JP2004060483A - Air-fuel ratio control system for engine - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To keep high accuracy in calibrating a canister model even when meeting calibration timing in the exposed state of a base air-fuel ratio error immediately after starting. <P>SOLUTION: A controller 11 estimates a fuel quantity desorbed from a canister 24 in purge processing using the canister model, computes the quantity of purge part correction to reduce air-fuel ratio deviation from a stoichiometric air-fuel ratio by performing purge processing at a target purge ratio based on the desorbed fuel quantity, and corrects the quantity of fuel supply to an engine with the quantity of purge part correction. The controller 11 further sets a second limiter narrower in width to the central value of the air-fuel ratio feedback correction quantity than a first limiter, performs purge cut-off when meeting calibration timing in the case of determining the presence of a base air-fuel ratio error by comparing the computed air-fuel ratio feedback correction quantity with the second limiter, and performs first calibration processing to the canister model based on data before and after the purge cut-off. <P>COPYRIGHT: (C)2004,JPO

Description

ただし、Ｔｐ：基本噴射パルス幅［ｍｓｅｃ］、
Ｔｆｂｙａ：目標当量比［無名数］、
Ｋａｔｈｏｓ：過渡補正量［ｍｓｅｃ］、
ＦＨＯＳ：パージ分補正係数（遅れなまし処理後の値）、
α：空燃比フィードバック補正係数、
αｍ：空燃比学習値、
Ｔｓ：無効パルス幅［ｍｓｅｃ］、
の式により燃料噴射弁４に与える燃料噴射パルス幅Ｔｉ［ｍｓｅｃ］を演算する。この演算は図２中の基本噴射パルス幅演算部Ｂ７、空燃比フィードバック補正係数演算部８、燃料噴射パルス幅演算部Ｂ９、空燃比学習制御部Ｂ１６における処理に対応する。
【００９１】
ここで、パージ分補正係数ＦＨＯＳで過渡補正量Ｋａｔｈｏｓを除いた部分（Ｔｐ×Ｔｆｂｙａ）を補正することにしているのは、パージにより供給される燃料は気体燃料であるため壁流補正を行う必要がないためである。
【００９２】
基本噴射パルス幅Ｔｐはほぼ理論空燃比の得られる噴射パルス幅で、吸入空気流量Ｑとエンジン回転速度Ｎから求めている。
【００９３】
目標当量比Ｔｆｂｙａは１．０を中心とする値で、冷却水温が低いときやエンジン始動直後には１．０より大きな値となる。エンジンの暖機完了後はＴｆｂｙａ＝１．０である。
【００９４】
空燃比フィードバック補正係数αは排気空燃比が理論空燃比を中心とするいわゆるウィンドウに収まるように空燃比センサ１５の出力に基づいて演算される値である。具体的には１００％を中心とする値で、定常偏差（ベース空燃比エラー）がないとした場合、検出された空燃比が理論空燃比よりもリッチ側にあるとき１００％よりも小さな値に、検出された空燃比が理論空燃比よりもリーン側にあるときに１００％よりも大きな値になる。
【００９５】
空燃比学習値αｍは空燃比フィードバック補正係数αに基づいて演算される値で、主に定常偏差を補償するために導入されている。
【００９６】
燃料噴射弁４から噴射された燃料の一部は吸気ポート壁や吸気弁傘裏部に壁流燃料を形成し、この壁流燃料は加速や減速といった過渡時に大きな応答遅れを有してシリンダに流入する。過渡補正量Ｋａｔｈｏｓはこの過渡時の壁流燃料の応答遅れ分を補償するための値である。
【００９７】
無効パルス幅Ｔｓは、燃料噴射弁４が所定のタイミングで噴射信号を受けてから実際に開弁するまでの作動遅れを補償するための値である。
【００９８】
また、（７）式はシーケンシャル噴射（４気筒ではエンジン２回転毎に１回、各気筒の点火順序に合わせて噴射）の場合の式であるため、数字の２が入っている。
【００９９】
図１２、図１３は目標パージ率ＭＰＲを設定するためのもので、一定時間毎（例えば１０ｍｓｅｃ毎）に実行する。この演算は図２中の目標パージ率設定部Ｂ１における処理に対応する。
【０１００】
ステップ６１では初回の較正処理後であるか否かを判定する。これは初回の較正処理の前後で目標パージ率の設定方法が異なるためである。初回の較正処理前であるときにはそのまま今回の処理を終了する。なお、初回の較正処理前の目標パージ率の設定については、後述するブートアップ制御のところで説明する。
【０１０１】
初回の較正処理後であるときにはステップ６２に進み、キャニスタモデルに基づき演算された脱離量Ｄｇ（図４ステップ６により演算されている）とパージ流量ＰＱとに基づきパージガスの空燃比（パージ空燃比）を演算する。ここで、パージ流量ＰＱは目標パージ率と吸入空気流量Ｑとの積であり、この場合の目標パージ率には前回値を用いる。
【０１０２】
このパージ空燃比に対しては次のような補正を施してもかまわない。すなわち、運転状態、例えばエンジン回転速度、エンジン負荷、吸入空気流量などのパラメータから、パージ空燃比の誤差を推定する。例えば、図１４を内容とするテーブルを検索することにより求める。図示のように吸入空気流量が少なくなるほど、またパージ率が小さくなるほどパージ空燃比誤差は大きくなる値である。あるいは、パージ空燃比誤差を、図１５に示すようにパージ空燃比とパージ空燃比誤差の関係を規定したテーブルを検索することにより求めるようにしてもよい。パージ空燃比誤差が求まったら、ステップ６２で求めたパージ空燃比をこの誤差に基づいて補正する。
【０１０３】
なお、パージ空燃比はＨＣセンサによって検出するようにしてもよいが、キャニスタモデルに基づき演算される脱離量等に基づき演算によって求めればパージ空燃比を安価かつ正確に演算することができる。
【０１０４】
ステップ６３ではこのようにして求めた（あるいは誤差補正後の）パージ空燃比に基づきパージ率変化量制限値を演算する。パージ率が変化すると排気空燃比が変化するが、このときの理論空燃比からの空燃比偏差が許容幅以内に収まるようにパージ率変化量制限値を演算する。理論空燃比からの空燃比偏差の許容幅は空燃比フィードバック制御により吸収可能な、排気エミッションを悪化させない幅に設定する。
【０１０５】
ステップ６４では、パージバルブ２７のサイズから規定されるパージ率上限値ＰＶＭＸを演算する。このようなパージ率上限値ＰＶＭＸを求めるのは、目標パージ率がパージバルブ２７を最大開度として得られるパージ率よりも大きな値に設定されてしまうと、パージ率と目標パージ率との不一致が生じ、ＦＨＯＳの演算の誤差が大きくなるため、理論空燃比からの空燃比偏差が増加する。これにより、排気エミッション悪化等の問題が生じるからである。具体的には、パージバルブ２７のサイズが一定の場合、パージバルブ２７の前後差圧が大きいほど流せるパージガスの流量も多くなることから、パージバルブ２７の前後差圧が大きいときにパージ率上限値ＰＶＭＸとして大きな値を演算する。
【０１０６】
ステップ６５では、燃料噴射弁４の性能に応じて決まる最小噴射パルス幅、目標パージ率の前回値、パージ分補正係数との関係から燃料噴射弁４の性能に基づくパージ率上限値ＴＩＭＮＭＸを演算する。パージ率が高くなるとパージによってエンジンに供給される燃料量が増加するので、燃料噴射弁４からの燃料噴射量がその分だけ減らされるように燃料噴射パルス幅を補正するが、燃料噴射弁４の噴射精度を確保するためには噴射パルス幅は所定の最小パルス幅よりも大きくなくてはならない。言い換えれば、燃料噴射パルス幅を最小パルス幅より大きくするためにはパージ率はある値よりも小さくしなくてはならない。このような理由から、燃料噴射弁４の噴射性能によってもパージ率の上限を規定する。
【０１０７】
また、ステップ６６では、現在の運転領域から想定しうる全ての運転領域を想定し、その中での最小パージ率を予測し、この最小パージ率とパージ率変化量制限値とからパージ率上限値ＰＲＭＮＭＸを演算する。例えば、アクセルペダルをあるだけ踏み込んでで加速した場合に目標パージ率はごく小さな値に設定するが、このアクセルペダルをあるだけ踏み込む直前に目標パージ率が大きな値に設定していると、パージ率の変化量が変化量制限値以下に制限されていることからパージ率を目標パージ率に追従させることができなくなる。この追従遅れは排気エミッション増大の原因等となることから、こうした追従遅れを生じないように想定しうる最小パージ率からもパージ率の上限を規定する必要がある。
【０１０８】
また、ステップ６７では、空燃比フィードバック補正係数αをモニターし、所定値以下であれば空燃比フィードバック補正係数αを所定値以上とするパージ率のうち最も大きな値をパージ率上限値ＡＬＰＭＸとして演算する。このような上限値ＡＬＰＭＸを設けるのは、空燃比フィードバック制御では空燃比フィードバック補正係数αは１００±２５％に収まるように制御されているが、空燃比フィードバック補正係数αが前記制限値近傍（例えば８０％）で制御されているような場合は、大量のパージを行っているとパージ以外の外乱を受けて前記制御範囲から外れやすくなるからである。
【０１０９】
ステップ６８では上記４つの上限値ＰＶＭＸ、ＴＩＭＮＭＸ、ＰＲＭＮＭＸ、ＡＬＰＭＸから最も小さい値を選択し、その値を最大パージ率ＰＲＭＡＸとして設定する。
【０１１０】
ステップ６９、７０では目標パージ率の前回値と最大パージ率ＰＲＭＡＸとを比較し、目標パージ率の前回値と最大パージ率ＰＲＭＡＸとが等しいときにはステップ６９よりステップ７３に進んで目標パージ率を前回値のままとし、目標パージ率の前回値が最大パージ率ＰＲＭＡＸよりも大きいときにはステップ６９、７０よりステップ７２に進んで目標パージ率をその前回値からパージ率変化量制限値を引いた値とし、目標パージ率の前回値が最大パージ率ＰＲＭＡＸよりも小さいときにはステップ６９、７０よりステップ７１に進んで目標パージ率をその前回値にパージ率変化量制限値を加えた値とする。
【０１１１】
図１３のステップ７４〜８１の処理については本願発明により追加した部分であり、後述する。
【０１１２】
このようにして、目標パージ率は、最大パージ率ＰＲＭＡＸを目標としてパージ率変化量制限値の範囲内でこれに追従するように設定され、排気エミッションを悪化させずに大量のパージを行う上で最良のパージ率が設定される。また、最大パージ率ＰＲＭＡＸを設定する際に、物理的な制限、現在の運転領域等で決まる上限値ＰＶＭＸ、ＴＩＭＮＭＸ、ＡＬＰＭＸだけでなく、運転領域が変化した場合でも遅れなくその領域での最大パージ率に移行できるように決定される上限値ＰＲＭＮＭＸをも考慮しているので、運転条件が変化しても排気エミッションを悪化させずに大量パージを行う上で最良のパージ率を設定することができる。
【０１１３】
このようにして設定された目標パージ率はパージバルブへのデューティ比に変換され、パージバルブ２７に出力される。
【０１１４】
ところで、キャニスタモデルに基づくパージ処理（モデル規範パージ処理）は、エンジン始動後に較正処理が一度も行われておらずキャニスタモデルで用いる初期値（初期吸着量）が存在しない間は、実行することができない。しかし、エンジン始動後より大量パージを実現するためには、たとえ較正処理前であってもパージ処理を実行する必要がある。そこで初回の較正処理が実行されるまでは、キャニスタモデルに基づくパージ処理に代えて次に示す制御（図１６に示すブートアップ制御）により目標パージ率を設定し、設定された目標パージ率でパージ処理を実行する。なお、このブートアップ制御では、パージによる排気空燃比への影響は空燃比フィードバック補正係数αによって吸収し、燃料噴射パルス幅の補正は行わない（ＦＨＯＳ＝１．０）。
【０１１５】
図１６はブートアップ制御を行うためのもので、一定時間毎（例えば１０ｍｓｅｃ毎）に実行する。このブートアップ制御は図２中のブートアップ制御部Ｂ１４、切換スイッチＢ１５における処理に対応する。
【０１１６】
ステップ９１で吸入空気流量Ｑを読み込み、ステップ９２で初回の較正処理前であるか否かを判定する。初回の較正処理前である場合には、ステップ９３で積算パージ流量（パージを開始してからの総パージ流量）ＳＱＰとパージ配管容積（キャニスタ２４からパージバルブ２７までの配管の容積）を比較する。積算パージ流量ＳＱＰがパージ配管容積を超えているときにはステップ９４に進み、これに対して超えていないときにはステップ９７に進む。
【０１１７】
ステップ９７では目標パージ率ＭＰＲとして初期パージ率（１％以下の小さな値）を設定する。このような小さな値に設定するのは、積算パージ流量ＳＱＰがパージ配管容積に達してない場合はパージ開始前にパージ配管内のガスがエンジンに供給されることになるが、このパージ配管内のガスの空燃比が不明であり、このままステップ９４〜９６に示す目標パージ率設定処理を行うとエンジンの燃焼安定性悪化等の問題を生じるからである。
【０１１８】
つまり、パージ開始時にパージ配管中に存在する低濃度のパージガスが供給され、これによる理論空燃比からの空燃比偏差が小さいと、さらに大量のパージが可能であると判断されて大きな目標パージ率ＭＰＲが設定されるが、このようにして大きな目標パージ率ＭＰＲが設定されてしまうと、配管内の低濃度のガスが全て供給されて本来の高濃度のパージガスが供給されるときに大量の脱離燃料が突然供給されることになり、エンジンの燃焼安定性等を悪化させる原因となるからである。
【０１１９】
積算パージ流量ＳＱＰが配管容積を超えたらステップ９４に進み実空燃比フィードバック偏差と目標空燃比フィードバック偏差との差を演算する。
【０１２０】
ここで、目標空燃比フィードバック偏差とは、空燃比フィードバック補正係数の目標値ｔαと空燃比フィードバック補正係数の中心値（１００％）との偏差（＝｜ｔα−１００｜％）をいい、実空燃比フィードバック偏差とは実際の空燃比フィードバック補正係数αと空燃比フィードバック補正係数の中心値との偏差（＝｜α−１００｜％）をいう。例えば、パージによる理論空燃比からの空燃比偏差を空燃比フィードバック制御で十分吸収できる範囲内で大量のパージ流量を確保することを目的として空燃比フィードバック補正係数αの目標値ｔαが８０％に設定されると、目標空燃比フィードバック偏差は｜８０−１００｜＝２０％に設定される。
【０１２１】
ステップ９５では図１７を内容とするテーブルを検索することにより上記目標空燃比フィードバック偏差と実空燃比フィードバック偏差との差に応じたパージ率変化量を算出する。パージ率変化量は、目標空燃比フィードバック偏差と実空燃比フィードバック偏差の差の絶対値が大きくなるほど大きな値を設定して、目標値への収束性を高めるのであるが、目標空燃比フィードバック偏差と実空燃比フィードバック偏差の差の正負によって、偏差の絶対値が同じであっても異なる値を設定し、空燃比フィードバック偏差の差が負側にずれた場合の方が大きな値（絶対値）をパージ率変化量に設定する。
【０１２２】
このようにパージ率変化量を空燃比フィードバック偏差の正負で異なる特性とするのは、空燃比フィードバック偏差の差が負側にずれている場合には空燃比フィードバック補正係数αが目標とする８０％よりも小さな値になっており、逆の正側にずれている場合と比べてパージ以外の外乱によってエンジン安定性、排気エミッションの悪化を招きやすく、不利な状態あるといえるからである。つまり、パージ率変化量を空燃比フィードバック偏差の差の正負に応じて特性を変えるのは、エンジンの燃焼安定性及び排気エミッション悪化防止の観点から、制御点を速やかに安全側に復帰させるためである。
【０１２３】
以上のようにしてパージ率変化量を演算したらステップ９６に進み、本ルーチン前回実行時に求めた目標パージ率に対して、ステップ９５で演算したパージ率変化量を付加し、新たな目標パージ率ＭＰＲを演算する。
【０１２４】
ステップ９８ではこのようにして得た目標パージ率ＭＰＲと吸入空気流量Ｑからパージ流量ＰＱを求め、これをステップ９９において積算パージ流量の前回値であるＳＱＰｚに加算することにより積算パージ流量ＳＱＰを更新する。
【０１２５】
したがって、ステップ９４〜９６の処理によると、キャニスタ２４の吸着状態によらず、最適な目標パージ率ＭＰＲを設定することができ、また、想定以上の濃度のパージガスが供給された場合でも、それによる排気空燃比の変化を受けて目標パージ率ＭＰＲが適宜変更され、常に最適な目標パージ率ＭＰＲを設定することができる。
【０１２６】
一方、初回の較正処理後にはステップ９２よりステップ１００以降に進み、積算パージ流量ＳＱＰに基づいてパージ処理が終了したか否かを判定する。すなわち、ステップ１００でパージ終了判定フラグ（ゼロに初期設定）をみる。エンジン始動直後にはパージ終了判定フラグ＝０であるので、ステップ１０１に進み脱離量Ｄｇ（図４のステップ６で演算されている）と目標パージ率ＭＰＲ（図１２、図１３により演算されている）を読み込む。ステップ１０２では、この読み込んだ目標パージ率ＭＰＲと吸入空気流量Ｑとからパージ流量ＱＰを算出し、これをステップ１０３において積算パージ流量の前回値であるＳＱＰｚに加算することにより積算パージ流量ＳＱＰを更新する。
【０１２７】
ステップ１０４では初回の較正処理タイミングであるか否かを判定する。ここで、初回の較正処理タイミングはステップ９２で初回の較正処理前でなくなった初めてのタイミングであるので、このときだけステップ１０５に進み、ステップ１０２で演算されているパージ流量ＰＱとステップ１０１で読み込んでいる脱離量Ｄｇとに基づいてパージ空燃比を演算し、このパージ空燃比からステップ１０６において図１８を内容とするテーブルを検索することにより、パージ終了判定値ＳＬＳＱＰを演算する。
【０１２８】
ここで、パージ終了判定値ＳＬＳＱＰはキャニスタ２４からのパージガス中にほぼ脱離燃料がなくなったと思われる積算パージ流量である。図１８のように、この判定値ＳＬＳＱＰはパージ空燃比が大きくなるほど小さくなる。
【０１２９】
ステップ１０７ではこのパージ終了判定値ＳＬＳＱＰと積算パージ流量ＳＱＰとを比較する。積算パージ流量ＳＱＰがパージ終了判定値ＳＬＳＱＰ未満であればそのまま今回の処理を終了する。やがて積算パージ流量ＳＱＰがパージ終了判定値ＳＬＳＱＰ以上になればキャニスタ２４からのパージガス中にほぼ脱離燃料がなくなったと判断し、パージが終了したことを表すためステップ１０８に進んでパージ終了判定フラグ＝１とする。
【０１３０】
パージ終了判定フラグ＝１になると次回からはステップ９１、９２、１００と流れてもステップ１０１以降に進むことができないので、そのまま処理を終了する。
【０１３１】
次に、パージ処理を行うことによる本実施形態の全体的な作用について先に説明する。
【０１３２】
パージ処理時、目標パージ率ＭＰＲはエンジン燃焼安定性低下、排気エミッション増大を起こさない範囲でできる限り大きな値が設定され、この目標パージ率ＭＰＲが実現されるようにパージバルブ２７が開かれる（図１２ステップ６１〜７３）。
【０１３３】
パージ処理中はキャニスタ２４から脱離した燃料を含んだパージガスがエンジンに供給されることになるので、エンジンコントローラ１１ではキャニスタ２４から脱離してくる燃料量Ｄｇを推定し、この脱離量Ｄｇ、目標パージ率、吸入吸気質量Ｑｇに基づいて、パージ燃料によって生じる空燃比フィードバック補正係数αの変化を予測する値であるパージ分補正係数ＦＨＯＳを演算し（図４ステップ１、３、４、６、７、８またはステップ１、３、５、６、７、８）、このパージ分補正係数ＦＨＯＳにより基本噴射パルス幅Ｔｐを補正することで、パージによる理論空燃比からの空燃比偏差が小さなものに抑えられる。
【０１３４】
このときキャニスタ２４からの脱離量Ｄｇは、（１）式から（４）式で表されるキャニスタモデルを用いて推定され、脱離量Ｄｇは短時間でかつ正確に推定される。ここで、キャニスタモデルに基づき演算される脱離量Ｄｇ等は誤差を含んでおり、キャニスタモデルの動作時間が長くなるにつれこの誤差が積算され大きくなるので、エンジンコントローラ１１は空燃比フィードバック補正係数αの変化よりキャニスタ２４から脱離された燃料量を推定し、この推定した脱離量から逆演した吸着量Ｙｒでキャニスタモデルの内部変数である吸着量の値を較正する。この較正処理は、パージ以外の外乱が小さく空燃比フィードバック係数αの変化をほぼ全てパージによるものとみなすことができ、かつ、パージによる排気空燃比への影響が比較的大きいときに実行される。
