JP4196494B2 - Control device for internal combustion engine - Google Patents

Description

式（１）からわかるように、位相進み補償後スロットル開度ＴＡｈは、上記実際のスロットル開度ＴＡｒを時間ｔについて微分して更に所定の係数Ｋｄを乗算し、その値を実際のスロットル開度ＴＡｒに加算して算出される値である。こうして算出される位相進み補償後スロットル開度ＴＡｈは、目標スロットル開度ＴＡｓｊの変化中においては、実際のスロットル開度ＴＡｒよりも同目標スロットル開度ＴＡｓｊに近い値になる。
【００５２】
ＥＣＵ９２は、目標スロットル開度ＴＡｓｊと上記位相進み補償後スロットル開度ＴＡｈとの差ｅ２を算出する。そして、ＥＣＵ９２は、その差ｅ２が「０」に近づくように、即ち位相進み補償後スロットル開度ＴＡｈが目標スロットル開度ＴＡｓｊに近づくようにスロットル用モータ２４を駆動制御する。
【００５３】
ここで、時間経過に伴い目標スロットル開度ＴＡｓｊが変化するときにおいて、位相進み補償後スロットル開度ＴＡｈ、及び実際のスロットル開度ＴＡｒがどのように推移するかを図３に示す。
【００５４】
図３に二点鎖線で示すように目標スロットル開度ＴＡｓｊが変化すると、それに応じて位相進み補償後スロットル開度ＴＡｈが細い実線で示すように、その目標スロットル開度ＴＡｓｊの近傍で推移する。このように推移する位相進み補償後スロットル開度ＴＡｈと、目標スロットル開度ＴＡｓｊとの差ｅ２が「０」に近づくようにスロットル用モータ２４を制御すると、実際のスロットル開度ＴＡｒは目標スロットル開度ＴＡｓｊの推移に対して太い実線で示すように所定の応答遅れをもって推移する。こうして実際のスロットル開度ＴＡｒに応答遅れを持たせるのは、そのスロットル開度ＴＡｒのオーバーシュートを防止するためである。
【００５５】
このように「均質燃焼」時のスロットル開度制御が行われると、エンジン１１における実際の吸気圧ＰＭｒ（吸入空気量）がスロットルバルブ２３の開度に対応したものとなる。ＥＣＵ９２は、実際のスロットル開度ＴＡｒ及び実際の吸気圧ＰＭｒ等から予測吸気圧ＰＭＦＷＤを算出する。この予測吸気圧ＰＭＦＷＤは、吸気バルブ１９の閉弁時における吸気圧を予測した値であって、後述する吸気圧算出ルーチンによって算出されるものである。
【００５６】
ＥＣＵ９２は、上記予測吸気圧ＰＭＦＷＤ及びエンジン回転数ＮＥに基づき基本燃料噴射量Ｑbse を算出する。そして、ＥＣＵ９２は、燃料噴射弁４０を駆動制御して、基本燃料噴射量Ｑbse から求められる最終燃料噴射量Ｑfin に対応した量の燃料を吸気行程中に燃焼室１６に噴射供給する。こうした燃料噴射によって「均質燃焼」が実行され、機関出力トルクが必要とされる値へと制御される。
【００５７】
従って、エンジン１１の均質燃焼運転時には、集約目標スロットル開度ＴＡｔが代入される目標スロットル開度ＴＡｓｊに基づきスロットル開度を調節することにより機関出力トルクが制御される。
【００５８】
次に、「成層燃焼」時の燃料噴射制御について説明する。
