits the pass band of the electric signal, and the sample-and-hold circuit 2140 samples the electric signal and digitizes the infrared intensity by the AD converter 2150. The CPU 2160 performs a flame detection process on the digitized infrared intensity.
[0116]
Next, processing performed by the CPU 2160 in FIG. 11 in the fourth embodiment of the present invention will be described with reference to FIGS.
[0117]
First, the process starts at a start 2600 of the process. In the initial value setting 2610, various parameters stored in each frequency average value storage 2740 and each frequency average deviation storage 2760 at the start are initialized. In the fourth embodiment, in the waveform buffer processing 2710, the infrared intensity signal is divided into analysis sections each having a predetermined time section length, and the Fourier transform 2720 performs processing using the discrete Fourier transform spectrum intensity. In the fourth embodiment, various parameters are initialized by the following equations.
[0118]
(Equation 7)

[0119]
(Equation 8)

[0120]

[0121]
Further, an initial detection threshold is set in advance in each frequency threshold setting 2770 from these estimated values by the following equation.
[0122]
(Equation 9)

[0123]
Where B (ω) _{M + 1} Is a detection threshold used for the fluctuation frequency ω in the (M + 1) th analysis section, and the constant α is a safety coefficient.

[0124]
In the peak value input and buffering 2620, a peak value of the number of samples corresponding to a predetermined time width of the analysis section is input and stored in an internal buffer of the CPU 2160.
[0125]
Next, a discrete Fourier transform is performed by a Fourier transform 2630 in a Fourier transform 2720 in FIG. Preferably, the FFT algorithm is used. Here, the infrared fluctuation spectrum X (ω) of the fluctuation frequency ω in the j-th (j = M + 1, M + 2,. _j Is obtained. At this time, the detection threshold B (ω) has already been obtained. _j (J = M + 1, M + 2,...) Are obtained by the respective frequency threshold settings 2770 in FIG. 17, so that the absolute value | X (ω) of the infrared fluctuation spectrum _j | And detection threshold B (ω) _j Are compared in each frequency level judgment 2780 in FIG.
[0126]
If the absolute value of the infrared fluctuation spectrum | X (ω) _j | Is the detection threshold B (ω) _j If it is smaller, the process proceeds to the update of the spectrum average value 2650.

[0127]
(Equation 10)

[0128]
Where K _x Is a coefficient for estimating the average value, and a constant larger than 1 is set. The result is stored in each frequency average value storage 2740 in FIG.

[0129]
[Equation 11]

[0130]
Where K _d Is a coefficient for estimating the average deviation, and a constant larger than 1 is set. The result is stored in each frequency average deviation storage 2760 in FIG.

[0131]
(Equation 12)