【０１３５】
また、パージバルブ２７が開かれてから脱離燃料がエンジンのシリンダに到達するまでに遅れがあり、また到達するまでに燃料の拡散もあるので、この遅れとなまし作用を考慮して上記のパージ分補正係数ＦＨＯＳに対して遅れ補正が施されている。
【０１３６】
また、上記キャニスタモデルを用いたパージ処理（モデル規範パージ処理）は、較正処理によって吸着量の初期値が求まるまではその効果を発揮することができないが、キャニスタモデルの初期値が演算されるまでは目標空燃比フィードバック偏差と実空燃比フィードバック偏差との差に応じて目標パージ率が設定され、この目標パージ率が実現されるようにパージバルブ２７が開かれる。これにより、較正処理によって初期値が演算される前であってもパージ処理を行うことができ、全領域で効果的なパージを行うことができる。
【０１３７】
これで先願装置（特願２００１−７１５６２号参照）におけるパージ処理と同様の部分の説明を終了し、次には本願発明で追加した図６のステップ１１１〜１１９、図１３のステップ７４〜８１、図１９、図２０、図２１、図２３、図２４、図２６、図２７に示すフローチャートを説明する。
【０１３８】
まず目標パージ率を設定するためのフローチャートの後半部である図１３から説明する。この部分の処理は図３１中のＡＮＤ回路Ｂ３８、パージ率低下処理部Ｂ３９、切換スイッチＢ４０における処理に対応する。
【０１３９】
ステップ７４では目標パージ率を変数ＭＰＲ０に移す。
【０１４０】
ステップ７５では較正実行可能判定フラグ（ゼロに初期設定）をみる。前述の図５では「定常条件」、「パージバルブ精度条件」、「パージ影響度条件」全てが成立した場合に較正処理実行可能タイミングであると判断したが、較正実行可能判定フラグはこのタイミングで１となるフラグである。較正実行可能判定フラグ＝０であるときにはステップ７８に進みＭＰＲ０の値をそのまま目標パージ率ＭＰＲとして設定する。
【０１４１】
一方、較正実行可能判定フラグ＝１であるときにはステップ７６、７７に進みデータサンプリング終了フラグ、アクティブ較正要求フラグ（いずれもゼロに初期設定）をみる。後述するように、ベース空燃比エラーがあると判定されたときにアクティブ較正要求フラグ＝１となり、このアクティブ較正要求フラグの０から１への切換タイミングで、パージカット前のデータをサンプリングした後にパージ率低下処理を開始する。このパージ率低下処理はアクティブ較正要求フラグが０から１へと切換わる直前の目標パージ率を初期値として、目標パージ率をゼロになるまで所定時間当たり所定値ずつ低下させる処理であり、このパージ率低下処理により目標パージ率がゼロとなった状態（パージカット状態）で安定化時間を経過したとき、パージカット時のデータをサンプリングし、これでパージカット前後のデータのサンプリングを終了するので、データサンプリング終了フラグ＝１としている。
【０１４２】
このため、アクティブ較正要求フラグ＝１となる前にはデータサンプリング終了フラグ＝０であり、このときにはステップ７６。７７よりステップ７８に進み、またデータサンプリング終了フラグ＝１となったときにはステップ７６よりステップ７８に進みステップ７８の操作を実行する。
【０１４３】
これに対して、アクティブ較正要求フラグ＝１となってからデータサンプリング終了フラグ＝１となる直前までの間はステップ７９に進んでパージ率低下処理を行う。
【０１４４】
ここで、ステップ７７のアクティブ較正要求フラグの設定について図１９のフローより説明する。
【０１４５】
図１９はアクティブ較正要求フラグを設定するためのもので、一定時間毎（例えば１０ｍｓｅｃ毎）に実行する。
【０１４６】
ステップ１２１ではモデルパージ実行中であるかどうかみる。モデルパージ実行中とはキャニスタモデルの初回の較正処理が終了しておりかつパージを禁止していない状態のことである。モデルパージ実行中であればステップ１２２に進み、パージ用リミッタ判定フラグをみる。このパージ用リミッタ判定フラグにはパージ用下限リミッタ判定フラグとパージ用上限リミッタ判定フラグの２つがあり、これら２つのフラグの設定については図２０、図２１のフローより説明する。
【０１４７】
図２０はパージ用下限リミッタ判定フラグを、また図２１はパージ用上限リミッタ判定フラグを設定するためのもので、いずれも一定時間毎（例えば１０ｍｓｅｃ毎）に実行する。これら図２０、図２１の設定処理と図１９の設定処理とが、図３１中のベース空燃比エラー判定部Ｂ３６、アクティブ較正要求部Ｂ３７における処理に対応する。
【０１４８】
空燃比フィードバック補正係数αには、従来よりαによる制御範囲を限定するため図２８に示したように、αの中心値（１００％）を中心として上下に第１リミッタ±ＬＭＴ１（例えば±２５％）を設けている。すなわち、αの第１上限リミッタは１２５％（＝１００＋ＬＭＴ１）、αの第１下限リミッタは７５％（＝１００−ＬＭＴ１）であり、演算されたαが、αの第１上限リミッタである１２５％を超えるときには、このαを１２５％に制限し、この逆に演算されたαが、αの第１下限リミッタである７５％を下回るときには、このαを７５％に制限している。
【０１４９】
本実施形態では、この第１リミッタ±ＬＭＴ１とは別に、パージ用にこれより小さい第２リミッタ±ＬＭＴ２を新たに導入している。すなわち、第２下限リミッタは１００−ＬＭＴ２、第２上限リミッタは１００＋ＬＭＴ２であり、図２８において、αの中心値である１００％と第１下限リミッタである１００−ＬＭＴ１との間に、１００−ＬＭＴ１からある程度の余裕代を持たせた位置に第２下限リミッタである１００−ＬＭＴ２がくるようにしている。すなわち、１００−ＬＭＴ２（第２下限リミッタ）と１００−ＬＭＴ１（第１下限リミッタ）の間をαの補正代として残し、パージ以外の外乱があった場合に、この残したαの補正代により理論空燃比からの空燃比偏差を小さなものとする。
【０１５０】
そして、αが第２下限リミッタ以下の領域にあるか否かあるいは第２上限リミッタ（図２８には示していない）以上の領域にあるか否かを示すため、フラグを設定する。すなわち、演算されたαが第２下限リミッタである１００−ＬＭＴ２以下になるとパージ用下限リミッタ設定フラグ＝１とし、この逆に演算されたαが第２上限リミッタである１００＋ＬＭＴ２以上となるとパージ用上限リミッタ設定フラグ＝１とする。
【０１５１】
ただし、第２下限リミッタ、第２上限リミッタにはそれぞれヒステリシスＨＹＳを設けている。これは、αが各リミッタ付近にいることにより各フラグがゼロとなったり１となったりを頻繁に繰り返すことを防止するためである。
【０１５２】
具体的に図２０から説明すると、ステップ１３１では空燃比フィードバック制御中であるか否かを判定する。空燃比フィードバック制御中でなければそのまま今回の処理を終了する。
【０１５３】
空燃比フィードバック制御中であるときには、ステップ１３２に進んで空燃比フィードバック補正係数αを読み込む。
【０１５４】
このαは図示しないフローにおいて空燃比センサ１５に基づいて演算している。ここでは、簡単のため比例動作の場合で説明すると、空燃比センサ１５により検出される実際の排気空燃比と理論空燃比との偏差を求め、この偏差に比例ゲインを乗算することにより求めればよい。このときαの波形は、実際の排気空燃比が理論空燃比よりリッチ側にあるあいだは、実際の排気空燃比を理論空燃比へと戻すため制御周期毎にαが階段状に小さくなってゆき、この逆に実際の排気空燃比が理論空燃比よりリーン側にあるあいだは、実際の空燃比を理論空燃比へと戻すため制御周期毎にαが階段状に大きくなってゆく。
【０１５５】
ステップ１３３ではパージ用下限リミッタ判定フラグ（エンジン始動時にゼロに初期設定）をみる。始動直後であればパージ用下限リミッタ判定フラグ＝０であることよりステップ１３４に進みαとパージ用下限リミッタである１００−ＬＭ２とを比較する。αが１００−ＬＭＴ２を超えているときにはそのまま今回の処理を終了し、αが１００−ＬＭＴ２以下になると、ステップ１３５に進んでパージ用下限リミッタ判定フラグ＝１として今回の処理を終了する。
【０１５６】
このパージ用下限リミッタ判定フラグ＝１により次回にはステップ１３３よりステップ１３６に進みαと１００−ＬＭＴ２＋ＨＹＳ（ＨＹＳは正の一定値）を比較する。αが１００−ＬＭＴ２＋ＨＹＳ以下であればステップ１３５に進んでパージ用下限リミッタ判定フラグ＝１のままとし、αが１００−ＬＭＴ２＋ＨＹＳを超えるとステップ１３７に進んでパージ用下限リミッタ判定フラグ＝０とする。
【０１５７】
図２１に移ると、ステップ１４１、１４２は図２０のステップ１３１、１３２と同様である。
【０１５８】
ステップ１４３ではαのパージ用上限リミッタ判定フラグ（エンジン始動時にゼロに初期設定）をみる。始動直後であればパージ用上限リミッタ判定フラグ＝０であることよりステップ１４４に進みαとパージ用上限リミッタである１００＋ＬＭＴ２とを比較する。αが１００＋ＬＭＴ２未満であるときにはそのまま今回の処理を終了し、αが１００＋ＬＭＴ２以上になると、ステップ１４５に進んでパージ用上限リミッタ判定フラグ＝１として今回の処理を終了する。
【０１５９】
このパージ用上限リミッタ判定フラグ＝１により次回にはステップ１４３よりステップ１４６に進みαと１００＋ＬＭＴ２−ＨＹＳ（ＨＹＳは正の一定値）を比較する。αが１００＋ＬＭＴ２−ＨＹＳ以上であれステップ１４５に進んでパージ用上限リミッタ判定フラグ＝１のままとし、αが１００＋ＬＭＴ２−ＨＹＳ未満になるとステップ１４７に進んでパージ用上限リミッタ判定フラグ＝０とする。
【０１６０】
図１９のステップ１２２に戻り、パージ用リミッタ判定フラグ（パージ用下限リミッタ判定フラグまたはパージ用上限リミッタ判定フラグ）＝１であるときにはステップ１２３に進みタイマ１（ゼロに初期設定）をインクリメントする。タイマ１は、継続してパージ用リミッタ判定フラグ＝１でいる時間を計測するためのものである。このため、インクリメントの途中でもパージ用リミッタ判定フラグ＝０となればステップ１２６に進んでタイマ１をゼロにクリアし、再びパージ用リミッタ判定フラグ＝１となればステップ１２３に進んでタイマ１をインクリメントする。
【０１６１】
ステップ１２４でタイマ１値と所定値を比較する。所定値はベース空燃比エラーがあるかどうかを判定するための値である。タイマ１値が所定値以上になるとベース空燃比エラーがあると判断し、ステップ１２５に進んでアクティブ較正処理（第２較正処理）を行うため、アクティブ較正要求フラグ（ゼロに初期設定）＝１とする。
【０１６２】
ただし、運転条件が過渡的に変化するときには、ベース空燃比エラーが存在していなくても一時的にパージ用リミッタ判定フラグ＝１となることがあるので、この場合にもベース空燃比エラーがあると誤判定されることがないように所定値を設定する。すなわち、過渡的にパージ用リミッタ判定フラグ＝１となる時間よりも長めの時間（例えば数秒）を所定値として定めておくことで、運転条件の過渡的変化による場合をベース空燃比エラーと誤判定することを回避する。
【０１６３】
これで、図１９の説明を終了するので、図１３に戻り、アクティブ較正要求フラグ＝１となった直後であればデータサンプリング終了フラグ＝０であるため、ステップ７６、７７よりステップ７９に進んでパージ率低下処理を行う。
【０１６４】
このパージ率低下処理は図２２に示したように、アクティブ較正要求フラグ＝１かつ較正実行可能判定フラグが０から１へと切換わったタイミングでの目標パージ率（図１２のステップ６１〜７３で演算されている）を初期値として所定時間Δｔ当たり所定値Δずつ減らし最終的にゼロにする制御である。これは、パージカットを行うための一方法であり、初期値からステップ的にゼロにするのではない。
【０１６５】
図２３、図２４（図１３ステップ７９のサブルーチン）はこのパージ率低下処理を行うためのものである。この処理は図３１中のパージ率低下処理部Ｂ３９、パージカット前データサンプリング部Ｂ４２、パージカット時データサンプリング部Ｂ４３における処理に対応する。
【０１６６】
図２３において、ステップ１５１、１５２では今回と前回の較正実行可能判定フラグをみる。今回に較正実行可能判定フラグ＝１かつ前回に較正実行可能判定フラグ＝０（つまり較正実行可能判定フラグが０より１へと切換わった直後）のときステップ１５３に進み、パージ率低下処理開始時の空燃比フィードバック補正係数α（図示しないαの演算フローにおいて演算されている）、パージ分補正係数ＦＨＯＳ（図４により演算されている）、目標パージ率ＭＰＲ０（図１３ステップ７４で得ている）、吸入空気質量Ｑｇ（吸入空気流量Ｑと吸気温度等から得ている）、空燃比学習値αｍ（図示しない空燃比学習値のマップからそのときの運転条件の属する小領域に入っている空燃比学習値を読み出してくる）を読み込んで、対応するメモリα１、ＦＨＯＳ１、ＭＰＲ１、Ｑｇ１、αｍ１に入れる（パージカット前データのサンプリング）。
【０１６７】
ステップ１５４では同じくパージ率低下処理開始時のＭＰＲ０の値を変数ＭＰＲＧに移す。これは、較正実行可能判定フラグが０より１へと切換わったタイミングでの目標パージ率を初期値としてサンプリングするためである。
【０１６８】
ステップ１５５ではタイマ２を一旦ゼロにクリアした後、ステップ１５６でインクリメントを開始する。
【０１６９】
今回、前回とも較正実行可能判定フラグ＝１であるときにはステップ１５３、１５４、１５５を飛ばしてステップ１５６の操作を実行する。
【０１７０】
ステップ１５７ではタイマ２値と所定時間Δｔを比較する。タイマ２値が所定時間Δｔ以上になると、ステップ１５８、１５９に進み、タイマ２値をゼロにクリアすると共に、
ＭＰＲＧ＝ＭＰＲＧｚ−Δ…（８）
ただし、ＭＰＲＧｚ：ＭＰＲＧの前回値、
の式により変数ＭＰＲＧを所定値Δ（正の値）だけ小さくする。タイマ２値が所定時間Δｔ未満のときにはステップ１６０に進み現在のＭＰＲＧを維持する。
【０１７１】
ステップ１５７〜１６０の操作により、ＭＰＲＧは所定時間Δｔが経過する毎に所定値Δずつ小さくなってゆく（図２２最下段参照）。
【０１７２】
ここで、パージ率を急激に低下させてのパージカットを行うと排気空燃比が変化し運転性や排気エミッションに影響するので、そうならないように所定時間Δｔと所定値Δの値は目標パージ率を設定する際のパージ率変化量制限値を用いて、またはこれに基づいて定めればよい。
【０１７３】
図２４に進み、ステップ１６１ではＭＰＲＧとゼロを比較する。ＭＰＲＧがゼロ以上であればステップ１６２に進みＭＰＲＧを目標パージ率ＭＰＲとして、これに対してＭＰＲＧが負の値になったときにはステップ１６３に進みゼロを目標パージ率ＭＰＲとして設定する。
【０１７４】
ステップ１６４では目標パージ率ＭＰＲとゼロを比較する。ＭＰＲ＝０であるときにはステップ１６５に進んでタイマ３（ゼロに初期設定）をインクリメントする。タイマ３はパージカット状態（目標パージ率ＭＰＲがゼロ）の連続経過時間を計測するためのものである。
【０１７５】
ステップ１６６ではタイマ３値と安定化時間を比較する。安定化時間はパージカット状態が安定するのを待つ時間である。
【０１７６】
タイマ３値が安定化時間以上になるとステップ１６７に進み、そのときの空燃比フィードバック補正係数α、空燃比学習値αｍを対応するメモリα２、αｍ２に入れる。これはパージカット時のデータをサンプリングするものである。これで、パージカット前後でのデータのサンプリングを終了するので、ステップ１６８ではデータサンプリング終了フラグ（ゼロに初期設定）＝１とすると共に、ステップ１６９でタイマ２、３をゼロにクリアする。
【０１７７】
このようにしてパージ率低下処理を終了したら図１３にのステップ８０に戻る。ステップ８０ではパージ終了判定フラグ（図１６により設定されている）をみる。パージ終了判定フラグ＝１のときには、空燃比学習を開始するためステップ８１に進み空燃比学習用のパージ率を目標パージ率ＭＰＲに入れる。これによって空燃比学習を開始するときには図２５に示したように目標パージ率ＭＰＲが小さくなる。
【０１７８】
図２６は空燃比学習開始フラグを設定するためのもので、一定時間毎（例えば１０ｍｓｅｃ毎）に実行する。
【０１７９】
ステップ１７１では空燃比学習開始フラグ（ゼロに初期設定）をみる。エンジン始動直後であれば空燃比学習開始フラグ＝０であることよりステップ１７２に進み、パージ終了判定フラグ（図１６により設定されている）をみる。
【０１８０】
パージ終了判定フラグ＝１であるときだけ空燃比学習を開始させるためステップ１７３に進み空燃比学習開始フラグ＝１とする。
【０１８１】
図２７は空燃比学習値の更新を行うためのもので、所定のタイミング毎（例えば所定のエンジン回転速度の間隔毎）に実行する。この空燃比学習値の更新と図２６の空燃比学習開始フラグ設定の各処理は図２中の空燃比学習制御部Ｂ１６における処理に対応する。
【０１８２】
ステップ１８１では空燃比学習開始フラグをみる。空燃比学習開始フラグ＝１であるときにはステップ１８２に進み空燃比学習許可条件が成立しているか否かを判定する。空燃比学習許可条件は従来と同様である。例えば空燃比フィードバック制御中であることや空燃比フィードバック制御に関わる空燃比センサ１５等に故障がないことなどを総て満足するときに空燃比学習許可条件が成立していると判断し、ステップ１８３に進んで空燃比学習値を更新する。この空燃比学習値の更新方法には次式によって更新する方法がある。
【０１８３】
αｍ＝αｍｚ＋（α−１００）×Ｍ…（９）
ただし、αｍ　：更新後の空燃比学習値、
αｍｚ：更新前の空燃比学習値、
Ｍ　　：学習割合（一定値）、
この場合、エンジンの回転速度と基本噴射パルス幅Ｔｐ（エンジン負荷相当）をパラメータとする全運転領域が複数の小さな領域に区分けされ、その各小領域毎に独立の空燃比学習値を有している。このため、（９）式右辺の更新前の空燃比学習値は更新タイミングでの運転条件が属する小領域に格納されている空燃比学習値であり、その同じ小領域に（９）式左辺の更新後の空燃比学習値が格納される。
【０１８４】
なお、実施形態は空燃比センサ１５に基づく空燃比フィードバック制御であるが、これに代えてＯ_２センサに基づく空燃比フィードバック制御であってもよく、この場合には、Ｏ_２センサの反転タイミングを空燃比学習値の更新タイミングとすればよい。
【０１８５】
次に、図６に戻ってステップ１１１〜１１９の処理を説明する。この部分の処理は図３１中のアクティブ較正処理部Ｂ４１、切換スイッチＢ４７における処理に対応する。
【０１８６】
ステップ１１１、１１２でアクティブ較正要求フラグ、データサンプリング終了フラグをみる。アクティブ較正要求フラグ＝０であるとき、アクティブ較正要求フラグ＝１であってもデータサンプリング終了フラグ＝０であるときにはステップ３２〜３６に進み先願装置と同様のパッシブ較正処理（第２較正処理）を実行する。
【０１８７】
アクティブ較正要求フラグ＝１かつデータサンプリング終了フラグ＝１であるときにはステップ１１３〜１１７でのアクティブ較正処理（第１較正処理）を行う。すなわち、ステップ１１３ではメモリに格納されているα１、α２、ＦＨＯＳ１、Ｑｇ１、αｍ１、αｍ２を読み込む。ここで、「１」が付いている値がパージカット前の、これに対して「２」が付いている値がパージカット時のデータである。
【０１８８】
ステップ１１４ではこれらパージカット前後のデータを用いて、
ＤＬＴ＝（α１＋αｍ１＋α２＋αｍ２−３）×ＦＨＯＳ１…（１０）
の式により全空燃比補正係数ＤＬＴを算出し、この全空燃比補正係数ＤＬＴとＭＰＲ１、Ｑｇ１とを用いて、
Ｄｇ＝Ｋ１×（１−ＤＬＴ＋Ｋ２×ＭＰＲ１）×Ｑｇ１…（１１）
の式により脱離量Ｄｇを算出する。
【０１８９】
ここで、（１０）式をどのようにして導いたかを説明する。
【０１９０】
（ア）空燃比学習制御を行わない場合：
実施形態は空燃比学習制御部Ｂ１６を備えるものであるが、空燃比学習制御を行わないものが考えの基本となるので、こちらから説明すると、このときには、
ＤＬＴ＝｛α１−（１−α２）｝×ＦＨＯＳ１…（ａ）
の式により全空燃比補正係数ＤＬＴを算出すればよい。ここで、（ａ）式右辺の（１−α２）がパージカット時における、αの中心値からのαの偏差、つまりベース空燃比エラーを表し、これがパージ実行中の値である（ａ）式右辺のα１に含まれてしまうことにより較正エラーとなるので、α１よりベース空燃比エラーを表す（１−α２）を差し引くことにより、ベース空燃比エラー分の誤差をα１から取り去っている。
【０１９１】
（イ）空燃比学習制御を行う場合：
このときには、空燃比学習値についても空燃比フィードバック補正係数αと同様に考える必要があるので、

の式により全空燃比補正係数ＤＬＴを算出する。ここで、（ｂ）式右辺の（１−αｍ２）はパージカット時における、αｍの中心値からのαｍの偏差、つまりこれもベース空燃比エラーを表し、これがパージ実行中の値である（ｂ）式右辺のαｍ１に含まれてしまうことにより較正エラーとなるので、αｍ１よりベース空燃比エラーを表す（１−αｍ２）を差し引くことにより、ベース空燃比エラー分の誤差をαｍ１から取り去っている。ただし、１．０を中心とする｛α１−（１−α２）｝と｛αｍ１−（１−αｍ２）｝の２つの値を足し合わせた合計は２となるので、２つの値の合計から１を差し引くことによって計算上のつじつま合わせをしている。（ｂ）式を整理すると上記（１０）式が得られる。