上記集約目標スロットル開度ＴＡｔは、「成層燃焼」時には、そのときのアクセル踏込量ＡＣＣＰ及びエンジン回転数ＮＥにて「均質燃焼」を実行すると仮定した場合での同「均質燃焼」（以下、「仮想均質燃焼」という）に適した目標スロットル開度として算出される。「成層燃焼」時には、上記「仮想均質燃焼」において上記集約目標スロットル開度ＴＡｔに基づきスロットル開度制御を行うことで得られるはずの実際のスロットル開度ＴＡｒを、同集約目標スロットル開度ＴＡｔに基づき仮想スロットル開度ＴＡｖとして算出する。
【００５９】
即ち、図３に示すように、「均質燃焼」時の目標スロットル開度ＴＡｓｊ（集約目標スロットル開度ＴＡｔ）の推移と、位相進み補償後スロットル開度ＴＡｈの推移とがほぼ等しいことから、まず「ＴＡｈ＝ＴＡｓｊ（ＴＡｈ＝ＴＡｔ）」であると仮定する。この仮定を条件に、上記式（１）と逆の手順により、目標スロットル開度ＴＡｓｊ（集約目標スロットル開度ＴＡｔ）から実際のスロットル開度ＴＡｒを算出し、そのスロットル開度ＴＡｒを仮想スロットル開度ＴＡｖとする。
【００６０】
更に、エンジン１１のスロットル開度が上記仮想スロットル開度ＴＡｖとなる状態での「均質燃焼（仮想均質燃焼）」時の吸気圧（予測吸気圧ＰＭＦＷＤ）である仮想吸気圧ＰＭｖを、上記仮想スロットル開度ＴＡｖ等に基づき算出する。そして、上記仮想吸気圧ＰＭｖ（仮想吸入空気量）及びエンジン回転数ＮＥ等に基づき「成層燃焼」時の基本燃料噴射量Ｑbse を算出する。
【００６１】
ＥＣＵ９２は、燃料噴射弁４０を駆動制御して、基本燃料噴射量Ｑbse から求められる最終燃料噴射量Ｑfin に対応した量の燃料を圧縮行程中に燃焼室１６に噴射供給する。上記最終燃料噴射量Ｑfin は、エンジン１１における燃料噴射量の制御目標値である。こうした最終燃料噴射量Ｑfin に基づく燃料噴射制御によって「成層燃焼」が実行され、機関出力トルクが必要とされる値へと制御される。
【００６２】
従って、エンジン１１の成層燃焼運転時には、集約目標スロットル開度ＴＡｔに基づき、「仮想均質燃焼」に対応する仮想スロットル開度ＴＡｖ及び仮想吸気圧ＰＭｖが算出され、この仮想吸気圧ＰＭｖ等から算出される最終燃料噴射量Ｑfin （基本燃料噴射量Ｑbse ）に基づき燃料噴射量を調節することにより機関出力トルクが制御される。このため、「成層燃焼」時には、上記集約目標スロットル開度ＴＡｔは、そのときの「成層燃焼」時に必要とされる機関出力トルクに対応した値として、後述する集約目標スロットル開度算出ルーチンにより算出される。
【００６３】
上記のように成層燃焼運転時には、「仮想均質燃焼」に対応する仮想スロットル開度ＴＡｖ及び仮想吸気圧ＰＭｖに応じて最終燃料噴射量Ｑfin が算出され、同最終燃料噴射量Ｑfin がこのとき実際に「均質燃焼」を実行した場合に得られる機関出力トルクに関連付けられる。そして、この最終燃料噴射量Ｑfin に基づき燃料噴射量が制御されるため、同最終燃料噴射量Ｑfin に対応した成層燃焼運転時の機関出力トルクが、このとき均質燃焼運転を行った場合での機関出力トルクに合致するようになる。これにより、燃焼方式を「成層燃焼」と「均質燃焼」との間で切り換える際等に、機関出力トルクに段差が生じるのを抑制することができる。
【００６４】