[0132]
Thus, a new detection threshold is calculated, and the process returns to the peak value input and buffering 2620.
[0133]
On the other hand, in FIG. 11, when the infrared source 2190 occurs when a fire occurs, the detection threshold value B (ω) set by the estimated average value and the estimated average deviation of the normal spectrum is used. _{j + 1} Higher spectra are observed. In such a case, the absolute value of the spectrum in FIG. 16 | X (ω) _j | And detection threshold B (ω) _j As a result of the comparison 2640 of each frequency level judgment 2780 in FIG. 17, if the spectrum is larger, the process proceeds to the flame detection alerting process 2670, the output when detected by the judgment output 2790 in FIG. It becomes.
[0134]
The effect of the fourth embodiment of the present invention described above will be described.
[0135]
In the fourth embodiment of the present invention, the change of the infrared peak value is decomposed into discrete frequency components by performing a discrete Fourier transform. For this reason, for example, a disturbance such as a rotating lamp that exhibits a periodic infrared fluctuation is observed as a steady spectrum intensity, and is included in the threshold value in advance as a spectrum average value.
[0136]
In general, a change in the infrared intensity of a flame has an irregularly varying frequency component exhibiting a 1 / f characteristic, and a frequency which is not affected by a periodic disturbance by using a frequency component other than a constant frequency due to the disturbance. Flame can be detected by the rise of the component. Therefore, in the third embodiment, when a periodic disturbance such as a rotating light is always present, there is a problem that the threshold value for the peak value is increased and the detection becomes difficult. If frequency components are used as in the example, even if there is a periodic disturbance, the influence on the detection performance is small.
[0137]
Further, by automatically determining the detection threshold value of the flame detection in the third embodiment, the effect of detecting the flame at the optimum detection threshold value without being affected by the intensity of the background infrared ray in the use environment is different from that of the fourth embodiment. It is similarly effective in the example.
[0138]
Further, even in the case where the infrared detection window is contaminated in the third embodiment, the effect of detecting the flame at the optimum detection threshold by lowering the detection threshold according to the contaminant is similarly obtained in the fourth embodiment. It is valid.
[0139]
In the third embodiment of the present invention, the threshold value for flame detection is set by using the peak value of the background infrared intensity, and in the fourth embodiment, the threshold value for flame detection is set by using as the spectral intensity feature value of the background infrared ray. It is not necessary to use it, and it is effective to use a feature amount such as a moving average value of a peak value, a filter bank output, and a linear prediction coefficient.
[0140]
In the third and fourth embodiments of the present invention, the detection threshold is determined from the average value of the characteristic amounts such as the peak value and the spectrum and the average deviation, but it is not always necessary to determine the detection threshold using these statistics. Instead, it can be easily analogized to determine the detection threshold using various statistics such as an average value expressing a normal distribution, an X-squared distribution, and the like.
[0141]
In the third and fourth embodiments of the present invention, in estimating the average value and the average deviation of the feature values such as the peak value and the spectrum, the average value and the average deviation are estimated by a simple low-pass filter. It is not necessary to use a simple estimation method, and it is easy to analogize to adopt a non-linear estimation method using a neural network, a singular value discrimination removal method, or the like.
[0142]
In the third and fourth embodiments of the present invention, flame detection is determined by a simple comparison between a feature value such as a peak value or a spectrum and a detection threshold. However, it is not always necessary to perform such a simple detection determination. It is easy to analogize using the various determination methods used in the voice section detection method, such as the determination based on the continuity of the signal detected as peculiar and the use of parameters using weighting for the spectrum. it can.
[0143]
In the third and fourth embodiments of the present invention, among the infrared rays radiated by the flame, an infrared filter that passes a wavelength range around 4.4 μm is used as an infrared wavelength range used for flame detection. The present invention is not required to be used, and the present invention is effective even when using other infrared wavelength ranges, visible light, ultraviolet light, or the like according to the purpose of use.
[0144]
FIG. 18 is a block diagram of a flame detection system according to a fifth embodiment of the present invention, and is a block diagram of a flame detection system configured by combining the first to fourth embodiments.
[0145]
In FIG. 18, reference numeral 2800 denotes a process of cutting out window data by dividing into a time (analysis) section in which infrared fluctuation waveforms of the first 4.4 μm wavelength region are superimposed on each other, and 2805 denotes infrared light of a second 3.3 μm wavelength region. A process of cutting out window data by dividing the variation waveform into analysis (time) sections superimposed on each other, 2810 denotes a Fourier transform process of the variation waveform in the first wavelength range, and 2815 denotes a Fourier transform of the variation waveform in the second wavelength range. Processing, 2820 is the average value estimation processing after the Fourier transform processing, 2825 is the average deviation estimation processing after the Fourier transformation processing, 2830 is the average value storage processing, 2835 is the average deviation storage processing, 2840 is the average value and the average deviation. 2850 is a level determination process based on a threshold, and 2860 is a fluctuation spectrum of a first 4.4 μm wavelength region and a second wavelength region. A process of detecting a phase difference with a 3.3 μm fluctuation spectrum; a process 2870 for detecting the phase difference; a phase difference randomness determining process for determining the randomness of the phase difference using the phase difference detected in the process 2860; A determination output process for obtaining a determination output based on the phase difference randomness determination and the level determination based on the threshold.
[0146]