【０１９２】
このようにして全空燃比補正係数ＤＬＴが得られると、脱離量の演算式は前述の（５ｂ）式と同じである。ただし、この場合の目標パージ率ＭＰＲ、吸入空気質量Ｑｇはパージカット前の値なければならないので、上記（１１）のようにＭＰＲ、Ｑｇに代えてＭＰＲ１、Ｑｇ１を用いている。
【０１９３】
ステップ１１６、１１７はパッシブ較正処理と同様である。すなわち、ステップ１１６では、ステップ１１５で演算した脱離量Ｄｇから上記（６）式により、キャニスタ２４の吸着量Ｙｒ（質量）を演算する。ステップ１１７では、キャニスタモデルに基づき脱離量Ｄｇを演算する際に使用する吸着量Ｙを、ステップ１１６で演算した吸着量Ｙｒに置き換える。
【０１９４】
これでアクティブ較正処理を終了するので、ステップ１１８、１１９ではデータサンプリング終了フラグ＝０、アクティブ較正要求フラグ＝０とする。アクティブ較正要求フラグ＝０より、次回に較正実行可能と判定されたときには図６においてステップ１１１よりステップ３２〜３６へと進みパッシブ較正処理を実行する。
【０１９５】
ここで、本実施形態のパージ実行中の作用を図２８、図２９の波形図を参照しながら説明する。
【０１９６】
図２８の右側は、ベース空燃比エラーはなく、空燃比学習機能は考えないものとした状態で、パージ分補正係数ＦＨＯＳが正しく演算されているときの空燃比フィードバック補正係数α、パージ分補正係数ＦＨＯＳの各波形を、、これに対して図２８の左側は同じくベース空燃比エラーはなく、空燃比学習機能は考えないものとした状態で、較正エラーに伴ってＦＨＯＳが正しく演算されていないときの空燃比フィードバック補正係数α、パージ分補正係数ＦＨＯＳの各波形をモデル的に示す。ただし、運転条件は理想的な定常でなく、目標パージ率が緩やかに変化しているように緩やかに変化している場合で示している。
【０１９７】
図２８右側に示したように、設定された目標パージ率のもとでパージ処理を行っているとき、パージ分補正係数ＦＨＯＳが正しく演算されていれば、このＦＨＯＳにより理論空燃比からの空燃比偏差が小さくなるため、空燃比フィードバック制御が働くまでもないことから、空燃比フィードバック補正係数αは補正代がフル（最大）となる位置（つまりαの中心値である１００％）の付近にとどまっている。
【０１９８】
これに対して、較正エラーに伴ってＦＨＯＳが正しく演算されていないときには空燃比フィードバック制御中における理論空燃比からの空燃比偏差が大きくなり、この大きくなった空燃比偏差を小さなものにしようと、空燃比フィードバック制御が働く。すなわち、図２８左側に示したようにαが第１下限リミッタである７５％からは少し余裕を残した位置で緩やかに振れており、このことは、この状態で理論空燃比からの空燃比偏差が小さくなっていることを現している。言い換えると、ＦＨＯＳが誤って１００％に近い値に演算されたために排気空燃比が理論空燃比よりリッチ側に傾き、このリッチ側に傾いた排気空燃比を理論空燃比に戻そうとαが１００％より小さくなっている。
【０１９９】
この場合に、本実施形態では、αの第１下限リミッタである７５％よりもαの中心値（１００％）までの幅が狭い第２下限リミッタである１００−ＬＭＴ２を設定しており、図２８左側のようにこの第２下限リミッタを横切りながらαが推移するとき、パージ用下限リミッタ判定フラグが０から１になったり、再び０に戻ったりを繰り返す（図２８左側、下から第２段目参照）。これは、αとヒステリシス付きの判定値との比較による。すなわち、αが判定値である１００−ＬＭＴ２を横切って小さくなったタイミングでパージ用下限リミッタ判定フラグが０から１に切換わる（図２０ステップ１３３、１３４、１３５）。その後にαが今度は１００−ＬＭＴ２を横切って大きくなる側に変化すると、１００−ＬＭＴ２＋ＨＹＳ以下にある間はパージ用下限リミッタ判定フラグ＝１のままであるが（図２０ステップ１３３、１３６、１３５）、αが１００−ＬＭＴ２＋ＨＹＳより大きくなったタイミングでパージ用下限リミッタ判定フラグが１から０に切換わる（図２０ステップ１３３、１３６、１３７）。αが再び１００−ＬＭＴ２を横切って小さくなると、パージ用下限リミッタ判定フラグが０から１に切換わる（図２０ステップ１３３、１３４、１３５）。このような繰り返しによりパージ用下限リミッタ判定フラグが図２８左側の下から第２段目に示すようになる。
【０２００】
そして、パージ用下限リミッタ判定フラグ＝１となっている連続時間（タイマ２値）が所定値以上になるとアクティブ較正要求フラグが０から１に切換わる（図１９ステップ１２４、１２５）。
【０２０１】
図２９は空燃比学習制御を行っているものの、何らの原因でエンジン始動後にベース空燃比エラーが露わになっており、しかも空燃比学習が開始される前にあるとした状態で、なおかつ較正エラーに伴ってパージ分補正係数ＦＨＯＳが正しく演算されていないときの空燃比フィードバック補正係数α、パージ分補正係数ＦＨＯＳ、空燃比学習値αｍの各波形をモデル的に示す。すなわちｔ１以前ではベース空燃比エラーに起因して排気空燃比が理論空燃比よりリーン側に傾き、このリーン側に傾いた排気空燃比を理論空燃比に戻そうと空燃比フィードバック補正係数αが１００％より大きくなっている。ただし、運転条件は緩やかな加速時のものである。
【０２０２】
なお、空燃比学習値αｍはエンジンの回転速度と負荷とで表される全運転領域をいくつかの小領域に分割し、その各小領域毎に別々の空燃比学習値を格納している場合を想定しており、このため、最下段に示したようにｔ１とｔ２の間で空燃比学習値αｍが変化している。これは、運転条件が別の小領域に移ったために読み出される空燃比学習値が異なったことを示している。
【０２０３】
また、目標パージ率ＭＰＲ、パージ分補正係数ＦＨＯＳ、空燃比フィードバック補正係数αとも、図２８に示したように実際には演算周期毎に階段状に動く値であるが、簡単のため図２８と相違して図２９では直線で示している。また、ｔ１、ｔ３のタイミングは縦の２つの破線のうち左側の破線位置のタイミングであり、右側の破線位置のタイミングは次の演算タイミング（従って２つの破線の幅が演算周期）を表している。
【０２０４】
さて、ｔ１のタイミング以前でアクティブ較正要求フラグ＝１となっており、較正実行可能判定フラグが０より１に切換わるｔ１のタイミングでパージ率低下処理が開始される（図１３ステップ７５、７６、７７、７９）。このパージ率低下処理では、パージ率低下処理開始時の目標パージ率を初期値として所定時間当Δｔたり所定値Δずつ小さくされてゆき（図２２参照）、ｔ２で目標パージ率がゼロに達する（図２９第４段目参照）。この場合、パージ分補正係数ＦＨＯＳは目標パージ率に比例する値であるので、このように目標パージ率を低下させると、ＦＨＯＳもＦＨＯＳの中心値である１００％に向かって小さくなる（図２９第５段目参照）。
【０２０５】
ここで、パージ率低下処理はパージカットを行うための一方法であり、パージカット前であるパージ率低下処理開始時には、そのときのα、ＦＨＯＳ、ＭＰＲ、αｍ、Ｑｇが、対応するメモリα１、ＦＨＯＳ１、ＭＰＲ１、αｍ１、Ｑｇ１にパージカット前のデータとしてサンプリングされる（図２３ステップ１５１、１５２、１５３）。
【０２０６】
また、目標パージ率がゼロになるとパージカット状態となり、この状態ですぐにパージカット時のデータをサンプリングするのではなく、安定化時間が経過したｔ３のタイミングでのα、αｍが、対応するメモリα２、αｍ２にパージカット時のデータとしてサンプリングされる（図２４ステップ１６４、１６５、１６６、１６７）。
【０２０７】
これでパージカット前後のデータサンプリングが終了するので、データサンプリング終了フラグが０から１へと切換わり（図２４ステップ１６６、１６８）、このパージカット前後のデータを用いｔ４のタイミングでアクティブ較正処理が行われる。
【０２０８】
このアクティブ較正処理では、上記（５ａ）、（５ｂ）式に代えて、上記（１０）、（１１）式により全空燃比補正係数ＤＬＴと脱離量Ｄｇが算出され、このようにして得た脱離量Ｄｇより逆演算によって吸着量Ｙｒを演算し、キャニスタモデル中の吸着量のデータをこの吸着量Ｙｒに置き換える（図６ステップ１１１、１１２、１１３、１１４、１１５、１１６、１１７）。この結果、アクティブ較正処理後はパージ分補正係数ＦＨＯＳが正しく演算される。
【０２０９】
アクティブ較正処理が終了したあとは、アクティブ較正要求フラグ＝０、データサンプリングフラグ＝０となり（図６ステップ１１８、１１９）、これによって目標パージ率がパージカット前の値へと戻される（図１３ステップ７５、７６、７７、７８）。すなわち目標パージ率は、パージ率変化量制限値に従ってパージカット前の値へと徐々に戻され、この目標パージ率の増大を受けてパージ分補正係数ＦＨＯＳも中心値である１００％から離れて小さくなってゆく。このときのＦＨＯＳはアクティブ較正処理の実行によって正しく演算されるため、αがその中心値である１００％へと収束しており、パージ以外の外乱に備えることになる。
【０２１０】
また、アクティブ較正処理が終了したあとに較正実行可能となったときには先願装置と同様の較正処理であるパッシブ較正処理が行われる（図３０参照）。
【０２１１】
このようにして、本実施形態（請求項１に記載の発明）によれば、空燃比学習制御を行っているものの、何からの原因でエンジン始動直後にベース空燃比エラーが露わになっている場合には、パッシブ較正処理を継続している限りキャニスタモデルの較正にベース空燃比エラー分の誤差が生じ、これによってパージ分補正係数ＦＨＯＳが不正確にしか演算されず、これに起因して理論空燃比からの空燃比偏差が大きくなり、この空燃比偏差を小さくするために空燃比フィードバック制御が働いて空燃比フィードバック補正係数αの補正代が小さくなるのであるが、本実施形態によれば、第１リミッタよりも空燃比フィードバック補正係数αの中心値までの幅が狭い第２リミッタ±ＬＭＴ２を設定し、演算される空燃比フィードバック補正係数αと第２リミッタとの比較によりベース空燃比エラーがあるか否かを判定し、ベース空燃比エラーがあると判定した場合に較正タイミングとなったとき、パージカットを行い、キャニスタモデルに対してこのパージカット前後のデータに基づいてアクティブ較正処理（第１較正処理）を行うので、エンジン始動後にベース空燃比エラーが露わになっている場合において較正タイミングとなっても、キャニスタモデルの較正を、ベース空燃比エラーの影響を受けることなく精度良く行うことができ、これによってパージ分補正係数ＦＨＯＳの演算精度が向上すると共に、空燃比フィードバック補正係数αの補正代を無駄に使うことがない。
【０２１２】
また、運転条件が過渡的に変化するときには、ベース空燃比エラーがなくても、演算される空燃比フィードバック補正係数αが第２リミッタを含んで中心値の側に収まらないことがあり、このとき即座にベース空燃比エラーがあると判定したのでは誤判定になってしまうが、本実施形態（請求項７に記載の発明）によれば、所定値（遅れ時間）の経過後にベース空燃比エラーがあるか否かの判定を行うので（図１９ステップ１２１、１２２、１２３、１２４、１２５）、ベース空燃比エラーと通常の空燃比エラー（ベース空燃比エラーはないのに過渡などでの理論空燃比からの空燃比偏差をもつこと）との分離を行うことが可能になり、ベース空燃比エラーの判定精度が向上する。
【０２１３】
また、本実施形態（請求項８に記載の発明）によれば、ベース空燃比エラーがある場合に較正タイミングとなったときの目標パージ率を初期値として所定時間間隔で所定値ずつゼロまで低下させる処理（パージ率低下処理）を行うので、ベース空燃比エラーによる較正エラーが生じ、これに伴ってパージ分補正係数ＦＨＯＳが正しく演算されていない場合においても急激に目標パージ率を低下させることがなく、かつ空燃比フィードバック制御が可能な範囲で目標パージ率を変化させることができる。
【０２１４】
また、アクティブ較正処理によれば、キャニスタモデルの状態の推定をベース空燃比の影響を受けずに精度良く行うことが可能であるため、アクティブ較正処理の結果をパージ処理が終了していることを確認するために用いることができる。すなわち、エンジン始動後初めての較正タイミングでアクティブ較正処理が行われた後は、キャニスタモデルにより演算される脱離量Ｄｇが精度の良いものとなり、これによって積算パージ流量ＳＱＰと比較するための判定値ＳＬＳＱＰが正確に演算されるため、パージ処理が終了したか否かの判定精度が向上する（図１６ステップ９２、１００〜１０８）。
【０２１５】
空燃比学習制御の精度を向上させるためには、パージなどの外乱要因を取り除いた状態で行うことが望ましい。一方、燃料タンク２１から発生する蒸発燃料の処理を行うためにパージ処理は必要であり、始動直後より可能な限り大量のパージ処理を行うことが車両から大気へのパージ燃料の放出防止につながる。このように空燃比学習制御とパージ処理の２つの制御は相反する要求を持つが、パージ燃料を十分に処理できた時点においてはパージ率（パージ流量）を必要最低限の値に保持し（図１３ステップ８０、８１、図２５）、空燃比学習値の更新を行うことで（図２６ステップ１７１〜１７３、図２７ステップ１８１〜１８３）、空燃比学習制御の精度の保証を行うことができている。
【０２１６】
このように、本実施形態（請求項９に記載の発明）によれば、アクティブ較正処理の結果をパージ処理の終了判定に活かすことで、パージ処理終了判定の精度を高め、このようにして高い精度でパージ終了判定を行った後で、パージ率を小さくして空燃比学習値の更新を行うので、空燃比学習制御の精度向上と始動直後よりの大量のパージ処理とを両立させることができる。
【０２１７】
さらに、キャニスタ２４からの燃料の脱離特性は、キャニスタ内の燃料吸着剤（活性炭等）の温度の影響を受けるので、脱離量演算式は吸着剤温度と脱離量の関係を考慮した式とする（請求項１５に記載の発明）。脱離量が多い領域では脱離量の演算誤差が排気空燃比に及ぼす影響が大きく特に高い演算精度が要求され、しかも、キャニスタ２４からの燃料の脱離現象が吸熱反応であるため、大量の脱離が行われるときはキャニスタ内の吸着剤温度が低下し脱離特性も変化する。そのため、吸着剤温度と脱離量の関係を考慮しないと脱離量が多い領域では脱離量の演算精度が低下するが、吸着剤温度も考慮に入れて脱離量を演算するようにキャニスタモデルを構成すれば、脱離量の演算精度の悪化を抑え、演算精度の悪化に起因する排気エミッション悪化等を防止できる。
【０２１８】
なお、吸着燃料量とパージ率と吸着剤温度とに基づき脱離燃料量を演算する脱離量演算式としては、フロイントリッヒ（Ｆｒｅｕｎｄｌｉｃｈ）の式をベースにキャニスタ２４の脱離現象に応用した式を用いることができる。また、吸着剤温度はキャニスタ２４からの燃料の脱離量に基づき吸着剤温度の降下量を求めることによって求めることができる（請求項１６に記載の発明）。
【０２１９】
また、本実施形態（請求項１１廃止に記載の発明）によれば、ベース空燃比エラーがない場合に較正タイミングとなったとき、キャニスタモデルに対してパージ実行中のデータに基づくパッシブ較正処理（第２較正処理）も行われる。キャニスタモデルは近似モデルであり、それによって演算される値は幾らかの誤差を含んだものとなり、また、キャニスタモデルは前回値を用いて新たな値を演算するため、その動作時間が長くなるにつれ誤差が積分され大きくなるが、このようなパッシブ較正処理を行うことで誤差を適宜修正し、高い演算精度を保つことができる。なお、このパッシブ較正処理は、理論空燃比からの空燃比偏差から推定される脱離量から吸着量を逆算し、内部変数の吸着量をこの逆算された吸着量で較正するというものであり、特殊なシーケンスを踏む必要が無く、パージ率を落とす必要もない。なお、吸着量はキャニスタモデルにおける脱離量演算式の逆演算によって求めることができる（請求項１７に記載の発明）。
【０２２０】
上記較正処理はパージ以外の外乱による理論空燃比からの空燃比偏差が生じないときに行われ、理論空燃比からの空燃比偏差を全てパージによるものとみなしてキャニスタからの燃料の脱離量を推定し、この推定脱離量から逆算したキャニスタ２４の実吸着量に基づき較正が行われるので、較正処理の精度が確保できる（請求項１８に記載の発明）。
【０２２１】
また、本実施形態（請求項１９、２０に記載の発明）によれば、脱離量の演算精度を高めるべく、較正処理はパージによる理論空燃比からの空燃比偏差が比較的大きなとき（パージ率が高く、パージ空燃比が濃いとき）に行われる。運転領域によっては十分な較正処理の演算精度が得られない場合があるが、このように十分な精度が得られる領域のみで較正処理を実行することにより、較正処理の精度を向上させ、キャニスタモデルの精度をさらに向上させることができる。
【０２２２】
以上、本発明の実施の形態について説明したが、上記実施形態の構成は本発明が適用される構成の一例を示したものであり、本発明の範囲を上記構成に限定するものではない。上述した通り、上記実施形態においては、キャニスタモデルによるパージ処理が可能となるまでは、図１６に示したパージ処理が補助的に実行されるが、図１６に示したパージ処理を継続して用いるようにしてもよい。
【図面の簡単な説明】
【図１】本発明に係る蒸発燃料処理装置の全体構成図。
【図２】エンジンコントローラにおけるパージ処理の概要を示したブロック図。
【図３】キャニスタモデルの構成を示したブロック図。
【図４】パージ分補正係数の演算を説明するためのフローチャート。
【図５】較正処理可能条件を示したフローチャート。
【図６】較正処理を説明するためのフローチャート。
【図７】キャニスタモデルに基づく脱離量の演算を説明するためのフローチャート。
【図８】遅れ補正を説明するためのフローチャート。
【図９】吸入空気流量と無駄時間の関係を示す特性図。
【図１０】吸入空気流量となまし係数の関係を示す特性図。
【図１１】遅れ補正における無駄時間処理の概要を示した図。
【図１２】目標パージ率の設定を説明するためのフローチャート。
【図１３】目標パージ率の設定を説明するためのフローチャート。
【図１４】吸入空気流量及びパージ率に対するパージ空燃比誤差の関係を示す特性図。
【図１５】パージ空燃比に対するパージ空燃比誤差の関係を示す特性図。
【図１６】ブートアップ制御を説明するためのフローチャート。
【図１７】空燃比フィードバック偏差の差（＝目標空燃比フィードバック偏差−実空燃比フィードバック偏差）とパージ率変化量の関係を示す特性図。
【図１８】パージ空燃比に対するパージ終了判定値の関係を示す特性図。
【図１９】アクティブ較正要求フラグの設定を説明するためのフローチャート。
【図２０】パージ用下限リミッタ判定フラグの設定を説明するためのフローチャート。
【図２１】パージ用上限リミッタ判定フラグの設定を説明するためのフローチャート。
【図２２】パージ率低下処理を説明するための波形図。
【図２３】パージ率低下処理を説明するためのフローチャート。
【図２４】パージ率低下処理を説明するためのフローチャート。
【図２５】パージ終了が判定されたときの目標パージ率の変化を示す波形図。
【図２６】空燃比学習開始フラグの設定を説明するためのフローチャート。
【図２７】空燃比学習値の更新を説明するためのフローチャート。
【図２８】正常に較正が行われたときと較正エラーがあるときとを比較して説明するための波形図。
【図２９】アクティブ較正処理の内容を示す波形図。
【図３０】パッシブ較正処理の内容を示す波形図。
【図３１】較正処理の概要を示したブロック図。
【符号の説明】
４　燃料噴射弁
５　排気通路
８　吸気通路
１１　エンジンコントローラ
１３　エアフローメータ
１５　空燃比センサ（空燃比検出手段）
２１　燃料タンク
２２　配管
２４　キャニスタ
２６　配管
２７　パージバルブ[0001]
[Industrial applications]
The present invention relates to an air-fuel ratio control device for an engine, and more particularly to an engine having an evaporated fuel processing device.
[0002]
[Prior art]
In the engine, the fuel vapor generated in the fuel tank while the engine is stopped is temporarily adsorbed on the activated carbon in the canister, and the intake pipe pressure (pressure lower than the atmospheric pressure) is used under predetermined operating conditions after the engine is started. A fuel vapor treatment device is provided for desorbing the fuel adsorbed on the activated carbon and guiding it to the intake passage downstream of the intake throttle valve for combustion treatment (see Japanese Patent Application Laid-Open No. 7-166978).