次に、最終燃料噴射量Ｑfin の算出手順について図４を参照して説明する。図４は、成層燃焼運転時及び均質燃焼運転時に仮想吸気圧ＰＭｖ及び予測吸気圧ＰＭＦＷＤに基づき最終燃料噴射量Ｑfin を算出するための燃料噴射量算出ルーチンを示すフローチャートである。この燃料噴射量算出ルーチンは、ＥＣＵ９２を通じて所定時間毎の時間割り込みにて実行される。
【００６５】
燃料噴射量算出ルーチンにおいて、ステップＳ２０１の処理は仮想吸気圧ＰＭｖ若しくは予測吸気圧ＰＭＦＷＤを算出するためのものである。このステップＳ２０１の処理が実行された後、ＥＣＵ９２は、ステップＳ２０２の処理として、仮想吸気圧ＰＭｖ若しくは予測吸気圧ＰＭＦＷＤを吸気圧ＰＭとして用い、下記の式（２）によって基本燃料噴射量Ｑbse を算出する。即ち、吸気圧ＰＭに吸気温補正係数Ｋtha 及び定数Ｋを乗算して基本燃料噴射量Ｑbse を算出する。
【００６６】
【数２】
Ｑbse ＝ＰＭ＊Ｋtha ＊Ｋ …（２）
なお、上記仮想吸気圧ＰＭｖ及び予測吸気圧ＰＭＦＷＤを算出する際には後述する体積効率ηｖが用いられるが、上記式（２）における吸気温補正係数Ｋtha は吸気温ＴＨＡの変化による体積効率ηｖの変化を補償するためのものである。ＥＣＵ９２は、吸気温センサ３７からの検出信号に基づき吸気温ＴＨＡを求めるとともに、上記吸気温補正係数Ｋtha を吸気温ＴＨＡに基づき図５のマップを参照して算出する。こうして算出される吸気温補正係数Ｋtha は、吸気温ＴＨＡが高くなるほど小さくなって「１．０」に近づくようになる。従って、補正後の基本燃料噴射量Ｑbse は、吸気温ＴＨＡが低くなるほど大きな値になる。
【００６７】
ステップＳ２０２の処理が実行された後、ステップＳ２０３に進む。ステップＳ２０３，Ｓ２０４の処理は、基本燃料噴射量Ｑbse 等から最終燃料噴射量Ｑfin を算出するためのものである。
【００６８】
ＥＣＵ９２は、ステップＳ２０３の処理として、モード補正係数Ｋmodeを算出する。このモード補正係数Ｋmodeは、「均質燃焼」と「成層燃焼」との燃焼効率の差に伴う要求燃料噴射量の差を補償するための補正係数であり、ＥＣＵ９２は、現在の燃焼方式に応じてモード補正係数Ｋmodeを算出する。このモード補正係数Ｋmodeは、燃焼効率が「成層燃焼」よりも低くなる「均質燃焼」時には、「Ｋmode＝１．０」に設定される。なお、「均質燃焼」時に「成層燃焼」時よりも燃焼効率が低くなるのは、「均質燃焼」では「成層燃焼」に比べてポンプ損失や冷却損失が大きくなるためである。
【００６９】
上記のようにステップＳ２０３の処理を実行し、モード補正係数Ｋmodeを算出すると、ＥＣＵ９２は、続くステップＳ２０４で、基本燃料噴射量Ｑbse にモード補正係数Ｋmodeを乗算して最終燃料噴射量Ｑfin を算出する。
【００７０】
上記のように最終燃料噴射量Ｑfin の算出にモード補正係数Ｋmodeを用いることで、燃焼方式毎の燃焼効率の違いに基づき最終燃料噴射量Ｑfin が調整され、燃焼効率の高い「成層燃焼」時には「均質燃焼」時に対して最終燃料噴射量Ｑfin が減量側に調整される。こうした燃焼方式毎の燃焼効率の違いを加味して算出される最終燃料噴射量Ｑfin に基づき燃料噴射制御を行うことで、いずれの燃焼方式を実行したときでも燃料噴射量制御に基づく機関出力トルク制御の精度が向上するようになる。