Ao, the present invention is not limited to the above embodiments, and various modifications are possible based on the spirit of the present invention, and these are not excluded from the scope of the present invention.
[0147]
【The invention's effect】
As described above, according to the present invention, the following effects can be obtained.
[0148]
(A) It is possible to accurately detect a true flame without erroneously detecting an infrared phenomenon that is not a flame as a flame.
[0149]
(B) First, a detection method similar to a normal infrared flame detection method is used, which detects an increase in the intensity or change in intensity of infrared light. Further, it is determined whether or not the cause of such an infrared phenomenon is caused by a flame, using a characteristic feature of the flame that a time lag occurs in changes in infrared intensity in different wavelength ranges. Infrared phenomena other than flame and infrared phenomena due to flame are reliably determined, and a flame detection output is performed only in the case of a flame.
[0150]
(C) First, a detection method similar to a normal infrared flame detection method is used, which detects an increase in the intensity or change in intensity of infrared light. Furthermore, it is determined whether or not the cause of such an infrared phenomenon is caused by a flame, using a flame-specific feature that causes a time lag in changes in infrared intensity in different wavelength ranges. The line phenomenon other than the flame and the infrared phenomenon due to the flame are reliably determined, and the flame detection output is performed only when the flame is detected.
[0151]
(D) It is possible to provide a flame detecting device and a flame detecting method which detect a flame at an optimal threshold value without being affected by the intensity of background infrared rays in the use environment or the contamination state of the infrared ray detection window.
[Brief description of the drawings]
FIG. 1 is a configuration diagram of a flame detection device according to a first embodiment of the present invention.
FIG. 2 is a configuration diagram of a flame detection system showing a first embodiment of the present invention.
FIG. 3 is a flowchart of a flame detection process according to the first embodiment of the present invention.
FIG. 4 is an explanatory diagram illustrating a relationship between an infrared intensity and an analysis window function of the flame detection system according to the first embodiment of the present invention.
FIG. 5 is an explanatory diagram of a fluctuation spectrum in each analysis section of a change in infrared intensity in a 4.4 μm wavelength region in the flame detection system according to the first embodiment of the present invention.
FIG. 6 is an explanatory diagram of infrared intensity change time waveforms in a 4.4 μm wavelength region and a 3.3 μm wavelength region in the flame detection system according to the first embodiment of the present invention.
FIG. 7 is a view showing data expressed by a cross spectrum (complex number) with respect to a phase difference between an infrared intensity of a 4.4 μm wavelength region and a 3.3 μm wavelength region of a disturbance and a flame which actually generate infrared rays.
8 is an explanatory diagram of a method for summing up a phase difference between an infrared fluctuation frequency component in a 4.4 μm wavelength region and a 3.3 μm wavelength region in a process 330 illustrated in FIG. 3;
FIG. 9 is a configuration diagram of a flame detection device according to a second embodiment of the present invention.
FIG. 10 is a diagram showing the behavior of each signal of the flame detection device according to the second embodiment of the present invention.
FIG. 11 is a configuration diagram of a flame detection device according to a third embodiment of the present invention.
FIG. 12 is an operation flowchart of a flame detection device according to a third embodiment of the present invention.
FIG. 13 is a processing block diagram inside a CPU of a flame detection device according to a third embodiment of the present invention.
FIG. 14 is an explanatory diagram of detection threshold calculation in a flame detection method according to a third embodiment of the present invention.
FIG. 15 is an explanatory diagram of a phenomenon at the time of contamination of an infrared detection window in a flame detection system according to a third embodiment of the present invention.
FIG. 16 is an operation flowchart of a flame detection device according to a fourth embodiment of the present invention.
FIG. 17 is a processing block diagram inside a CPU of a flame detection device according to a fourth embodiment of the present invention.
FIG. 18 is a configuration diagram of a flame detection system according to a fifth embodiment of the present invention.
[Explanation of symbols]
100, 105, 900, 905, 2100 Infrared filter
110, 115, 910, 915, 2110 Thermopile element
120, 125, 920, 925, 2120 Amplifier circuit (Amp)
130, 135, 955, 957, 970, 980, 2130 Low-pass filter (LPF)
140, 145, 2140 Sample hold circuit (S / H)
150, 155, 2150 AD converter (AD)
160, 2160 Central processing unit (CPU)
170,998,2170 Protected sapphire glass
180,2180 Background infrared source in normal condition
190,995,2190 Infrared source in case of fire
200, 205, 210, 215, 220, 230, 240 processing (FIG. 2) 300, 305, 310, 315, 320, 325, 330, 335, 340, 350, 360 processing (FIG. 3)
400 infrared intensity signal
410 Window function of first analysis interval
420 Window function of second analysis interval
430 window function of third analysis interval
500 Varying frequency components of infrared intensity
510 Plane with slope
Distribution of cross spectrum when 700 60W incandescent lamp is used as infrared source
710 Cross spectrum distribution when using actual flame as infrared source
930,932,935,937 Band Pass Filter (BPL)
940,942,945,947 Voltage comparator (Comp)
950,952 Exclusive OR circuit
960 adder (Mix)
965,980 Comparator (Comp)
970 High-pass filter (HPF)
972 Rectifier circuit
985 AND circuit
990 Flame judgment output
998 Protected sapphire glass
1000, 1010, 1030, 1040 Bandpass filter output
1005,1015,1035,1045 Output of voltage comparator
1020, 1050 Output of exclusive OR circuit
1025, 1055 Low-pass filter output
2200, 2210, 2220, 2230, 2240, 2250, 2260, 2270 Processing (FIG. 12)
2300, 2310, 2320, 2330, 2340, 2350, 2360, 2370 Processing (FIG. 13)
2400 Background infrared fluctuation waveform
2410 Estimated average value
2420 Estimated mean deviation
2430 Detection threshold
2500 Waveform of background infrared fluctuation observed when there is no contamination on the infrared detection window
2510 Waveform of background infrared fluctuation observed when there is contamination in the infrared detection window
2600, 2610, 2620, 2630, 2640, 2660, 2670, 2680 Processing (FIG. 16)
2700, 2710, 2720, 2730, 2740, 2750, 2760, 2770, 2780, 2790 Processing (FIG. 17)
2800, 2805, 2810, 2815, 2820, 2825, 2830, 2835, 2840, 2850, 2860, 2870, 2880 Processing (FIG. 18)