[0003]
In the following, "purge" or "purge processing" refers to desorbing fuel adsorbed in the canister and supplying it to the engine, and "purge fuel" refers to fuel desorbed from the canister, along with purge fuel as well as purge fuel. Fresh air supplied to the engine may also be referred to as “purge gas”.
[0004]
In such a conventional device, the effect of the purge on the exhaust air-fuel ratio is treated as a disturbance, and the oxygen concentration sensor provided in the exhaust passage is used even if the exhaust air-fuel ratio may deviate from the stoichiometric air-fuel ratio due to the disturbance. The air-fuel ratio feedback control is performed to return the exhaust air-fuel ratio to the stoichiometric air-fuel ratio, so that the conversion efficiency of the three-way catalyst provided in the exhaust passage is not reduced.
[0005]
[Problems to be solved by the invention]
By the way, since the change of the air-fuel ratio feedback correction coefficient α (air-fuel ratio feedback correction amount) due to the supply of the purge fuel to the engine can be predicted, the predictable air-fuel ratio feedback correction coefficient α is used as the purge correction coefficient FHOS (purge The purge correction coefficient FHOS corrects the basic value of the fuel amount supplied to the engine so that the air-fuel ratio deviation from the stoichiometric air-fuel ratio due to the purge is reduced.
(1) An equation for calculating the amount of fuel adsorbed on the canister,
(2) an equation for calculating the amount of fuel desorbed from the canister based on the amount of adsorbed fuel and the target purge rate;
An apparatus for estimating the amount of fuel desorbed from the canister by using the canister model composed of the above, and calculating the purge correction coefficient FHOS based on the desorbed fuel amount, the target purge rate and the intake air mass Was previously proposed by the same applicant as the present application (see Japanese Patent Application No. 2001-71562).
[0006]
According to the purging process (model reference purging process) by the prior application apparatus, the disturbance due to the purging is exclusively associated with the purge correction coefficient FHOS, and there is no base air-fuel ratio error and the FHOS is accurately calculated. For example, even if the purge process is performed before the start of the air-fuel ratio feedback control, the exhaust air-fuel ratio is controlled to the stoichiometric air-fuel ratio. That is, when the air-fuel ratio feedback control condition is satisfied during the purge process and the air-fuel ratio feedback control is started, the air-fuel ratio feedback correction coefficient α moves as if there is no purge.
[0007]
For the air-fuel ratio feedback correction coefficient α, a first limiter that defines a limit of the correction margin is provided in advance, for example, ± 25%, and α is increased from 100%, which is the central value of α, to 125% (upper limiter). According to the conventional device, the influence of the purge on the exhaust air-fuel ratio is determined by the feedback of the air-fuel ratio, where 75% (lower limiter) of 100% is the full correction allowance for decreasing α. Since the correction coefficient α compensated, the amount of correction of α for disturbance was reduced by that much.However, according to the prior application, the air-fuel ratio feedback correction coefficient α became free with respect to purge, and α Has been fully used for disturbances other than purge.
[0008]
Also, by setting the maximum purge rate as a target purge rate within a certain limit depending on the flow rate of the purge valve or the like, a large amount of purge can be performed earlier than immediately after the engine is started.
[0009]
However, since the canister model is merely an approximation model, the values calculated using the canister model (desorption amount, adsorption amount, etc.) are slightly different from the actual values. Further, since the canister model is a recurrence formula as described later, as the operation time of the canister model increases, the error is integrated, and the difference between the calculated value and the actual value increases. Therefore, in order to eliminate this deviation and maintain the calculation accuracy of the canister model high, at the calibration timing, the value of the adsorption amount of the canister, which is one of the internal variables of the canister model, is calibrated based on the data during the purge process. are doing.
[0010]
On the other hand, the deviation of the base air-fuel ratio from the stoichiometric air-fuel ratio due to the deterioration and dispersion of components such as the air flow meter and the fuel injection valve with the lapse of time and the non-linearity of the injection pulse width-flow rate characteristics of the fuel injection valve are defined as the base air-fuel ratio error. In other words, in order to eliminate this base air-fuel ratio error, learning control of the air-fuel ratio is performed during stoichiometric operation (operation during which the stoichiometric air-fuel ratio is set as the target air-fuel ratio).
[0011]
When the air-fuel ratio learning control is combined with the purging process of the prior application, which performs a large amount of purging immediately after the engine is started, the following problem occurs. In other words, if the air-fuel ratio learning control is working properly even if there is a base air-fuel ratio error, the air-fuel ratio learning value absorbs the base air-fuel ratio error, so that there is no base air-fuel ratio error. If there is a significant base air-fuel ratio error immediately after the start of the engine for some reason, it is possible that the correction allowance by the air-fuel ratio feedback correction coefficient α alone is not enough to make the correction. If it cannot be corrected, exhaust emissions will deteriorate.
[0012]
On the other hand, since the purging is required to flow in large quantities after the start for the purpose of reducing the deterioration of exhaust emission due to the purging, it is desirable to start the purging process early as in the prior application.
[0013]
Therefore, when the calibration timing comes after a large amount of purge is started immediately after the start in a state where the base air-fuel ratio error has occurred, the calibration process is performed by regarding the base air-fuel ratio error as being due to the purge. Therefore, a calibration error larger by the base air-fuel ratio error occurs, and the purge correction coefficient FHOS is not accurately calculated. That is, an error occurs in the purge correction coefficient FHOS under the influence of the calibration error accompanying the base air-fuel ratio error. Even when the error occurs in the purge correction coefficient FHOS, the air-fuel ratio feedback control is performed. If it is medium, the air-fuel ratio deviation from the stoichiometric air-fuel ratio due to the purge is eliminated by the air-fuel ratio feedback correction coefficient α.
[0014]
This is because, according to the prior application, the effect of the purge on the exhaust air-fuel ratio was compensated by the purge correction coefficient FHOS, so that the correction allowance of the air-fuel ratio feedback correction coefficient α could be fully used. If an error occurs in the purge correction coefficient FHOS due to a calibration error in a state where the air-fuel ratio error is exposed, it means that the correction allowance of the air-fuel ratio feedback correction coefficient α becomes narrow again.
[0015]
Therefore, the present invention is based on the prior application, and the second limiter as a purge limiter is provided between the center value of the air-fuel ratio feedback correction coefficient α and the original limiter (first limiter) for the air-fuel ratio feedback correction coefficient α. It is determined whether there is a base air-fuel ratio error by comparing the calculated air-fuel ratio feedback correction coefficient α with the second limiter, and when it is determined that there is a base air-fuel ratio error, the calibration timing is reached. By performing a purge cut and performing a calibration process (first calibration process) on the canister model based on the data before and after the purge cut, the canister model is calibrated in a state where a base air-fuel ratio error is exposed immediately after starting. Even when timing is reached, the effects of base air-fuel ratio errors are eliminated and the accuracy of canister model calibration is improved. Keeping (calculation accuracy FHOS also kept high) that is an object.
[0016]
[Means for solving the problem]
According to the first aspect of the present invention, the air-fuel ratio detecting means for detecting the air-fuel ratio in the exhaust gas and the air-fuel ratio feedback correction amount are adjusted so that the exhaust air-fuel ratio detected by the air-fuel ratio detecting means coincides with the stoichiometric air-fuel ratio. An air-fuel ratio feedback correction amount calculating means for calculating; a first limiter setting means for setting a first limiter having a predetermined width from a center value of the air-fuel ratio feedback correction amount; and an air-fuel ratio feedback correction amount including the first limiter. Air-fuel ratio feedback correction amount limiting means for limiting the calculated air-fuel ratio feedback correction amount so as to fall within the center value side, and an air-fuel ratio for correcting the amount of fuel supplied to the engine with the limited air-fuel ratio feedback correction amount An air-fuel ratio control device for an engine, comprising: a canister that adsorbs fuel vapor generated in a fuel tank; A purge valve for opening and closing a pipe that communicates the intake passage of the canister and the engine, a purge valve control means for controlling the purge valve so that the target purge ratio is obtained,
at least,
(A) an adsorbed fuel amount calculation formula for calculating the amount of fuel adsorbed on the canister based on the previous value of the amount of fuel adsorbed on the canister and the previous value of the amount of fuel desorbed from the canister;
(B) a desorbed fuel amount calculation formula for calculating the amount of fuel desorbed from the canister based on the adsorbed fuel amount calculated by the calculated fuel amount calculation formula and the target purge rate;
And a canister model composed of
Purge correction that calculates the purge correction based on the desorbed fuel calculated using the canister model so that the air-fuel ratio deviation from the stoichiometric air-fuel ratio is reduced by performing the purge processing at the target purge rate. An amount calculating means, a purge amount correcting means for correcting the amount of fuel supplied to the engine with the purge amount correcting amount, a calibration timing determining means for determining whether or not a calibration timing is reached, and an air-fuel ratio feedback rather than a first limiter. Means for setting a second limiter having a narrow width up to a central value of the correction amount, and a base air-fuel ratio for determining whether there is a base air-fuel ratio error by comparing the calculated air-fuel ratio feedback correction amount with the second limiter Error determination means, purge cut means for performing a purge cut when the calibration timing comes when there is a base air-fuel ratio error from the determination result, And a first calibration processing means for performing first calibration processing based on the purge cut before and after the data to Yanisutamoderu.
[0017]
[Action and effect]
If the base air-fuel ratio error is exposed immediately after the start of the engine, it rarely occurs.However, in the rare case, since it was not considered in the prior application, calibration was performed in that rare case. When the timing comes, as long as the calibration process (second calibration process) of the prior application is continued, the calibration process cannot eliminate the calibration error associated with the base air-fuel ratio error. Since the correction amount for the purge cannot be correctly calculated due to the calibration error, the air-fuel ratio deviation from the stoichiometric air-fuel ratio increases, and the correction allowance for the air-fuel ratio feedback correction amount narrows to reduce the air-fuel ratio deviation. On the other hand, according to the first aspect of the present invention, when the calibration timing is reached when the base air-fuel ratio error is exposed immediately after the start of the engine, the purge cut is performed, and the canister model is removed. Since the first calibration process is performed based on the data before and after the purge cut, even when the base air-fuel ratio error is exposed after the engine is started, the calibration of the canister model can be performed even when the calibration timing comes. The correction can be performed with high accuracy without being affected by an error, thereby improving the calculation accuracy of the correction amount for the purge amount, and avoiding wasteful use of the correction allowance for the air-fuel ratio feedback correction amount.
[0018]
BEST MODE FOR CARRYING OUT THE INVENTION
Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings.
[0019]
FIG. 1 shows the overall configuration of an engine provided with an evaporated fuel processing device.
[0020]
Reference numeral 1 denotes an engine body, and a fuel injection valve 4 is provided for each cylinder in the intake passage 2 downstream of the intake throttle valve 3. The fuel injection valve 4 has a predetermined air-fuel ratio according to an operating condition according to an injection signal from the engine controller 11. As described above, the fuel is injected and supplied during the intake. Specifically, a signal from the crank angle sensor 12, a signal of the intake air flow rate from the air flow meter 13, a signal of the engine cooling water temperature from the water temperature sensor 14, and the like are input to the engine controller 11 mainly composed of a microcomputer, The controller 11 calculates a basic injection pulse width Tp for obtaining an air-fuel mixture of a basic air-fuel ratio (stoichiometric air-fuel ratio) based on these signals, performs various corrections such as an increase in water temperature, and calculates a fuel injection pulse width Ti. Then, the fuel injection valve 4 is opened at a predetermined timing for a period corresponding to the Ti.
[0021]
The exhaust passage 5 includes a three-way catalyst (not shown). The three-way catalyst performs reduction of NOx in the exhaust gas and oxidation of HC and CO with maximum conversion efficiency when the air-fuel ratio in the exhaust gas is in a narrow range around the stoichiometric air-fuel ratio. Therefore, the engine controller 11 calculates the air-fuel ratio feedback correction coefficient α so that the air-fuel ratio in the exhaust gas detected by the air-fuel ratio sensor 15 matches the stoichiometric air-fuel ratio. The feedback control of the air-fuel ratio is performed by correcting the basic injection pulse width Tp (the basic value of the amount of fuel supplied to the engine) with the correction amount).
[0022]
In this case, for the air-fuel ratio feedback correction coefficient α, a first limiter ± LMT1 (± 25%) having a predetermined width around the center value of α (100%) is previously provided as a range in which α can move, When the calculated α becomes larger than the center value and exceeds the upper limit limiter 100 + LMT1, this upper limit limiter is used. When the calculated α becomes smaller than the center value and becomes smaller than the lower limit limiter 100−LMT1, this lower limiter is used. Limited.
[0023]
In addition, in order to eliminate base air-fuel ratio errors caused by deterioration with time and variations of components such as the air flow meter 13 and the fuel injection valve 4 and non-linearity of the injection pulse width-flow rate characteristic of the fuel injection valve 4, the engine controller 11 During the stoichiometric operation (operation during which the stoichiometric air-fuel ratio is set to the target air-fuel ratio), learning control of the air-fuel ratio is performed.
[0024]
On the other hand, the evaporative fuel processing device is for processing the evaporative fuel generated in the fuel tank 21, and includes a canister 24, a pipe 22 communicating the canister 24 and the fuel tank 21, a canister 24 and an intake throttle valve 3 of the engine. A pipe 26 communicating with the downstream intake passage 2 and a purge valve 27 for opening and closing the pipe 26 are provided.
[0025]
The fuel vapor generated in the fuel tank 21 is guided to the canister 24 via the pipe 22, and only the fuel component is adsorbed by the fuel adsorbent (activated carbon) 24 a in the canister 24, and the remaining air is released from the air release port 25. Released outside. In order to burn the fuel adsorbed on the activated carbon 24a, the purge valve 27 is opened, and fresh air is introduced into the canister 24 from the atmosphere release port 25 by using the intake pipe pressure (pressure lower than the atmosphere) developed downstream of the intake throttle valve 3. Is introduced. As a result, the fuel adsorbed on the activated carbon 24a by the fresh air is desorbed, and introduced into the intake passage 2 of the engine 1 via the pipe 26 together with the fresh air.
[0026]
When performing a process (purge process) of desorbing the fuel adsorbed by the canister 24 and supplying the desorbed fuel to the engine (purge process), the engine controller 21 sets the target purge rate as high as possible within a range that does not deteriorate the engine combustion stability and exhaust emission. (The ratio of the purge flow rate to the intake air flow rate) is set, and the opening degree of the purge valve 27 is controlled so that the target purge rate is realized after the engine is started.
[0027]
Furthermore, after the purge process is started, the purge becomes a disturbance, and an air-fuel ratio deviation from the stoichiometric air-fuel ratio occurs. In the initial stage of the purge, the exhaust air-fuel ratio deviates from the stoichiometric air-fuel ratio to the rich side, so the air-fuel ratio feedback correction coefficient α moves to a side smaller than the center value until the deviated air-fuel ratio returns to the stoichiometric air-fuel ratio. When the target purge rate is set as high as possible after the start-up, and when performing a large amount of purge, α moves close to 75% which is the lower limit limiter of α.
[0028]
In this case, if disturbance other than purge occurs and the exhaust air-fuel ratio deviates from the stoichiometric air-fuel ratio and becomes rich, α moves to a smaller side, but immediately reaches the lower limiter and becomes smaller thereafter. No. In other words, since the correction allowance for α to reduce α to the disturbance due to the purge has already been almost exhausted, only a small amount of correction remains for disturbances other than the purge. If the exhaust air-fuel ratio cannot be returned to the stoichiometric air-fuel ratio even if it is used, the exhaust emission deteriorates.
[0029]
For this reason, the engine controller 11 estimates the amount of desorbed fuel from the canister using a canister model described later, and based on the estimated desorbed fuel amount Dg, the target purge rate, and the intake air mass, a purge correction coefficient. By calculating the FHOS and correcting the basic injection pulse width Tp with the purge correction coefficient FHOS, the target air purge ratio is set as high as possible after the engine is started, and even if a large amount of purge is performed, the stoichiometric air-fuel ratio is maintained. The deviation of the air-fuel ratio from is reduced. That is, since the effect of the purge on the exhaust air-fuel ratio is compensated by the purge correction coefficient FHOS, the air-fuel ratio feedback correction coefficient α operates in the same manner as when there is no purge. It is not necessary to use the correction allowance for the disturbance, and the correction allowance remains full, and has strong robustness against disturbances other than the purge.
[0030]
As will be described later, the change in the air-fuel ratio feedback correction coefficient α can be predicted by supplying the desorption amount Dg calculated by the canister model to the engine, and the predicted α is the purge correction coefficient FHOS. is there. Therefore, the predicted α increases as the target purge rate increases, increases as the desorption amount Dg increases, and decreases as the intake air mass Qg increases. Therefore, the purge correction coefficient FHOS also increases as the target purge rate increases, increases as the desorption amount Dg increases, and decreases as the intake air mass Qg increases.
[0031]
In addition, it is necessary to perform a calibration process on the canister model, and at that time, a calibration error occurs. The biggest cause of the calibration error comes from the base air-fuel ratio error. This is because when performing the calibration process of the canister model, it is assumed that the air-fuel ratio deviation from the stoichiometric air-fuel ratio is caused entirely by the purge, and even if the air-fuel ratio learning control is performed. However, if the calibration process is performed just because the base air-fuel ratio error is exposed immediately after the engine is started for some reason, the air-fuel ratio feedback correction coefficient α is changed due to the base air-fuel ratio error. This is because the calibration process is performed as if the purging was performed entirely.
[0032]
Therefore, the engine controller 11 sets the second limiter ± LMT2 which is narrower than the first limiter ± LMT1 to the center value of the air-fuel ratio feedback correction coefficient α, and calculates the calculated air-fuel ratio feedback correction coefficient α and the second limiter. It is determined whether or not there is a base air-fuel ratio error by the comparison.If it is determined that there is a base air-fuel ratio error, when the calibration timing comes, the calibration process is not the same as that of the prior application, but the base air-fuel ratio error is determined. Perform a new calibration process in an unaffected manner.
[0033]
The pipe 22 includes a bypass valve 29 that bypasses a vacuum cut valve 23 that is opened when the pressure in the passage on the fuel tank 21 side becomes lower than the atmospheric pressure. The pipe 26 includes a pressure sensor 30 that measures the pressure in the pipe. Drain cut valves 28 are provided at the atmosphere release ports 25 of the canisters 24, respectively. These drain valves 28 are required for diagnosis of the evaporating apparatus and have no direct relation to the purging processing.
[0034]
The control performed by the engine controller 11 will first be outlined with reference to the block diagrams shown in FIGS.
[0035]
FIG. 2 is a block diagram schematically showing a part related to the purge processing and the air-fuel ratio control. In FIG. 2, a target purge rate setting unit B1 calculates a maximum purge rate that can be set in the current operation area based on performance limits of components related to the purge processing, and sets the target purge rate so as to follow this maximum purge rate. Set. However, an abrupt change in the purge rate causes a large change in the exhaust air-fuel ratio (increase in the air-fuel ratio deviation from the stoichiometric air-fuel ratio) and causes a deterioration in the exhaust emission. As described above, the amount of change in the purge rate is limited to a predetermined amount (purge rate change amount limit value) or less.
[0036]
The duty ratio calculator B2 calculates the duty ratio of the purge valve 27 necessary to achieve the target purge rate, and the purge valve driver B3 drives the purge valve 27 at the duty ratio calculated by the duty ratio calculator B2.
[0037]
On the other hand, in the desorption amount calculation unit B4, the amount of fuel desorbed from the canister 24 when purging is performed at the target purge rate using a physical model of the canister 24 (hereinafter, referred to as a canister model) described later. Is calculated.
[0038]
The purge amount correction coefficient calculation unit B5 determines the purge amount correction coefficient based on the estimated desorption amount (desorption fuel amount), the target purge rate, and the intake air flow rate such that the air-fuel ratio deviation from the stoichiometric air-fuel ratio due to the purge becomes small. Calculate FHOS. The delay correction unit B6 performs a delay correction including a dead time correction and a smoothing process in consideration of the intake air flow rate on the purge correction coefficient FHOS, and sets the correction coefficient after the delay correction as the purge correction coefficient FHOS. .
[0039]
The basic injection pulse width calculation unit B7 calculates the basic injection pulse width Tp from the intake air flow rate and the engine rotation speed, and the air-fuel ratio feedback correction coefficient calculation unit B8 calculates the air-fuel ratio feedback correction coefficient α based on the actual exhaust air-fuel ratio. Then, the fuel injection pulse width calculation unit B9 performs various corrections and the air-fuel ratio feedback correction coefficient α on the basic injection pulse width Tp as described above, and also uses the purge correction coefficient FHOS after the delay correction. Correction is performed to calculate the final fuel injection pulse width Ti. The fuel injection valve driving section B10 drives the fuel injection valve 4 with the fuel injection pulse width Ti calculated in this manner.
[0040]
As shown in FIG. 3, the canister model includes an adsorption amount calculation unit B21, a reference desorption amount calculation unit B22, a purge flow equivalent desorption amount calculation unit B23, and an activated carbon temperature calculation unit B24. Each calculation is performed every time.
[0041]
The adsorption amount calculation unit B21 calculates the current value Y of the amount of fuel adsorbed on the canister 24 by the following equation.