Claims (10)

In an infrared type flame detection device that detects a flame by observing the intensity of infrared rays emitted from the flame,
(A) means for measuring a time lag of output between infrared detectors having different wavelength sensitivities, and (b) means for determining flame as a result of the output from the means. Flame detector. By measuring the time lag between the infrared intensity change near the wavelength of 4.4 μm mainly caused by the carbon dioxide resonance phenomenon at the time of combustion and the wavelength near 3.0 μm mainly caused by the high-temperature combustion residue, the flame is measured. 2. The infrared flame detecting apparatus according to claim 1, wherein a phenomenon other than the flame and the flame is determined. 3. The infrared flame detecting device according to claim 1, wherein a phase difference of the infrared intensity change is measured when the increase in the infrared intensity is detected. 3. The infrared flame detecting device according to claim 1, wherein a phase difference of the infrared intensity change is measured when an increase in the infrared intensity change width is detected. The infrared flame detection apparatus according to any one of claims 1 to 4, wherein the infrared intensity change or the phase difference of the infrared intensity change is observed by Fourier transform. In an infrared flame detection method for detecting a flame by observing the intensity of infrared rays emitted from the flame,
An infrared flame detection method, wherein a flame is determined by measuring a time lag of an output between infrared detectors having different wavelength sensitivities. 2. An infrared flame detecting apparatus according to claim 1, wherein an average intensity of the background radiation infrared rays during a time when no flame is generated is observed, and a flame detection threshold value is determined from the average intensity. . 2. The infrared detecting apparatus according to claim 1, wherein an average variation of the background radiation infrared rays during a time when no flame is generated is observed, and a flame detection threshold value is determined from the average variation. The infrared flame detection apparatus according to claim 7, wherein a flame detection threshold value is determined in accordance with a state of fouling of a window for transmitting infrared light for detecting infrared light intensity from the average intensity. . 9. The infrared flame detecting apparatus according to claim 8, wherein a flame detection threshold value is determined from the average fluctuation in accordance with a state of fouling of a window that transmits infrared light for detecting infrared light intensity.

Images