[0042]
[Adsorption amount calculation formula]
Y = Yz-Dgz (1)
Here, Y: the current value of the adsorption amount,
Yz: previous value of adsorption amount,
Dgz: previous value of desorption amount,
This adsorption amount calculation formula is to calculate the current (current) adsorption amount Y [g] by subtracting Dgz, which is the fuel amount desorbed last time, from the previous value Yz of the adsorption amount (adsorbed fuel amount).
[0043]
The reference desorption amount calculation unit B22 calculates the desorption amount Dgk [g] at the time of the reference purge flow rate by the following equation.
[0044]
[Calculation formula for desorption amount at standard purge flow rate]
Dgk = (Y / A) ＾ n (T) (2)
Here, Y: adsorption amount,
A: desorption constant,
n (T): desorption index,
T: activated carbon temperature,
This equation is based on the concept of the adsorption / desorption phenomenon (Freundrich's equation) applied to the fuel desorption phenomenon from the canister 24, whereby the fuel desorption characteristics from the canister 24 can be expressed almost exactly. The Freundlich equation is described in "Theory on Surfaces II" (Maruzen, Tsukada), pp. 25-27 and p. 108-115.
[0045]
In the purge flow rate equivalent desorption amount calculation unit B23, the desorption amount Dg [g] is calculated from the following equation.
[0046]
[Desorption amount calculation formula according to purge flow rate]
Dg = K × PQ × Dgk (3)
Where K: constant,
PQ: purge flow rate (= target purge rate x intake air flow rate),
Dgk: Desorption amount at reference flow rate,
Since the purge flow rate PQ and the desorption amount are substantially proportional to each other, the desorption amount calculation formula according to the purge flow rate PQ calculates the desorption amount by linear approximation. Here, the desorption amount Dgk at the reference flow rate is obtained by equation (2), and the desorption amount Dg is calculated by multiplying this by the purge flow rate PQ in equation (3). , (3) may be combined into one expression.
[0047]
The activated carbon temperature calculating section B24 calculates the activated carbon temperature T [K] by the following equation.
[0048]
[Activated carbon temperature calculation formula]
T = Tz−Kt1 × (Yz2−Yz) + Kt2 × (Tz−Ta) (4)
Here, T: the current value of the activated carbon temperature,
Tz: previous value of activated carbon temperature,
Kt1: Endothermic coefficient,
Yz2: value of the amount of adsorption two times before,
Yz: previous value of adsorption amount,
Kt2: heat transfer component coefficient,
Ta: ambient temperature of the canister,
The activated carbon temperature calculation formula is obtained by calculating the temperature in the past (the first term on the right side), the temperature drop (the second term on the right side) in proportion to the desorption amount (Yz2−Yz), and the temperature rise due to heat transfer (the third term on the right side). Section). The reason for calculating the activated carbon temperature T in this way is that the desorption index n (T) in the equation (2) is affected by the activated carbon temperature T. In particular, when the desorption amount is large, the amount of drop of the activated carbon temperature T is large. This is because the influence on the desorption characteristics of the fuel in the canister 24 cannot be ignored.
[0049]
Therefore, the canister model is composed of the four expressions (1) to (4), and three expressions when the expressions (2) and (3) are put together.
[0050]
Returning to FIG. 2, the canister model expresses the desorption characteristics of the canister with high accuracy as described above. However, since it is an approximate model, values calculated using this model (desorption amount, adsorption amount, etc.) ) Is a value slightly shifted from the actual value. Further, since the canister model is a recurrence formula using the previous calculation result as shown in the above equation (1), the error is integrated as the model operation time becomes longer, and the deviation between the calculated value and the actual value increases. I do.
[0051]
Therefore, in order to calibrate this deviation and maintain the calculation accuracy of the canister model high, if the calibration timing determination unit B11 determines that it is time to execute the calibration process, the calibration processing unit B12 is one of the internal variables of the canister model. The value of the adsorption amount of the canister 4 is calibrated. Specifically, as shown in FIG. 31, the calibration timing determination unit B11 regards almost all deviations from the central value (100%) of the air-fuel ratio feedback correction coefficient α during the air-fuel ratio feedback control to be due to the purge. When the condition that can perform the calibration process is satisfied, it is determined that it is the timing at which the calibration process can be executed, and the switch B13 is turned ON. The data sampling unit B32 causes the air-fuel ratio feedback correction coefficient α and the purge amount correction at that timing (during the purging process). The coefficient FHOS, the target purge rate MPR, and the intake air mass Qg are read, the desorption amount calculation unit B33 estimates the desorption amount of fuel from the canister 24 based on the data during the execution of the purge, and the adsorption amount calculation unit B34 The adsorption amount is calculated back from the estimated desorption amount. Then, the calibrating unit B35 calibrates the value of the adsorption amount of the canister model with this value.
[0052]
On the other hand, in the case where the air-fuel ratio learning control unit B16 (see FIG. 2) is provided and the base air-fuel ratio error is exposed immediately after the engine is started for some reason, even when the calibration timing is reached, the timing is also satisfied. Similarly to the prior application, when the calibration process based on the data during the execution of the purge (hereinafter, this calibration process is referred to as “passive calibration process”) is performed, the amount of change in the air-fuel ratio feedback correction coefficient α due to the base air-fuel ratio error is also increased. In this embodiment, the calibration process is performed on the basis of the data before and after the purge cut, separately from the passive calibration processing unit B31 including blocks B32 to B35. An active calibration processing unit B41 for performing the “calibration processing”) is additionally provided.
[0053]
Here, the active calibration processing unit B41 includes a pre-purge cut data sampling unit B42, a purge cut data sampling unit B43, a desorption amount calculation unit B44, an adsorption amount calculation unit B45, and a calibration unit B46. In order to operate, a base air-fuel ratio error determination section B36, an active calibration request section B37, an AND circuit B38, a purge rate reduction processing section B39, and changeover switches B40 and B47 are provided.
[0054]
That is, the base air-fuel ratio error determination unit 36 determines whether or not there is a base air-fuel ratio error based on the air-fuel ratio feedback correction coefficient α and the second limiter ± LMT2. The calibration request unit B37 outputs a high-level signal to the AND circuit B38. In the AND circuit B38, only when the signal from the active calibration request unit B37 is at the high level and the signal from the calibration timing determination unit is at the high level, the two changeover switches B40 and B47 are switched from the illustrated state to the opposite side. Switch.
[0055]
The purge rate lowering processing section 39 performs a process of reducing the target purge rate MPR immediately before the changeover of the changeover switch B40 by a predetermined value per a predetermined time as an initial value and finally reducing the target purge rate to zero. The target purge rate decreasing in this manner is output to the changeover switch 15 via the changeover switch B40 after the changeover of the changeover switch B40.
[0056]
At the time of the purge rate lowering processing by the purge rate lowering processing section 39, at the start of the purge rate lowering processing, the data sampling section B42 before purge cut performs the air-fuel ratio feedback correction coefficient α, the air-fuel ratio learning value αm, and the purge correction coefficient FHOS at that time. , The target purge rate MPR and the intake air mass Qg are sampled as data before purge cut. Further, when the target purge rate MPR becomes zero due to the purge rate lowering process, the purge cut time data sampling unit B43 samples α and αm at that time as purge cut data.
[0057]
The desorbing unit calculation unit B44 estimates the amount of fuel desorbed from the canister 24 using the data before and after the purge cut, and the adsorption amount calculation unit B45 calculates the adsorption amount from the estimated desorption amount. After the changeover of the changeover switch B47, the calibrating unit B46 calibrates the value of the suction amount of the canister model with the calculated value via the changeover switch B47.
[0058]
Hereinafter, the specific contents of the purge processing and the air-fuel ratio control performed by the engine controller 11 will be described below with reference to the flowchart.
[0059]
FIG. 4 is for calculating the purge correction coefficient, and is executed at regular intervals (for example, every 10 msec).
[0060]
In step 1, it is determined whether or not a purge permission condition is satisfied. If the purge permission condition is not satisfied, the routine proceeds to step 2, where the purge correction coefficient FHOS is set to 1.0, and the current process is terminated. Step 1 may be omitted, and the determination in step 3 may be performed immediately after the start.
[0061]
When the purge permission condition is satisfied, the process proceeds to step 3, and it is determined whether or not it is a timing at which the calibration process of the adsorption amount, which is an internal variable of the canister model, can be executed. When disturbance due to factors other than purge is small and the effect of the purge on the air-fuel ratio feedback correction coefficient α is relatively large, that is, almost all deviations from the center value of the air-fuel ratio feedback correction coefficient α can be regarded as the effects of the purge. If it is possible, it is determined that it is time to execute the calibration process.
[0062]
Specifically, when all of the “steady condition”, “purge valve accuracy condition”, and “purge influence condition” shown in FIG. 5 are satisfied, it is determined that the calibration process is executable, and one of these conditions is determined. If none of them is established, it is determined that the calibration process cannot be executed. This determination processing corresponds to the processing in the calibration timing determination unit B11 in FIG.
[0063]
As shown in FIG. 5, the “steady-state conditions” include misfire conditions (the engine has not misfired), fuel cut conditions (there is no fuel cut), blow-by conditions (there is no blow-by gas), EGR conditions (exhaust gas recirculation rate is constant), intake throttle valve opening area and engine rotational speed conditions (intake throttle valve opening area and engine rotational speed are constant), purge rate conditions (purge rate is constant) ). Then, when it is determined that all of these conditions are satisfied and the disturbance other than the purge is small, it is determined that the steady condition is satisfied.
[0064]
Further, a purge flow rate condition (the purge flow rate is equal to or more than a predetermined amount) is set as the “purge valve accuracy condition”. When the purge flow rate is small, the control accuracy of the purge flow rate is reduced, and the calculation accuracy in the calibration process described later is reduced. Therefore, when the purge flow rate is smaller than a predetermined amount, it is determined that the purge valve accuracy condition is not satisfied.
[0065]
The “purge influence degree condition” includes a purge establishment condition (purging is performed), a purge concentration condition (the concentration of the purge gas is higher than a predetermined concentration, for example, the α change amount per 1% of the purge rate is 1%). The purge rate condition (the purge rate is equal to or more than a predetermined value, for example, the purge rate is 30% or more) is set. When all of these conditions are satisfied and it is determined that the effect of the purge on the air-fuel ratio is relatively large, it is determined that the purge influence degree condition has been satisfied.
[0066]
If it is determined in step 1 that it is time to execute the calibration process, the process proceeds to step 4 and executes the calibration process. This calibration process will be described with reference to the flow chart of FIG. 6. The calibration process estimates the amount of fuel desorbed from the canister 24 from the change in the air-fuel ratio feedback correction coefficient α, and further calculates the amount of fuel desorbed from the estimated desorption amount to the canister 24 by an inverse operation. The amount of adsorbed fuel is calculated, and the value of the amount of adsorption, which is an internal variable of the canister model, is calibrated to the value of the amount of adsorption determined by the inverse operation. This calibration process is a passive calibration process shown in FIG. This corresponds to the processing in the section B31.
[0067]
The processing of steps 111 to 119 is a part added according to the present invention, and will be described later.
[0068]
In step 31, it is determined whether the purging is being performed. The determination as to whether or not the purging is being performed is based on the assumption that the arithmetic processing in the subsequent steps 32 to 35 is performing the purging. Therefore, if these processings are performed when the purging is not performed, the correct calibration is performed. Is no longer possible. Therefore, when the purge is not being executed, this routine is terminated and the calibration process is not performed.
[0069]
When it is determined that the purging is being performed, the routine proceeds to step 32, where the intake air mass Qg [g] obtained from the intake air flow rate Q and the intake air temperature, the target purge rate MPR, the purge correction coefficient FHOS, and the air-fuel ratio feedback correction coefficient α Read.
[0070]
In step 33, the air-fuel ratio feedback correction coefficient α and the purge correction coefficient FHOS (data during purge execution) are used.
DLT = α × FHOS (5a)
The total air-fuel ratio correction coefficient DLT is calculated by the following equation, and the total air-fuel ratio correction coefficient DLT, the target purge rate MPR, and the intake air mass Qg are calculated in step 34.
Dg = K1 × (1-DLT + K2 × MPR) × Qg (5b)
Here, Dg: desorption amount,
DLT: total air-fuel ratio correction coefficient,
MPR: target purge rate,
K1: coefficient (constant determined by the properties of desorbed fuel)
K2: coefficient (constant determined by the properties of air)
Qg: intake air weight,
The desorption amount Dg [g] is calculated by the following equation.
[0071]
The equation (5b) shows the deviation of the total air-fuel ratio correction coefficient from 1.0 (center value) (the first and second terms on the right side), the target purge rate MPR (the third term on the right side) at that time, and the intake air. This is an equation for calculating the amount Dg of fuel desorbed from the canister 4 from the mass Qg. That is, the air-fuel ratio feedback correction coefficient α and the purge correction coefficient FHOS are both values having a center value of 1.0, and the center value of the total air-fuel ratio correction coefficient DLT which is a value obtained by multiplying α by FHOS. The desorption amount is estimated by regarding all deviations from (1.0) as purging.
[0072]
Then, in step 35, from the desorption amount Dg calculated in step 34,
Yr = KD × Dg ＾ (1 / n (T)) (6)
Here, n (T): desorption index,
KD: desorption coefficient,
T: activated carbon temperature,
The amount of adsorption Yr (mass) of the canister 24 is calculated by the following equation. Expression (6) is an inverse operation of Expression (2), which is one of the expressions constituting the above-mentioned canister model.
[0073]
In step 36, the adsorption amount Y used in calculating the desorption amount Dg based on the canister model is replaced with the adsorption amount Yr calculated in step 35. As a result, the value of the adsorption amount used in the canister model can be calibrated to a correct value, and the calculation accuracy of the subsequent desorption amount Dg can be improved.
[0074]
When the calibration process (second calibration process) is completed in this way, the process returns to FIG. 4, and in step 6, the desorption amount Dg from the canister 4 is calculated using the canister model. The calculation of the desorption amount Dg will be described with reference to the flowchart of FIG. This processing corresponds to the processing in the desorption amount calculation unit B4 in FIG.
[0075]
In step 41, the present value Y of the amount of fuel adsorbed in the canister is calculated by the above equation (1). However, when the calibration processing shown in FIG. 6 is executed, the calculation of the equation (1) is not performed, or the value calculated by the equation (1) is ignored, and in the subsequent calculations, the suction amount calculated by the above calibration processing Yr is used as the adsorption amount Y.
[0076]
In step 42, the desorption amount Dgk at the time of the reference purge flow rate is calculated from the adsorption amount Y by the above equation (2), and in step 43, the desorption amount by the above equation (3) from the reference desorption amount Dgk and the purge flow rate PQ. Calculate Dg. In step 44, the activated carbon temperature T is calculated from the desorption amount Dg by the above equation (4).
[0077]
When the calculation of the desorption amount Dg and the activated carbon temperature T is completed in this way, the process returns to FIG. 4. In step 7, the purge correction coefficient FHOS is calculated based on the desorption amount Dg, the intake air mass Qg, and the target purge rate. . The calculated purge correction coefficient FHOS is sequentially stored in a predetermined data storage location (see FIG. 11) in the engine controller 11.
[0078]
The change in the air-fuel ratio feedback correction coefficient α can be predicted by supplying the desorption amount Dg calculated by the canister model to the engine, and the predicted α is the purge correction coefficient FHOS. Therefore, FHOS is the same unit as α, and the central value of FHOS is also 1.0 (= 100%), the same as α. For example, if the amount Dg of desorption from the canister 24 increases or the target purge rate increases and the amount of fuel supplied to the engine increases, the air-fuel ratio of the exhaust shifts to a richer side than the stoichiometric air-fuel ratio, , The air-fuel ratio feedback correction coefficient α is expected to change to a value smaller than 100%. Accordingly, the purge correction coefficient FHOS correspondingly changes to a value smaller than 100%. Further, even if the desorption amount Dg is equal to the target purge rate, when the intake air mass Qg increases, the air-fuel ratio of the exhaust shifts to the lean side from the stoichiometric air-fuel ratio, and the air-fuel ratio feedback correction coefficient α is set to return to the original value. Since it is expected to change to the larger side, the purge correction coefficient FHOS correspondingly changes to the larger side.
[0079]
In step 8, the purge correction coefficient FHOS is subjected to dead time correction and delay correction composed of smoothing processing. The reason for performing the dead time correction is that there is a delay according to the transfer speed of the purge gas and the distance between the purge valve 27 and the engine cylinder between the time when the purge valve 27 is opened and the time when the purge gas reaches the cylinder of the engine. The reason for performing the annealing process is that the fuel that has desorbed from the canister 24 diffuses before reaching the cylinder of the engine.
[0080]
This delay correction will be described with reference to the flowchart of FIG. 8. Delay correction corresponds to the processing in the delay correction unit B6 in FIG.
[0081]
In step 51, the intake air flow rate is calculated from the output of the air flow meter 13, and in steps 52 and 53, the dead time and the smoothing coefficient are obtained by searching tables having the contents shown in FIGS. As the intake air flow rate increases, the intake flow velocity increases, so the dead time decreases as the intake air flow rate increases.If the intake air flow rate increases and the intake flow velocity increases, the rate at which the desorbed fuel diffuses also increases. Since the speed increases, the smoothing coefficient is set to a value that increases as the intake air flow rate increases.
[0082]
In step 54, a value corresponding to the moving speed of the purge gas is calculated from the dead time. The value corresponding to the purge gas moving speed is calculated as the reciprocal of the dead time obtained in step 22.
[0083]
In step 55, the purge correction coefficient FHOS stored in the data storage location (see FIG. 11) in the engine controller 11 corresponding to the distance between the purge valve 27 and the cylinder of the engine is read. The data is shifted toward the cylinder by the speed equivalent value. In step 57, the average value of the data overflowing from the data storage location due to the data shift is determined.
[0084]
In step 58, the average value of the overflow data obtained in step 57 is subjected to an averaging process using the averaging coefficient obtained in step 52, and the value after the averaging process is set as the purge correction coefficient FHOS again. . The smoothing process is a general smoothing process using a first-order lag system, and the degree of smoothing increases as the smoothing coefficient decreases.
[0085]
FIG. 11 is a diagram showing an outline of the dead time correction in the delay correction. In the drawing, black circles and white circles indicate data before the data shift and data after the data shift, respectively.
[0086]
As shown in the figure, the memory of the engine controller 11 is provided with a data storage location corresponding to the distance between the purge valve 27 and the cylinder of the engine, and the purge amount calculated according to the amount of fuel desorbed from the canister 24 is provided. The correction coefficient FHOS is sequentially stored in the storage location. In the dead time correction, these data are shifted to the cylinder side by an amount equivalent to the purge gas transfer speed (the reciprocal of the dead time), and the overflow caused by the data shift from the data storage location reaches the cylinder and is supplied to the purge gas. Is a correction coefficient corresponding to Then, a value obtained by performing a smoothing process on the average value of the overflowed data is used for correcting a fuel injection pulse width Ti described later. In this way, by combining the dead time correction and the smoothing process, the arrival delay of the purge gas can be accurately corrected.
[0087]
When the delay correction is completed in this way, the process returns to FIG. 4, and the process at the calibration executable timing is ended.
[0088]
On the other hand, if it is determined in step 3 that the calibration process cannot be performed, the process proceeds to step 5 and determines whether the calibration process has been performed in the past. Such a determination is made because there is not yet an initial value (initial adsorption amount) required to operate the canister model when the calibration process has never been performed. This is to prevent the purging process based on the model from being performed. As a result of the determination, if the calibration process has been performed at least once in the past, the operations of steps 6 to 8 are executed. If the calibration process has never been performed, the operation of step 2 is performed. To end.
[0089]
It should be noted that purging is not performed when the calibration process has never been performed, and the purging process is executed by a purging process (boot-up control shown in FIG. 16) that does not use a canister model described later.
[0090]
Using the purge correction coefficient FHOS calculated in this way, in a fuel injection pulse width Ti calculation routine (not shown)

Here, Tp: basic injection pulse width [msec],
Tfbya: target equivalent ratio [unnamed number],
Kathos: transient correction amount [msec],
FHOS: purge correction coefficient (value after delay smoothing),
α: air-fuel ratio feedback correction coefficient,
αm: Air-fuel ratio learning value,
Ts: invalid pulse width [msec],
The fuel injection pulse width Ti [msec] given to the fuel injection valve 4 is calculated by the following equation. This calculation corresponds to the processing in the basic injection pulse width calculation unit B7, the air-fuel ratio feedback correction coefficient calculation unit 8, the fuel injection pulse width calculation unit B9, and the air-fuel ratio learning control unit B16 in FIG.
[0091]
Here, the part (Tp × Tfbya) except for the transient correction amount Kathos is corrected by the purge correction coefficient FHOS, because the fuel supplied by the purge is a gaseous fuel, so that wall flow correction must be performed. Because there is no.
[0092]
The basic injection pulse width Tp is an injection pulse width at which a stoichiometric air-fuel ratio can be obtained, and is obtained from the intake air flow rate Q and the engine speed N.
[0093]
The target equivalent ratio Tfbya is a value centered at 1.0, and becomes a value larger than 1.0 when the cooling water temperature is low or immediately after the engine is started. After completion of the warm-up of the engine, Tfbya = 1.0.
[0094]
The air-fuel ratio feedback correction coefficient α is a value calculated based on the output of the air-fuel ratio sensor 15 so that the exhaust air-fuel ratio falls within a so-called window centered on the stoichiometric air-fuel ratio. Specifically, assuming that there is no steady-state deviation (base air-fuel ratio error) at a value centered on 100%, when the detected air-fuel ratio is richer than the stoichiometric air-fuel ratio, the value becomes smaller than 100%. When the detected air-fuel ratio is leaner than the stoichiometric air-fuel ratio, the value becomes greater than 100%.
[0095]
The air-fuel ratio learning value αm is a value calculated based on the air-fuel ratio feedback correction coefficient α, and is introduced mainly to compensate for a steady-state deviation.
[0096]
Part of the fuel injected from the fuel injection valve 4 forms wall-flow fuel on the intake port wall and the back of the intake valve umbrella. Inflow. The transient correction amount Kathos is a value for compensating for the response delay of the wall flow fuel during this transition.
[0097]
The invalid pulse width Ts is a value for compensating an operation delay from when the fuel injection valve 4 receives an injection signal at a predetermined timing until the fuel injection valve 4 actually opens.
[0098]
Since the equation (7) is an equation for sequential injection (injection is performed once for every two revolutions of the engine in the case of four cylinders in accordance with the ignition order of each cylinder), the numeral 2 is included.
[0099]
FIGS. 12 and 13 are for setting the target purge rate MPR, and are executed at regular intervals (for example, every 10 msec). This calculation corresponds to the processing in the target purge rate setting unit B1 in FIG.
[0100]
In step 61, it is determined whether or not after the first calibration process. This is because the method of setting the target purge rate differs before and after the first calibration process. If it is before the first calibration process, the current process ends. The setting of the target purge rate before the first calibration process will be described later in the boot-up control.
[0101]
If it is after the first calibration process, the routine proceeds to step 62, in which the air-fuel ratio of the purge gas (purge air-fuel ratio) is calculated based on the desorption amount Dg calculated based on the canister model (calculated in step 6 in FIG. 4) and the purge flow rate PQ. ) Is calculated. Here, the purge flow rate PQ is a product of the target purge rate and the intake air flow rate Q, and the previous value is used as the target purge rate in this case.
[0102]
The purge air-fuel ratio may be corrected as follows. That is, the error of the purge air-fuel ratio is estimated from the operating state, for example, parameters such as the engine speed, the engine load, and the intake air flow rate. For example, it is obtained by searching a table having the contents shown in FIG. As shown in the drawing, the purge air-fuel ratio error increases as the intake air flow rate decreases and the purge rate decreases. Alternatively, the purge air-fuel ratio error may be obtained by searching a table that defines the relationship between the purge air-fuel ratio and the purge air-fuel ratio error as shown in FIG. When the purge air-fuel ratio error is determined, the purge air-fuel ratio determined in step 62 is corrected based on the error.
[0103]
Although the purge air-fuel ratio may be detected by an HC sensor, the purge air-fuel ratio can be calculated at low cost and accurately by calculating the purge air-fuel ratio based on a desorption amount calculated based on a canister model.
[0104]
In step 63, the purge rate change amount limit value is calculated based on the purge air-fuel ratio thus obtained (or after error correction). When the purge rate changes, the exhaust air-fuel ratio changes. The purge rate change amount limit value is calculated so that the air-fuel ratio deviation from the stoichiometric air-fuel ratio at this time falls within an allowable range. The allowable range of the air-fuel ratio deviation from the stoichiometric air-fuel ratio is set to a range that can be absorbed by the air-fuel ratio feedback control and does not deteriorate the exhaust emission.
[0105]
In step 64, a purge rate upper limit value PVMX defined from the size of the purge valve 27 is calculated. The reason why the purge rate upper limit value PVMX is obtained is that if the target purge rate is set to a value larger than the purge rate obtained by setting the purge valve 27 to the maximum opening degree, the purge rate and the target purge rate will not match. , And FHOS, the air-fuel ratio deviation from the stoichiometric air-fuel ratio increases. This causes a problem such as deterioration of exhaust emission. Specifically, when the size of the purge valve 27 is constant, the larger the differential pressure across the purge valve 27, the larger the flow rate of the purge gas that can flow. Therefore, when the differential pressure across the purge valve 27 is large, the purge rate upper limit value PVMX is large. Calculate the value.
[0106]
In step 65, the purge rate upper limit TIMNMX based on the performance of the fuel injection valve 4 is calculated from the relationship between the minimum injection pulse width determined according to the performance of the fuel injection valve 4, the previous value of the target purge rate, and the purge correction coefficient. . When the purge rate increases, the amount of fuel supplied to the engine by the purge increases. Therefore, the fuel injection pulse width is corrected so that the fuel injection amount from the fuel injection valve 4 is reduced by that amount. In order to ensure injection accuracy, the injection pulse width must be larger than a predetermined minimum pulse width. In other words, in order to make the fuel injection pulse width larger than the minimum pulse width, the purge rate must be made smaller than a certain value. For this reason, the upper limit of the purge rate is also defined by the injection performance of the fuel injection valve 4.
[0107]
In step 66, all possible operating ranges are assumed from the current operating range, the minimum purge rate in the current operation range is predicted, and the purge rate upper limit value is calculated from the minimum purge rate and the purge rate change amount limit value. Calculate PRMNMX. For example, the target purge rate is set to a very small value when the accelerator pedal is depressed as much as possible to accelerate, but if the target purge rate is set to a large value immediately before depressing the accelerator pedal as much as possible, the purge rate becomes Is limited to the change amount limit value or less, the purge rate cannot follow the target purge rate. Since the following delay causes an increase in exhaust emissions, it is necessary to define the upper limit of the purge rate from the minimum purge rate that can be assumed so as not to cause such a delay.
[0108]
In step 67, the air-fuel ratio feedback correction coefficient α is monitored, and if the air-fuel ratio feedback correction coefficient α is equal to or less than a predetermined value, the largest value among the purge rates that sets the air-fuel ratio feedback correction coefficient α to a predetermined value or more is calculated as the purge rate upper limit value ALPMX. . The reason for providing such an upper limit value ALPMX is that in the air-fuel ratio feedback control, the air-fuel ratio feedback correction coefficient α is controlled to be within 100 ± 25%, but the air-fuel ratio feedback correction coefficient α is close to the limit value (for example, In the case where the control is performed at 80%), if a large amount of purging is performed, disturbances other than the purging are received, and it is likely to be out of the control range.
[0109]
In step 68, the smallest value is selected from the four upper limit values PVMX, TIMNMX, PRMNMX, and ALPMX, and that value is set as the maximum purge rate PRMAX.
[0110]
In steps 69 and 70, the previous value of the target purge rate is compared with the maximum purge rate PRMAX. When the previous value of the target purge rate is equal to the maximum purge rate PRMAX, the process proceeds from step 69 to step 73 where the target purge rate is set to the previous value. When the previous value of the target purge rate is larger than the maximum purge rate PRMAX, the process proceeds from step 69 or 70 to step 72, where the target purge rate is set to a value obtained by subtracting the purge rate change amount limit value from the previous value. When the previous value of the purge rate is smaller than the maximum purge rate PRMAX, the process proceeds to steps 71 from steps 69 and 70, and the target purge rate is set to a value obtained by adding the purge rate change amount limit value to the previous value.
[0111]
The processing of steps 74 to 81 in FIG. 13 is a part added according to the present invention, and will be described later.
[0112]
In this manner, the target purge rate is set so as to follow the maximum purge rate PRMAX within the range of the purge rate change amount limit value, and to perform a large amount of purge without deteriorating the exhaust emission. The best purge rate is set. Further, when setting the maximum purge rate PRMAX, not only the upper limit values PVMX, TIMNMX, and ALPMX determined by the physical limitation, the current operation area, and the like, but also the maximum purge rate in that area without delay even if the operation area changes. Since the upper limit value PRMNMX, which is determined so as to be able to shift to the rate, is also taken into consideration, it is possible to set the best purge rate for performing a large quantity purge without deteriorating the exhaust emission even when the operating conditions change. .
[0113]
The target purge rate set in this way is converted into a duty ratio for the purge valve and output to the purge valve 27.
[0114]
By the way, the purging process based on the canister model (model reference purging process) may be executed as long as no calibration process has been performed after the engine is started and there is no initial value (initial adsorption amount) used in the canister model. Can not. However, in order to realize a large amount of purge after the engine is started, it is necessary to execute the purge process even before the calibration process. Therefore, until the first calibration process is performed, the target purge rate is set by the following control (boot-up control shown in FIG. 16) instead of the purge process based on the canister model, and the purge is performed at the set target purge rate. Execute the process. In this boot-up control, the effect of the purge on the exhaust air-fuel ratio is absorbed by the air-fuel ratio feedback correction coefficient α, and the fuel injection pulse width is not corrected (FHOS = 1.0).
[0115]
FIG. 16 is for performing boot-up control, and is executed at regular intervals (for example, every 10 msec). This boot-up control corresponds to the processing in the boot-up control unit B14 and the changeover switch B15 in FIG.
[0116]
At step 91, the intake air flow rate Q is read, and at step 92, it is determined whether or not it is before the first calibration process. If it is before the first calibration processing, in step 93, the integrated purge flow rate (total purge flow rate after starting the purge) SQP is compared with the purge pipe volume (the pipe volume from the canister 24 to the purge valve 27). When the accumulated purge flow rate SQP exceeds the purge pipe volume, the process proceeds to step 94, and when it does not exceed, the process proceeds to step 97.
[0117]
In step 97, an initial purge rate (a small value of 1% or less) is set as the target purge rate MPR. The reason for setting such a small value is that if the accumulated purge flow rate SQP has not reached the purge pipe volume, the gas in the purge pipe will be supplied to the engine before the purge is started. This is because the air-fuel ratio of the gas is unknown, and if the target purge rate setting processing shown in steps 94 to 96 is performed as it is, a problem such as deterioration of combustion stability of the engine occurs.
[0118]
That is, at the start of the purge, a low-concentration purge gas existing in the purge pipe is supplied. If the air-fuel ratio deviation from the stoichiometric air-fuel ratio is small, it is determined that a larger amount of purge is possible, and the large target purge rate MPR is determined. However, if a large target purge rate MPR is set in this way, a large amount of desorption occurs when all low-concentration gases in the pipe are supplied and the original high-concentration purge gas is supplied. This is because fuel is suddenly supplied, which causes deterioration of combustion stability and the like of the engine.
[0119]
When the accumulated purge flow rate SQP exceeds the pipe volume, the routine proceeds to step 94, where the difference between the actual air-fuel ratio feedback deviation and the target air-fuel ratio feedback deviation is calculated.
[0120]
Here, the target air-fuel ratio feedback deviation refers to the deviation (= | tα−100 |%) between the target value tα of the air-fuel ratio feedback correction coefficient and the center value (100%) of the air-fuel ratio feedback correction coefficient. The fuel ratio feedback deviation is a deviation (= | α−100 |%) between the actual air-fuel ratio feedback correction coefficient α and the center value of the air-fuel ratio feedback correction coefficient. For example, the target value tα of the air-fuel ratio feedback correction coefficient α is set to 80% in order to secure a large purge flow rate within a range in which the air-fuel ratio deviation from the stoichiometric air-fuel ratio due to the purge can be sufficiently absorbed by the air-fuel ratio feedback control. Then, the target air-fuel ratio feedback deviation is set to | 80-100 | = 20%.
[0121]
At step 95, a purge rate change amount corresponding to the difference between the target air-fuel ratio feedback deviation and the actual air-fuel ratio feedback deviation is calculated by searching a table having the contents shown in FIG. The amount of change in the purge rate is set to a larger value as the absolute value of the difference between the target air-fuel ratio feedback deviation and the actual air-fuel ratio feedback deviation increases, thereby improving the convergence to the target value. Depending on the sign of the difference in the actual air-fuel ratio feedback deviation, a different value is set even if the absolute value of the deviation is the same, and a larger value (absolute value) is obtained when the difference in the air-fuel ratio feedback deviation is shifted to the negative side. Set to the purge rate change amount.
[0122]
The reason why the purge rate change amount has different characteristics depending on whether the air-fuel ratio feedback deviation is positive or negative is that the air-fuel ratio feedback correction coefficient α is set to the target 80% when the air-fuel ratio feedback deviation is shifted to the negative side. This is because the engine stability and exhaust emission are more likely to be degraded due to disturbances other than purge, as compared with the case where the value is shifted to the opposite positive side, and it can be said that there is a disadvantageous condition. That is, the reason why the characteristics of the purge rate change amount are changed according to the sign of the difference in the air-fuel ratio feedback deviation is to quickly return the control point to the safe side from the viewpoint of engine combustion stability and prevention of exhaust emission deterioration. is there.
[0123]
After calculating the purge rate change amount as described above, the routine proceeds to step 96, where the purge rate change amount calculated in step 95 is added to the target purge rate obtained in the previous execution of this routine, and a new target purge rate MPR Is calculated.
[0124]
In step 98, the purge flow rate PQ is obtained from the target purge rate MPR and the intake air flow rate Q thus obtained, and in step 99, the cumulative purge flow rate SQP is updated by adding the purge flow rate PQ to the previous value of the cumulative purge flow rate SQPz. I do.
[0125]
Therefore, according to the processing of steps 94 to 96, the optimum target purge rate MPR can be set regardless of the suction state of the canister 24, and even if a purge gas with a concentration higher than expected is supplied, The target purge rate MPR is appropriately changed in response to the change in the exhaust air-fuel ratio, and the optimal target purge rate MPR can always be set.
[0126]
On the other hand, after the first calibration process, the process proceeds from step 92 to step 100 and thereafter, and it is determined whether or not the purge process has been completed based on the accumulated purge flow rate SQP. That is, in step 100, a purge end determination flag (initial setting to zero) is checked. Immediately after the start of the engine, the purge end determination flag is 0, so the routine proceeds to step 101, where the desorption amount Dg (calculated in step 6 in FIG. 4) and the target purge rate MPR (calculated in FIGS. 12 and 13). Read). In step 102, the purge flow rate QP is calculated from the read target purge rate MPR and the intake air flow rate Q, and is added to the previous value of the cumulative purge flow rate SQPz in step 103 to update the cumulative purge flow rate SQP. I do.
[0127]
In step 104, it is determined whether or not it is the first calibration processing timing. Here, since the first calibration processing timing is the first timing that has disappeared before the first calibration processing in step 92, the process proceeds to step 105 only at this time, and the purge flow rate PQ calculated in step 102 and read in step 101. The purge air-fuel ratio is calculated based on the desorption amount Dg, and a purge end determination value SLSQP is calculated from the purge air-fuel ratio by searching a table having the contents shown in FIG.
[0128]
Here, the purge end determination value SLSQP is an integrated purge flow rate at which there is almost no desorbed fuel in the purge gas from the canister 24. As shown in FIG. 18, the determination value SLSQP decreases as the purge air-fuel ratio increases.
[0129]
In step 107, the purge end determination value SLSQP is compared with the accumulated purge flow rate SQP. If the accumulated purge flow rate SQP is less than the purge end determination value SLSQP, the current process is terminated. If the accumulated purge flow rate SQP eventually becomes equal to or greater than the purge end determination value SLSQP, it is determined that there is almost no desorbed fuel in the purge gas from the canister 24, and the flow proceeds to step 108 to indicate that the purge has been completed. Let it be 1.
[0130]
When the purge end determination flag is set to 1, the flow ends with steps 91, 92 and 100 from the next time.
[0131]
Next, the overall operation of the present embodiment by performing the purge process will be described first.
[0132]
At the time of the purge process, the target purge rate MPR is set to a value as large as possible within a range that does not cause a decrease in engine combustion stability and an increase in exhaust emission, and the purge valve 27 is opened so as to realize the target purge rate MPR (FIG. 12). Steps 61 to 73).
[0133]
During the purge process, the purge gas containing the fuel desorbed from the canister 24 is supplied to the engine. Therefore, the engine controller 11 estimates the amount Dg of the fuel desorbed from the canister 24, and estimates the amount Dg of the desorbed fuel. Based on the target purge rate and the intake air mass Qg, a purge correction coefficient FHOS, which is a value for predicting a change in the air-fuel ratio feedback correction coefficient α caused by the purge fuel, is calculated (steps 1, 3, 4, 6, and 6 in FIG. 4). 7, 8 or steps 1, 3, 5, 6, 7, 8), by correcting the basic injection pulse width Tp by the purge correction coefficient FHOS, the air-fuel ratio deviation from the stoichiometric air-fuel ratio due to the purge can be reduced. Can be suppressed.
[0134]
At this time, the desorption amount Dg from the canister 24 is estimated using the canister model represented by the equations (1) to (4), and the desorption amount Dg is accurately estimated in a short time. Here, the desorption amount Dg or the like calculated based on the canister model includes an error. Since the error increases and increases as the operation time of the canister model increases, the engine controller 11 determines the air-fuel ratio feedback correction coefficient α. The amount of fuel desorbed from the canister 24 is estimated based on the change of the canister 24, and the value of the amount of adsorption, which is an internal variable of the canister model, is calibrated based on the adsorption amount Yr, which is reversed from the estimated amount of desorption. This calibration process is executed when disturbances other than the purge are small and almost all changes in the air-fuel ratio feedback coefficient α can be regarded as being caused by the purge, and the effect of the purge on the exhaust air-fuel ratio is relatively large.
[0135]
Further, there is a delay from when the purge valve 27 is opened to when the desorbed fuel reaches the cylinder of the engine, and there is also diffusion of fuel before the desorption fuel reaches the cylinder. Delay correction is performed on the minute correction coefficient FHOS.
[0136]
Further, the purging process using the canister model (model reference purging process) cannot exert its effect until the initial value of the adsorption amount is obtained by the calibration process, but the purge process is not performed until the initial value of the canister model is calculated. The target purge rate is set in accordance with the difference between the target air-fuel ratio feedback deviation and the actual air-fuel ratio feedback deviation, and the purge valve 27 is opened so that the target purge rate is realized. Thus, the purge process can be performed even before the initial value is calculated by the calibration process, and an effective purge can be performed in all regions.
[0137]
This concludes the description of the same part as the purging process in the prior application (see Japanese Patent Application No. 2001-71562). Next, steps 111 to 119 in FIG. 6 and steps 74 to 81 in FIG. , FIG. 19, FIG. 20, FIG. 21, FIG. 23, FIG. 24, FIG. 26, and FIG.
[0138]
First, a description will be given with reference to FIG. The processing in this part corresponds to the processing in the AND circuit B38, the purge rate lowering processing unit B39, and the changeover switch B40 in FIG.
[0139]
In step 74, the target purge rate is moved to a variable MPR0.
[0140]
In step 75, a calibration executable flag (initialized to zero) is checked. In FIG. 5 described above, when all of the “steady condition”, “purge valve accuracy condition”, and “purge influence degree condition” are satisfied, it is determined that the calibration process is executable, but the calibration executable flag is set to 1 at this timing. Is a flag. If the calibration feasibility flag is 0, the routine proceeds to step 78, where the value of MPR0 is set as it is as the target purge rate MPR.
[0141]
On the other hand, when the calibration executable determination flag is 1, the process proceeds to steps 76 and 77 to check the data sampling end flag and the active calibration request flag (both are initially set to zero). As will be described later, when it is determined that there is a base air-fuel ratio error, the active calibration request flag is set to 1, and at the time when the active calibration request flag is switched from 0 to 1, the data before the purge cut is sampled and purged. Start rate reduction processing. This purge rate lowering process is a process of lowering the target purge rate by a predetermined value per predetermined time until the target purge rate becomes zero, with the target purge rate immediately before the active calibration request flag is switched from 0 to 1 as an initial value. When the stabilization time has elapsed in a state where the target purge rate has become zero due to the rate reduction processing (purge cut state), the data at the time of the purge cut is sampled, and the sampling of data before and after the purge cut is completed. It is assumed that the data sampling end flag = 1.
[0142]
Therefore, before the active calibration request flag is set to 1, the data sampling end flag is set to 0. In this case, the process proceeds from step 76.77 to step 78, and when the data sampling end flag is set to 1, the process proceeds from step 76. Proceed to 78 to execute the operation of step 78.
[0143]
On the other hand, during the period from when the active calibration request flag is set to 1 until immediately before the data sampling end flag is set to 1, the routine proceeds to step 79, where the purge rate is reduced.
[0144]
Here, the setting of the active calibration request flag in step 77 will be described with reference to the flow in FIG.
[0145]
FIG. 19 is for setting the active calibration request flag, and is executed at regular intervals (for example, every 10 msec).
[0146]
In step 121, it is determined whether or not the model purge is being executed. The execution of the model purge is a state in which the initial calibration process of the canister model has been completed and purging is not prohibited. If the model purge is being executed, the process proceeds to step 122, and the purge limiter determination flag is checked. There are two purge limiter determination flags, a purge lower limiter determination flag and a purge upper limiter determination flag. The setting of these two flags will be described with reference to the flowcharts of FIGS.
[0147]
FIG. 20 is for setting the purge lower limiter determination flag, and FIG. 21 is for setting the purge upper limiter determination flag, all of which are executed at regular intervals (for example, every 10 msec). The setting process of FIGS. 20 and 21 and the setting process of FIG. 19 correspond to the processes of the base air-fuel ratio error determination unit B36 and the active calibration request unit B37 in FIG.
[0148]
As shown in FIG. 28, the air-fuel ratio feedback correction coefficient α has a first limiter ± LMT1 (eg, ± 25%) which is vertically increased and decreased about a central value (100%) of α in order to limit the control range based on α. ). That is, the first upper limiter of α is 125% (= 100 + LMT1), the first lower limiter of α is 75% (= 100−LMT1), and the calculated α is the first upper limiter of α 125%. When α exceeds 75%, the α is limited to 125%. When the calculated α falls below 75% which is the first lower limiter of α, the α is limited to 75%.
[0149]
In this embodiment, apart from the first limiter ± LMT1, a second limiter ± LMT2 smaller than the first limiter ± LMT2 is newly introduced for purging. That is, the second lower limiter is 100-LMT2 and the second upper limiter is 100 + LMT2. In FIG. 28, 100-LMT1 is between 100%, which is the center value of α, and 100-LMT1, which is the first lower limiter. Therefore, the second lower limiter 100-LMT2 is located at a position where a certain margin is provided. In other words, the interval between 100-LMT2 (second lower limiter) and 100-LMT1 (first lower limiter) is left as a correction allowance for α, and if there is a disturbance other than purging, the theoretical correction allowance for α remains. The air-fuel ratio deviation from the air-fuel ratio is made small.
[0150]
Then, a flag is set to indicate whether α is in a region equal to or smaller than the second lower limiter or in a region equal to or larger than the second upper limiter (not shown in FIG. 28). That is, when the calculated α becomes equal to or less than 100−LMT2 which is the second lower limiter, the purge lower limiter setting flag = 1, and when the calculated α becomes equal to or more than 100 + LMT2 which is the second upper limiter, the purge upper limit is set. It is assumed that the limiter setting flag = 1.
[0151]
However, the second lower limiter and the second upper limiter are each provided with a hysteresis HYS. This is to prevent frequent repetition of each flag becoming zero or one due to α being near each limiter.
[0152]
Referring to FIG. 20 specifically, in step 131, it is determined whether the air-fuel ratio feedback control is being performed. If the air-fuel ratio feedback control is not being performed, the current process ends.
[0153]
If the air-fuel ratio feedback control is being performed, the routine proceeds to step 132, where the air-fuel ratio feedback correction coefficient α is read.
[0154]
This α is calculated based on the air-fuel ratio sensor 15 in a flow not shown. Here, for the sake of simplicity, a description will be given of the case of the proportional operation. For example, a deviation between the actual exhaust air-fuel ratio detected by the air-fuel ratio sensor 15 and the stoichiometric air-fuel ratio may be obtained, and the deviation may be multiplied by a proportional gain. . At this time, while the actual exhaust air-fuel ratio is richer than the stoichiometric air-fuel ratio, the waveform of α gradually decreases in a stepwise manner in each control cycle in order to return the actual exhaust air-fuel ratio to the stoichiometric air-fuel ratio. On the other hand, while the actual exhaust air-fuel ratio is leaner than the stoichiometric air-fuel ratio, α gradually increases stepwise at each control cycle to return the actual air-fuel ratio to the stoichiometric air-fuel ratio.
[0155]
In step 133, a purge lower limiter determination flag (initial setting to zero when the engine is started) is checked. If it is immediately after the start, since the purge lower limiter determination flag is 0, the routine proceeds to step 134, where α is compared with 100-LM2, which is the purge lower limiter. If α exceeds 100-LMT2, the current processing is terminated as it is. If α becomes 100-LMT2 or less, the routine proceeds to step 135, where the purge lower limiter determination flag is set to 1 and the current processing is terminated.
[0156]
Next, the process proceeds from step 133 to step 136 based on the purge lower limiter determination flag = 1, and compares α with 100−LMT2 + HYS (HYS is a positive constant value). If α is equal to or less than 100−LMT2 + HYS, the routine proceeds to step 135, where the purging lower limiter judgment flag = 1 is maintained. If α exceeds 100−LMT2 + HYS, the routine proceeds to step 137, where the purging lower limiter judging flag = 0.
[0157]
Turning to FIG. 21, steps 141 and 142 are the same as steps 131 and 132 in FIG.
[0158]
In step 143, an upper limit limiter determination flag for purging α (initial setting to zero when the engine is started) is checked. Immediately after the start, since the purge upper limiter determination flag is 0, the routine proceeds to step 144, where α is compared with 100 + LMT2, which is the purge upper limiter. When α is less than 100 + LMT2, the current process is terminated as it is. When α is 100 + LMT2 or more, the process proceeds to step 145 to set the purge upper limiter determination flag = 1 and terminate the current process.
[0159]
Next, the process proceeds from step 143 to step 146 according to the purge upper limiter determination flag = 1, and compares α with 100 + LMT2-HYS (HYS is a positive constant value). If α is equal to or greater than 100 + LMT2-HYS, the routine proceeds to step 145, where the upper-limit purge judgment flag for purging is kept at 1, and if α is less than 100 + LMT2-HYS, the routine proceeds to step 147, where the upper-limit purging judgment flag = 0.
[0160]
Returning to step 122 in FIG. 19, when the purge limiter determination flag (the lower limit limiter determination flag for purge or the upper limiter determination flag for purge) = 1, the process proceeds to step 123 and the timer 1 (initial setting to zero) is incremented. The timer 1 is for measuring the time during which the purge limiter determination flag is 1 continuously. Therefore, if the purge limiter determination flag becomes 0 even during the increment, the routine proceeds to step 126, where the timer 1 is cleared to zero. If the purge limiter determination flag becomes 1 again, the routine proceeds to step 123 and the timer 1 is incremented. I do.
[0161]
In step 124, the timer 1 value is compared with a predetermined value. The predetermined value is a value for determining whether there is a base air-fuel ratio error. When the timer 1 value becomes equal to or more than a predetermined value, it is determined that there is a base air-fuel ratio error, and the process proceeds to step 125 to perform the active calibration process (second calibration process). Therefore, the active calibration request flag (initial setting to zero) = 1 I do.
[0162]
However, when the operating condition changes transiently, the purge limiter determination flag may be temporarily set to 1 even if the base air-fuel ratio error does not exist. Therefore, the base air-fuel ratio error also exists in this case. The predetermined value is set so as not to be erroneously determined to be. That is, by setting a predetermined time longer (for example, several seconds) longer than the time during which the purge limiter determination flag = 1 becomes erroneously determined as a base air-fuel ratio error due to a transient change in operating conditions. Avoid doing that.
[0163]
Now, the description of FIG. 19 is completed. Returning to FIG. 13, immediately after the active calibration request flag = 1, since the data sampling end flag = 0, the process proceeds from step 76, 77 to step 79. A purge rate lowering process is performed.
[0164]
As shown in FIG. 22, the purge rate lowering process is performed at the target purge rate at the timing when the active calibration request flag = 1 and the calibration executable determination flag is switched from 0 to 1 (at steps 61 to 73 in FIG. 12). (Calculated) is set as an initial value, and is reduced by a predetermined value Δ per predetermined time Δt, and finally reduced to zero. This is one method for performing a purge cut, and is not a stepwise reduction from an initial value to zero.
[0165]
FIG. 23 and FIG. 24 (subroutine of step 79 in FIG. 13) are for performing this purge rate lowering process. This processing corresponds to the processing in the purge rate lowering processing section B39, the data sampling section B42 before purge cut, and the data sampling section B43 during purge cut in FIG.
[0166]
In FIG. 23, in steps 151 and 152, the current and previous calibration executable determination flags are checked. When the calibration execution feasibility determination flag is set to 1 this time and the calibration execution feasibility determination flag is set to 0 last time (that is, immediately after the calibration execution feasibility determination flag is switched from 0 to 1), the process proceeds to step 153, and the purge rate lowering process starts , The purge correction coefficient FHOS (calculated in FIG. 4), and the target purge rate MPR0 (obtained in step 74 in FIG. 13). , The intake air mass Qg (obtained from the intake air flow rate Q and the intake air temperature), the air-fuel ratio learning value αm (from the map of the air-fuel ratio learning value not shown), (Learning values are read) and stored in the corresponding memories α1, FHOS1, MPR1, Qg1, and αm1 (sampling of data before purge cut). ).
[0167]
In step 154, the value of MPR0 at the start of the purge rate lowering process is similarly moved to a variable MPRG. This is because the target purge rate at the timing when the calibration feasibility flag is switched from 0 to 1 is sampled as an initial value.
[0168]
After the timer 2 is once cleared to zero in step 155, the increment is started in step 156.
[0169]
This time, when the calibration execution possibility determination flag is set to 1 last time, steps 153, 154, and 155 are skipped, and the operation of step 156 is executed.
[0170]
In step 157, the value of the timer 2 is compared with a predetermined time Δt. When the timer 2 value becomes equal to or longer than the predetermined time Δt, the process proceeds to steps 158 and 159 to clear the timer 2 value to zero,
MPRG = MPRGz-Δ (8)
Where MPRGz: the previous value of MPRG,
The variable MPRG is reduced by a predetermined value Δ (positive value) according to the following equation. When the value of the timer 2 is less than the predetermined time Δt, the process proceeds to step 160 to maintain the current MPRG.
[0171]
By the operations of steps 157 to 160, the MPRG decreases by a predetermined value Δ each time the predetermined time Δt elapses (see the lowermost row in FIG. 22).
[0172]
Here, if the purge cut is performed with the purge rate sharply lowered, the exhaust air-fuel ratio changes and affects drivability and exhaust emission. Therefore, the values of the predetermined time Δt and the predetermined value Δ are set so that the target purge rate does not become so. May be determined using or based on the purge rate change amount limit value at the time of setting.
[0173]
Proceeding to FIG. 24, in step 161, MPRG is compared with zero. If MPRG is equal to or greater than zero, the routine proceeds to step 162, where MPRG is set as the target purge rate MPR, and if MPRG becomes a negative value, the routine proceeds to step 163, where zero is set as the target purge rate MPR.
[0174]
In step 164, the target purge rate MPR is compared with zero. If MPR = 0, the routine proceeds to step 165, where the timer 3 (initial setting to zero) is incremented. The timer 3 is for measuring the continuous elapsed time in the purge cut state (the target purge rate MPR is zero).
[0175]
In step 166, the value of the timer 3 is compared with the stabilization time. The stabilization time is a time to wait for the purge cut state to stabilize.
[0176]
When the value of the timer 3 becomes equal to or longer than the stabilization time, the routine proceeds to step 167, where the air-fuel ratio feedback correction coefficient α and the air-fuel ratio learning value αm are stored in the corresponding memories α2 and αm2. This is to sample data at the time of purge cut. Since the data sampling before and after the purge cut is completed, the data sampling end flag (initial setting to zero) is set to 1 in step 168, and the timers 2 and 3 are cleared to zero in step 169.
[0177]
When the purge rate lowering process is completed in this way, the process returns to step 80 in FIG. In step 80, a purge end determination flag (set in FIG. 16) is checked. When the purge end determination flag is 1, the routine proceeds to step 81 to start the air-fuel ratio learning, and the purge rate for the air-fuel ratio learning is set to the target purge rate MPR. As a result, when the air-fuel ratio learning is started, the target purge rate MPR decreases as shown in FIG.
[0178]
FIG. 26 is for setting the air-fuel ratio learning start flag, and is executed at regular intervals (for example, every 10 msec).
[0179]
In step 171, an air-fuel ratio learning start flag (initial setting to zero) is checked. If it is immediately after the engine is started, the air-fuel ratio learning start flag is 0, so the routine proceeds to step 172, and the purge end determination flag (set by FIG. 16) is checked.
[0180]
In order to start the air-fuel ratio learning only when the purge end determination flag = 1, the routine proceeds to step 173, where the air-fuel ratio learning start flag = 1 is set.
[0181]
FIG. 27 is for updating the air-fuel ratio learning value, and is executed at predetermined timings (for example, at predetermined engine rotation speed intervals). The processes of updating the air-fuel ratio learning value and setting the air-fuel ratio learning start flag in FIG. 26 correspond to the processes in the air-fuel ratio learning control unit B16 in FIG.
[0182]
In step 181, the air-fuel ratio learning start flag is checked. When the air-fuel ratio learning start flag is 1, the routine proceeds to step 182, where it is determined whether the air-fuel ratio learning permission condition is satisfied. The air-fuel ratio learning permission condition is the same as the conventional one. For example, when it is determined that the air-fuel ratio feedback control is being performed and that the air-fuel ratio sensor 15 and the like related to the air-fuel ratio feedback control do not have any failures, it is determined that the air-fuel ratio learning permission condition is satisfied. To update the air-fuel ratio learning value. As a method for updating the air-fuel ratio learning value, there is a method for updating by the following equation.
[0183]
αm = αmz + (α−100) × M (9)
Where αm is the updated air-fuel ratio learning value,
αmz: Air-fuel ratio learning value before update,
M: learning rate (constant value),
In this case, the entire operation range using the engine speed and the basic injection pulse width Tp (corresponding to the engine load) as parameters is divided into a plurality of small regions, and each of the small regions has an independent air-fuel ratio learning value. I have. For this reason, the air-fuel ratio learning value before updating on the right side of equation (9) is the air-fuel ratio learning value stored in the small area to which the operating condition at the update timing belongs. The updated air-fuel ratio learning value is stored.
[0184]
In the embodiment, the air-fuel ratio feedback control based on the air-fuel ratio sensor 15 is used. ₂ Air-fuel ratio feedback control based on a sensor may be used. ₂ The reversal timing of the sensor may be set as the update timing of the air-fuel ratio learning value.
[0185]
Next, returning to FIG. 6, the processing of steps 111 to 119 will be described. The processing in this portion corresponds to the processing in the active calibration processing section B41 and the changeover switch B47 in FIG.
[0186]
In steps 111 and 112, the active calibration request flag and the data sampling end flag are checked. When the active calibration request flag is 0, and when the active calibration request flag is 1 and the data sampling end flag is 0, the process proceeds to steps 32 to 36, and the same passive calibration process as the prior application device (second calibration process) is performed. Execute
[0187]
When the active calibration request flag = 1 and the data sampling end flag = 1, the active calibration processing (first calibration processing) in steps 113 to 117 is performed. That is, in step 113, α1, α2, FHOS1, Qg1, αm1, and αm2 stored in the memory are read. Here, the value with “1” is the data before the purge cut, while the value with “2” is the data at the time of purge cut.
[0188]
In step 114, using the data before and after the purge cut,
DLT = (α1 + αm1 + α2 + αm2-3) × FHOS1 (10)
A total air-fuel ratio correction coefficient DLT is calculated by the following equation, and using this total air-fuel ratio correction coefficient DLT and MPR1, Qg1,
Dg = K1 × (1-DLT + K2 × MPR1) × Qg1 (11)
The desorption amount Dg is calculated by the following equation.
[0189]
Here, how the equation (10) is derived will be described.
[0190]
(A) When the air-fuel ratio learning control is not performed:
Although the embodiment includes the air-fuel ratio learning control unit B16, a unit that does not perform the air-fuel ratio learning control is a basic idea.
DLT = {α1- (1-α2)} × FHOS1 (a)
May be used to calculate the total air-fuel ratio correction coefficient DLT. Here, (1−α2) on the right side of the equation (a) represents the deviation of α from the central value of α during the purge cut, that is, the base air-fuel ratio error, and this is the value during the execution of the purge (equation (a)). Since a calibration error is caused by being included in α1 on the right side, an error corresponding to the base air-fuel ratio error is removed from α1 by subtracting (1−α2) representing the base air-fuel ratio error from α1.
[0191]
(B) When performing air-fuel ratio learning control:
At this time, it is necessary to consider the air-fuel ratio learning value in the same way as the air-fuel ratio feedback correction coefficient α.

The total air-fuel ratio correction coefficient DLT is calculated by the following equation. Here, (1−αm2) on the right side of the equation (b) represents a deviation of αm from the central value of αm at the time of purge cutting, that is, also represents a base air-fuel ratio error, which is a value during execution of purge (b). Since a calibration error is caused by being included in αm1 on the right side of the equation, the error of the base air-fuel ratio error is removed from αm1 by subtracting (1−αm2) representing the base air-fuel ratio error from αm1. However, since the sum of the two values of {α1− (1−α2)} and {αm1− (1−αm2)} centered at 1.0 is 2, the sum of the two values is 1 Is calculated by subtracting By rearranging the equation (b), the above equation (10) is obtained.
[0192]
When the total air-fuel ratio correction coefficient DLT is obtained in this manner, the equation for calculating the desorption amount is the same as the above-described equation (5b). However, in this case, since the target purge rate MPR and the intake air mass Qg must be values before the purge cut, MPR1 and Qg1 are used instead of MPR and Qg as in (11) above.
[0193]
Steps 116 and 117 are similar to the passive calibration processing. That is, in step 116, the adsorption amount Yr (mass) of the canister 24 is calculated from the desorption amount Dg calculated in step 115 by the above equation (6). In step 117, the adsorption amount Y used in calculating the desorption amount Dg based on the canister model is replaced with the adsorption amount Yr calculated in step 116.
[0194]
Since the active calibration process is completed, the data sampling end flag = 0 and the active calibration request flag = 0 in steps 118 and 119. When it is determined from the active calibration request flag = 0 that calibration can be executed next time, the process proceeds from step 111 to steps 32 to 36 in FIG. 6 to execute passive calibration processing.
[0195]
Here, the operation of the present embodiment during the execution of the purge will be described with reference to the waveform diagrams of FIGS.
[0196]
The right side of FIG. 28 shows a state in which there is no base air-fuel ratio error and the air-fuel ratio learning function is not considered, and the air-fuel ratio feedback correction coefficient α and the purge amount correction coefficient when the purge correction coefficient FHOS is correctly calculated. When the FHOS is not correctly calculated due to the calibration error, the waveforms of the FHOS are shown on the left side of FIG. 28 with no base air-fuel ratio error and no air-fuel ratio learning function considered. The waveforms of the air-fuel ratio feedback correction coefficient α and the purge correction coefficient FHOS are modeled. However, the operating condition is not ideal steady state, but shows a case where the target purge rate changes gradually as if it changes gradually.
[0197]
As shown on the right side of FIG. 28, when the purge process is performed under the set target purge rate, if the purge correction coefficient FHOS is correctly calculated, the air-fuel ratio from the stoichiometric air-fuel ratio is calculated using the FHOS. Since the deviation is small, it is not necessary to perform the air-fuel ratio feedback control. Therefore, the air-fuel ratio feedback correction coefficient α stays near the position where the correction allowance is full (maximum) (that is, 100%, which is the central value of α). ing.
[0198]
On the other hand, when the FHOS is not correctly calculated due to the calibration error, the air-fuel ratio deviation from the stoichiometric air-fuel ratio during the air-fuel ratio feedback control increases, and it is attempted to reduce the increased air-fuel ratio deviation. The air-fuel ratio feedback control works. That is, as shown on the left side of FIG. 28, α slowly fluctuates at a position with a little margin from 75%, which is the first lower limiter, which indicates that the air-fuel ratio deviation from the stoichiometric air-fuel ratio in this state. This means that is getting smaller. In other words, since the FHOS is erroneously calculated to a value close to 100%, the exhaust air-fuel ratio leans toward the rich side from the stoichiometric air-fuel ratio, and α is set to 100 to return the rich-side exhaust air-fuel ratio to the stoichiometric air-fuel ratio. %.
[0199]
In this case, in the present embodiment, 100-LMT2 is set as the second lower limiter having a smaller width to the central value (100%) of α than 75% which is the first lower limiter of α. When α changes while crossing the second lower limiter as shown on the left side of FIG. 28, the purge lower limiter determination flag repeatedly changes from 0 to 1 and returns to 0 again (FIG. 28, left, second stage from the bottom). Eyes). This is based on a comparison between α and a judgment value with hysteresis. That is, the purge lower limiter determination flag is switched from 0 to 1 at the timing when α becomes smaller across the determination value 100-LMT2 (steps 133, 134, and 135 in FIG. 20). Thereafter, when α changes to a side that becomes larger across 100-LMT2, the purge lower limiter determination flag = 1 remains at 100-LMT2 + HYS or less (steps 133, 136, and 135 in FIG. 20). , Α become greater than 100−LMT2 + HYS, the purge lower limiter determination flag is switched from 1 to 0 (steps 133, 136, 137 in FIG. 20). When α decreases again across 100-LMT2, the purge lower limiter determination flag is switched from 0 to 1 (steps 133, 134, 135 in FIG. 20). By such repetition, the purge lower limiter determination flag becomes as shown in the second row from the lower left in FIG.
[0200]
When the continuous time (timer 2 value) in which the purge lower limiter determination flag = 1 is equal to or longer than a predetermined value, the active calibration request flag is switched from 0 to 1 (FIGS. 19, steps 124 and 125).
[0201]
FIG. 29 shows that the air-fuel ratio learning control is performed, but the base air-fuel ratio error is exposed after the engine is started for some reason, and that the air-fuel ratio learning is started before the learning is started. The waveforms of the air-fuel ratio feedback correction coefficient α, the purge correction coefficient FHOS, and the air-fuel ratio learning value αm when the purge correction coefficient FHOS is not correctly calculated due to an error are modeled. That is, before t1, the exhaust air-fuel ratio leans toward the lean side from the stoichiometric air-fuel ratio due to the base air-fuel ratio error, and the air-fuel ratio feedback correction coefficient α becomes 100 to return the lean air-fuel ratio to the stoichiometric air-fuel ratio. %. However, the operating conditions are those for moderate acceleration.
[0202]
The air-fuel ratio learning value αm is obtained by dividing the entire operating region represented by the engine speed and load into several small regions, and storing a different air-fuel ratio learning value for each of the small regions. Therefore, the air-fuel ratio learning value αm changes between t1 and t2 as shown at the bottom. This indicates that the learned air-fuel ratio learning value is different because the operating condition has shifted to another small area.
[0203]
Also, as shown in FIG. 28, the target purge rate MPR, the purge correction coefficient FHOS, and the air-fuel ratio feedback correction coefficient α actually move stepwise at each calculation cycle as shown in FIG. In contrast, FIG. 29 shows a straight line. The timing of t1 and t3 is the timing of the left broken line position of the two vertical broken lines, and the timing of the right broken line position represents the next calculation timing (the width of the two broken lines is the calculation cycle). .
[0204]
By the way, before the timing of t1, the active calibration request flag is set to 1, and the purge rate lowering process is started at the timing of t1 when the calibration execution determination flag is switched from 0 to 1 (steps 75 and 76 in FIG. 13). 77, 79). In the purge rate lowering process, the target purge rate at the start of the purge rate lowering process is decreased by a predetermined time Δt or a predetermined value Δ as an initial value (see FIG. 22), and the target purge rate reaches zero at t2 (see FIG. 22). (See the fourth row in FIG. 29). In this case, since the purge correction coefficient FHOS is a value proportional to the target purge rate, if the target purge rate is reduced in this way, the FHOS also decreases toward 100%, which is the center value of FHOS (FIG. 29 (See the fifth row).
[0205]
Here, the purge rate lowering process is a method for performing the purge cut. At the start of the purge rate lowering process before the purge cut, α, FHOS, MPR, αm, Qg at that time are stored in the corresponding memory α1, FHOS1, MPR1, αm1, and Qg1 are sampled as data before purge cut (steps 151, 152, and 153 in FIG. 23).
[0206]
When the target purge rate becomes zero, a purge cut state is established. In this state, the data at the time of the purge cut is not sampled immediately, but α and αm at the timing t3 when the stabilization time has elapsed are stored in the corresponding memory. The data is sampled as data at the time of purge cut to α2 and αm2 (steps 164, 165, 166, and 167 in FIG. 24).
[0207]
Since the data sampling before and after the purge cut is completed, the data sampling end flag is switched from 0 to 1 (steps 166 and 168 in FIG. 24), and the active calibration process is performed at the timing of t4 using the data before and after the purge cut. Done.
[0208]
In this active calibration process, the total air-fuel ratio correction coefficient DLT and the desorption amount Dg are calculated by the above equations (10) and (11) instead of the above equations (5a) and (5b), and thus obtained. The adsorption amount Yr is calculated from the desorption amount Dg by an inverse operation, and the data of the adsorption amount in the canister model is replaced with the adsorption amount Yr (steps 111, 112, 113, 114, 115, 116 and 117 in FIG. 6). As a result, after the active calibration process, the purge correction coefficient FHOS is correctly calculated.
[0209]
After the completion of the active calibration process, the active calibration request flag = 0 and the data sampling flag = 0 (steps 118 and 119 in FIG. 6), whereby the target purge rate is returned to the value before the purge cut (step 13 in FIG. 13). 75, 76, 77, 78). That is, the target purge rate is gradually returned to the value before the purge cut in accordance with the purge rate change amount limit value, and in response to the increase in the target purge rate, the purge correction coefficient FHOS becomes smaller from the central value of 100%. It will become. Since the FHOS at this time is correctly calculated by executing the active calibration process, α converges to 100%, which is the center value thereof, and prepares for disturbance other than purge.
[0210]
In addition, when the calibration can be executed after the active calibration process is completed, a passive calibration process, which is a calibration process similar to that of the prior application, is performed (see FIG. 30).
[0211]
In this manner, according to the present embodiment (the invention according to claim 1), although the air-fuel ratio learning control is performed, a base air-fuel ratio error becomes apparent immediately after the engine is started for some reason. In this case, as long as the passive calibration process is continued, an error corresponding to the base air-fuel ratio error occurs in the calibration of the canister model, whereby the purge correction coefficient FHOS is calculated only inaccurately. The air-fuel ratio deviation from the stoichiometric air-fuel ratio increases, and the air-fuel ratio feedback control works to reduce the air-fuel ratio deviation, thereby reducing the correction allowance of the air-fuel ratio feedback correction coefficient α. , The second limiter ± LMT2 having a smaller width to the center value of the air-fuel ratio feedback correction coefficient α than the first limiter is set, and the calculated air-fuel ratio feedback correction coefficient α is calculated. It is determined whether or not there is a base air-fuel ratio error by comparison with the second limiter. If it is determined that there is a base air-fuel ratio error, the purge cut is performed when the calibration timing is reached. Since the active calibration process (first calibration process) is performed based on the data before and after the cut, even when the base air-fuel ratio error is exposed after the engine is started, the calibration of the canister model is performed even when the calibration timing comes. This can be performed accurately without being affected by the air-fuel ratio error, thereby improving the calculation accuracy of the purge correction coefficient FHOS and avoiding wasteful use of the correction allowance for the air-fuel ratio feedback correction coefficient α.
[0212]
Also, when the operating conditions change transiently, the calculated air-fuel ratio feedback correction coefficient α may not fall on the center value side including the second limiter even if there is no base air-fuel ratio error. Immediately determining that there is a base air-fuel ratio error would result in an erroneous determination. However, according to the present embodiment (the invention according to claim 7), the base air-fuel ratio error is determined after a predetermined value (delay time) has elapsed. (Steps 121, 122, 123, 124, and 125 in FIG. 19), a base air-fuel ratio error and a normal air-fuel ratio error (there is no theoretical air in a transient state even though there is no base air-fuel ratio error). (Having an air-fuel ratio deviation from the fuel ratio) can be performed, and the accuracy of determining a base air-fuel ratio error is improved.
[0213]
Further, according to the present embodiment (the invention according to claim 8), when there is a base air-fuel ratio error, the target purge rate at the time of calibration timing is reduced to zero by a predetermined value at predetermined time intervals as an initial value. In this case, the target purge rate may be rapidly reduced even when the calibration error due to the base air-fuel ratio error occurs and the purge correction coefficient FHOS is not correctly calculated. And the target purge rate can be changed in a range where the air-fuel ratio feedback control is possible.
[0214]
Further, according to the active calibration process, it is possible to accurately estimate the state of the canister model without being affected by the base air-fuel ratio. Can be used to confirm. That is, after the active calibration process is performed at the first calibration timing after the engine is started, the desorption amount Dg calculated by the canister model becomes accurate, and thus, the determination value for comparison with the integrated purge flow rate SQP Since the SLSQP is accurately calculated, the accuracy of determining whether or not the purge process has ended is improved (steps 92, 100 to 108 in FIG. 16).
[0215]
In order to improve the accuracy of the air-fuel ratio learning control, it is desirable to perform the control in a state where disturbance factors such as purge are removed. On the other hand, a purge process is necessary to process the evaporated fuel generated from the fuel tank 21, and performing the purge process as much as possible immediately after the start of the engine leads to the prevention of release of the purge fuel from the vehicle to the atmosphere. As described above, the two controls of the air-fuel ratio learning control and the purge process have conflicting requirements. However, when the purge fuel can be sufficiently processed, the purge rate (purge flow rate) is held at the minimum necessary value (see FIG. By updating the air-fuel ratio learning value (13 steps 80 and 81, FIG. 25) (FIG. 26 steps 171 to 173, FIG. 27 steps 181 to 183), it is possible to guarantee the accuracy of the air-fuel ratio learning control. I have.
[0216]
As described above, according to the present embodiment (the invention according to claim 9), the accuracy of the purge process end determination is increased by utilizing the result of the active calibration process in the determination of the end of the purge process. Since the air-fuel ratio learning value is updated by reducing the purge rate after the purge end determination is made with high accuracy, it is possible to achieve both improvement in the air-fuel ratio learning control accuracy and a large amount of purge processing immediately after the start. .
[0219]
Further, the desorption characteristics of the fuel from the canister 24 are affected by the temperature of the fuel adsorbent (eg, activated carbon) in the canister. Therefore, the desorption amount calculation equation is an equation that takes into account the relationship between the adsorbent temperature and the desorption amount. (The invention according to claim 15). In a region where the desorption amount is large, the calculation error of the desorption amount has a large effect on the exhaust air-fuel ratio, and particularly high calculation accuracy is required. Further, since the desorption phenomenon of the fuel from the canister 24 is an endothermic reaction, a large amount of When desorption is performed, the temperature of the adsorbent in the canister decreases and the desorption characteristics also change. Therefore, if the relationship between the adsorbent temperature and the amount of desorption is not considered, the accuracy of calculation of the amount of desorption decreases in a region where the amount of desorption is large, but the canister is calculated so as to calculate the amount of desorption taking into account the temperature of the adsorbent. By configuring the model, it is possible to suppress the deterioration of the calculation accuracy of the desorption amount, and to prevent the deterioration of the exhaust emission caused by the deterioration of the calculation accuracy.
[0218]
The desorption amount calculation formula for calculating the desorption amount based on the adsorbed fuel amount, the purge rate, and the adsorbent temperature is based on Freundlich's formula and is applied to the desorption phenomenon of the canister 24 based on Freundrich's formula. Can be used. Further, the adsorbent temperature can be obtained by calculating the amount of decrease in the adsorbent temperature based on the amount of fuel desorbed from the canister 24 (the invention according to claim 16).
[0219]
Further, according to the present embodiment (the invention described in claim 11), when the calibration timing comes when there is no base air-fuel ratio error, the passive calibration process ( A second calibration process is also performed. The canister model is an approximation model, and the value calculated by the canister model includes some errors.The canister model calculates a new value using the previous value, so that the operation time increases as the operation time increases. The error is increased by integration, but by performing such a passive calibration process, the error can be corrected appropriately and high calculation accuracy can be maintained. In addition, this passive calibration process is to reversely calculate the adsorption amount from the desorption amount estimated from the air-fuel ratio deviation from the stoichiometric air-fuel ratio, and to calibrate the adsorption amount of the internal variable with the reversely calculated adsorption amount, There is no need to follow a special sequence, and no need to lower the purge rate. The amount of adsorption can be obtained by the inverse operation of the desorption amount calculation formula in the canister model (the invention according to claim 17).
[0220]
The above calibration process is performed when an air-fuel ratio deviation from the stoichiometric air-fuel ratio does not occur due to disturbances other than the purge. Since the calibration is performed based on the actual adsorption amount of the canister 24 estimated and calculated backward from the estimated desorption amount, the accuracy of the calibration process can be secured (the invention according to claim 18).
[0221]
Further, according to the present embodiment (the invention according to claims 19 and 20), the calibration process is performed when the air-fuel ratio deviation from the stoichiometric air-fuel ratio by purge is relatively large (purging (When the rate is high and the purge air-fuel ratio is high). Depending on the operation region, sufficient calculation accuracy of the calibration process may not be obtained.However, by performing the calibration process only in the region where the sufficient accuracy is obtained, the accuracy of the calibration process is improved, and the canister model is improved. Can be further improved.
[0222]
As described above, the embodiments of the present invention have been described. However, the configuration of the above embodiment is an example of the configuration to which the present invention is applied, and the scope of the present invention is not limited to the above configuration. As described above, in the above-described embodiment, the purge process shown in FIG. 16 is supplementarily executed until the purge process by the canister model becomes possible, but the purge process shown in FIG. 16 is continuously used. You may do so.
[Brief description of the drawings]
FIG. 1 is an overall configuration diagram of a fuel vapor processing apparatus according to the present invention.
FIG. 2 is a block diagram showing an outline of a purge process in an engine controller.
FIG. 3 is a block diagram showing a configuration of a canister model.
FIG. 4 is a flowchart for explaining calculation of a purge correction coefficient.
FIG. 5 is a flowchart showing calibration processing possible conditions.
FIG. 6 is a flowchart illustrating a calibration process.
FIG. 7 is a flowchart for explaining a calculation of a desorption amount based on a canister model.
FIG. 8 is a flowchart for explaining delay correction.
FIG. 9 is a characteristic diagram showing a relationship between an intake air flow rate and a dead time.
FIG. 10 is a characteristic diagram showing a relationship between an intake air flow rate and a smoothing coefficient.
FIG. 11 is a diagram showing an outline of dead time processing in delay correction.
FIG. 12 is a flowchart for explaining setting of a target purge rate.
FIG. 13 is a flowchart for explaining setting of a target purge rate.
FIG. 14 is a characteristic diagram showing a relationship between a purge air-fuel ratio error with respect to an intake air flow rate and a purge rate.
FIG. 15 is a characteristic diagram showing a relationship between a purge air-fuel ratio and a purge air-fuel ratio error.
FIG. 16 is a flowchart for explaining boot-up control.
FIG. 17 is a characteristic diagram showing the relationship between the difference between the air-fuel ratio feedback deviations (= target air-fuel ratio feedback deviation−actual air-fuel ratio feedback deviation) and the purge rate change amount.
FIG. 18 is a characteristic diagram showing a relationship between a purge air-fuel ratio and a purge end determination value.
FIG. 19 is a flowchart for explaining setting of an active calibration request flag.
FIG. 20 is a flowchart illustrating setting of a purge lower limiter determination flag.
FIG. 21 is a flowchart for explaining setting of a purge upper limiter determination flag.
FIG. 22 is a waveform chart for explaining a purge rate lowering process.
FIG. 23 is a flowchart illustrating a purge rate lowering process.
FIG. 24 is a flowchart illustrating a purge rate lowering process.
FIG. 25 is a waveform chart showing a change in a target purge rate when it is determined that a purge has been completed.
FIG. 26 is a flowchart illustrating the setting of an air-fuel ratio learning start flag.
FIG. 27 is a flowchart for explaining updating of an air-fuel ratio learning value.
FIG. 28 is a waveform chart for comparing and explaining a case where calibration is normally performed and a case where there is a calibration error.
FIG. 29 is a waveform chart showing the contents of an active calibration process.
FIG. 30 is a waveform chart showing the contents of a passive calibration process.
FIG. 31 is a block diagram showing an outline of a calibration process.
[Explanation of symbols]
4 Fuel injection valve
5 Exhaust passage
8 Intake passage
11 Engine controller
13 Air flow meter
15 Air-fuel ratio sensor (air-fuel ratio detecting means)
21 Fuel tank
22 Piping
24 Canister
26 Piping
27 Purge valve

Claims (20)

Air-fuel ratio detection means for detecting the air-fuel ratio in the exhaust gas, and air-fuel ratio feedback correction amount calculation means for calculating the air-fuel ratio feedback correction amount so that the exhaust air-fuel ratio detected by the air-fuel ratio detection means matches the stoichiometric air-fuel ratio When,
First limiter setting means for setting a first limiter having a predetermined width from a center value of the air-fuel ratio feedback correction amount;
Air-fuel ratio feedback correction amount limiting means for limiting the calculated air-fuel ratio feedback correction amount such that the air-fuel ratio feedback correction amount falls on the side of the center value including the first limiter;
An air-fuel ratio feedback correction means for correcting the amount of fuel supplied to the engine with the limited air-fuel ratio feedback correction amount;
A canister that adsorbs fuel vapor generated in the fuel tank;
A purge valve that opens and closes a pipe that communicates the canister with the intake passage of the engine;
Purge valve control means for controlling a purge valve so that a target purge rate is obtained;
at least,
(A) an adsorbed fuel amount calculation formula for calculating the amount of fuel adsorbed on the canister based on the previous value of the amount of fuel adsorbed on the canister and the previous value of the amount of fuel desorbed from the canister;
(B) a canister model comprising an adsorbed fuel amount calculated by an adsorbed fuel amount calculation formula and a desorbed fuel amount calculation formula for calculating an amount of fuel desorbed from the canister based on the target purge rate;
Purge correction that calculates the purge correction based on the desorbed fuel calculated using the canister model so that the air-fuel ratio deviation from the stoichiometric air-fuel ratio is reduced by performing the purge processing at the target purge rate. Quantity calculation means,
Purge amount correction means for correcting the amount of fuel supplied to the engine with the purge amount correction amount;
Calibration timing determination means for determining whether or not the calibration timing,
Means for setting a second limiter having a smaller width to the center value of the air-fuel ratio feedback correction amount than the first limiter;
Base air-fuel ratio error determining means for determining whether there is a base air-fuel ratio error by comparing the calculated air-fuel ratio feedback correction amount with the second limiter;
A purge cut means for performing a purge cut when the calibration timing comes when there is a base air-fuel ratio error from the determination result;
An air-fuel ratio control device for an engine, comprising: first calibration processing means for performing a first calibration process on a canister model based on data before and after the purge cut. The first calibration processing means includes:
When the calibration timing is reached when there is a base air-fuel ratio error, pre-purge cut data sampling means for sampling data at that time as pre-purge cut data,
A purge cut data sampling means for sampling the air-fuel ratio feedback correction amount at the time of the purge cut as purge cut data,
Desorbed fuel amount estimating means for estimating the amount of fuel desorbed from the canister based on the data before and after the purge cut,
Adsorption fuel amount back calculation means for back calculating the amount of fuel adsorbed in the canister from the estimated desorbed fuel amount,
2. The air-fuel ratio control device for an engine according to claim 1, further comprising calibration means for calibrating the value of the adsorbed fuel amount, which is an internal variable in the canister model, with the value of the back calculated adsorbed fuel amount. 3. The engine air-fuel ratio control device according to claim 1, wherein an air-fuel ratio learning value is added as purge cut data when the air-fuel ratio learning control is performed. 3. The air-fuel ratio control device for an engine according to claim 2, wherein the data before the purge cut is at least an air-fuel ratio feedback correction amount, a purge correction amount, and a target purge rate. 3. The air-fuel ratio control device for an engine according to claim 2, wherein an air-fuel ratio learning value is added to the data before purge cut when the air-fuel ratio learning control is performed. 3. The engine air-fuel ratio control device according to claim 2, wherein the sampling timing of the data sampling means at the time of purge cut is a timing at which a stabilization time has elapsed since the target purge rate became zero. When the calculated air-fuel ratio feedback correction amount, including the second limiter, does not fall on the side of the center value, the base air-fuel ratio error determination means includes a time measuring means for measuring a time during which the out-of-range condition continues, and 3. The air-fuel ratio control device for an engine according to claim 1, further comprising comparison means for determining whether there is a base air-fuel ratio error by comparing the time with a predetermined value. The purge cut means is a purge rate reduction processing means for performing processing of reducing the target purge rate to zero at predetermined time intervals by a predetermined value as an initial value when the calibration timing is reached when there is a base air-fuel ratio error. The engine air-fuel ratio control device according to claim 1 or 2, wherein: When the calibration timing comes for the first time after the start of the engine, the calibration process at the initial calibration timing is performed in the first calibration process, and the calibration process is performed according to the desorption amount calculated using the canister model calibrated by the first calibration process. A determination value is calculated, and it is determined whether the purge process has been completed based on the determination value.When it is determined that the purge process has been completed, the purge rate or the purge flow rate is reduced to update the air-fuel ratio learning value. 3. The air-fuel ratio control device for an engine according to claim 1, wherein the control is performed. The air-fuel ratio of an engine according to claim 9, wherein the determination value is a determination value for an integrated purge flow rate, and when the integrated purge flow rate from engine start is equal to or greater than the determination value, it is determined that the purge process has been completed. Control device. 3. The air-fuel ratio of an engine according to claim 1, wherein a second calibration process is performed on the canister model based on data during execution of the purge when the calibration timing comes when there is no base air-fuel ratio error. Control device. The second calibration processing means includes:
Desorbed fuel amount estimating means for estimating the amount of fuel desorbed from the canister based on data during the execution of the purge;
Adsorption fuel amount back calculation means for back calculating the amount of fuel adsorbed in the canister from the estimated desorbed fuel amount,
12. The air-fuel ratio control device for an engine according to claim 11, comprising calibration means for calibrating the value of the adsorbed fuel amount, which is an internal variable in the canister model, with the value of the back calculated adsorbed fuel amount. 12. The air-fuel ratio control device for an engine according to claim 10, wherein the data during the execution of the purge is at least an air-fuel ratio feedback correction amount, a purge amount correction amount, and a target purge rate. 3. The air-fuel ratio control device for an engine according to claim 1, wherein a hysteresis is provided in the second limiter. The desorbed fuel amount calculation formula of the canister model calculates the desorbed fuel amount from the canister based on the adsorbed fuel amount calculated by the adsorbed fuel amount calculation formula, the target purge rate, and the temperature of the fuel adsorbent in the canister. The air-fuel ratio control device for an engine according to claim 1 or 2, wherein: The canister model further includes an adsorbent temperature calculation formula for calculating the temperature drop of the fuel adsorbent based on the amount of fuel desorbed from the canister, and calculating the temperature of the fuel adsorbent based on the temperature drop. The air-fuel ratio control device for an engine according to claim 15, wherein: The means for back-calculating the amount of fuel adsorbed on the canister calculates the amount of fuel adsorbed on the canister from the estimated amount of desorbed fuel by the inverse operation of the desorbed fuel amount calculation formula of the canister model. 3. The air-fuel ratio control device for an engine according to claim 1, wherein: The means for estimating the desorbed fuel amount estimates the desorbed fuel amount from the canister based on the air-fuel ratio deviation from the stoichiometric air-fuel ratio when the air-fuel ratio deviation from the stoichiometric air-fuel ratio does not occur due to disturbance other than the purge. The engine air-fuel ratio control device according to claim 1 or 2, wherein: Means for estimating the amount of desorbed fuel is based on the fact that the purge air-fuel ratio is richer than a predetermined value and the air-fuel ratio deviation from the stoichiometric air-fuel ratio due to the purge sufficiently appears. 3. The air-fuel ratio control device for an engine according to claim 1, wherein a fuel amount is estimated. Means for estimating the amount of desorbed fuel is based on the deviation of air-fuel ratio from stoichiometric air-fuel ratio when the purge rate is higher than a predetermined value and the air-fuel ratio deviation from stoichiometric air-fuel ratio due to purge is sufficient. 3. The air-fuel ratio control device for an engine according to claim 1, wherein the amount is estimated.

Images