【００７１】
ＥＣＵ９２は、続くステップＳ２０５の処理として、成層燃焼運転の実行中か否かを判断する。そして、成層燃焼運転中でなければ当該燃料噴射量算出ルーチンを一旦終了し、成層燃焼運転中であればステップＳ２０６の処理を実行した後に同燃料噴射量算出ルーチンを一旦終了する。上記ステップＳ２０６の処理は、「成層燃焼」時に燃料系の経年変化等に伴う燃料噴射量の適正値に対するずれを補償するためのものである。
【００７２】
ＥＣＵ９２は、ステップＳ２０６の処理として、下記の式（３）に基づき最終燃料噴射量Ｑfin を補正する。
【００７３】
【数３】

式（３）において、最終燃料噴射量Ｑfin に加算される項、即ち「｛（ｑｇｔｊ／ｑｇｓｊ）−１｝＊Ｃ＊（６００／ＮＥ）」により、「成層燃焼」時の上記燃料噴射量の適正値に対するずれが補償されることとなる。なお、この補償に関しては後で詳しく説明する。式（３）において、成層用学習値ｑｇｔｊは、後述するＩＳＣ学習ルーチンによって上記燃料噴射量の適正値からのずれ量に対応した値として学習されるものである。また、均質用学習値ｑｇｓｊは、上記ＩＳＣ学習ルーチンによって吸気系の経年変化等に伴う吸入空気量の適正値からのずれ量に対応した値として学習されるものである。また、式（３）において、重み係数Ｃ及びエンジン回転数ＮＥによって定まる「Ｃ＊（６００／ＮＥ）」という値は、式（３）に基づく最終燃料噴射量Ｑfin の補正が過剰になるのを抑制し、同補正の適正化を図るためのものである。
【００７４】
次に、燃料噴射量算出ルーチンにおけるステップＳ２０１の処理について図６及び図７を参照して詳しく説明する。図６及び図７は、予測吸気圧ＰＭＦＷＤ及び仮想吸気圧ＰＭｖを算出するための吸気圧算出ルーチンを示すフローチャートである。この吸気圧算出ルーチンは、燃料噴射量算出ルーチンのステップＳ２０１に進む毎にＥＣＵ９２を通じて実行される。
【００７５】
吸気圧算出ルーチンにおいて、ＥＣＵ９２は、ステップＳ３０１（図６）の処理として、成層燃焼運転中であるか否かを判断する。そして、成層燃焼運転中であれば、ステップＳ３０２の処理として、集約目標スロットル開度ＴＡｔと実際のスロットル開度ＴＡｒとに基づき、「仮想均質燃焼」時のスロットル開度ＴＡｔである仮想スロットル開度ＴＡｖを算出する。その後、ステップＳ３０３に進む。また、上記ステップＳ３０１の処理において、成層燃焼運転中でない旨判断されるときには直接ステップＳ３０３に進む。
【００７６】
ＥＣＵ９２は、ステップＳ３０３の処理として、現在の実際のスロットル開度ＴＡｒ若しくは仮想スロットル開度ＴＡｖと、エンジン回転数ＮＥとに基づき定常時吸気圧ＰＭbse を算出する。この定常時吸気圧ＰＭbse は、上記スロットル開度ＴＡｒ，ＴＡｖ及びエンジン回転数ＮＥの状態にあって、エンジン１１を定常運転したときの吸気圧である。なお、定常時吸気圧ＰＭbse は、均質燃焼運転時には実際のスロットル開度ＴＡｒ及びエンジン回転数ＮＥに基づき算出され、成層燃焼運転時には仮想スロットル開度ＴＡｖ及びエンジン回転数ＮＥに基づき算出される。
【００７７】
ＥＣＵ９２は、ステップＳ３０４の処理として、大気圧補正係数Ｋpa１を大気圧ＰＡに基づき図８のマップを参照して算出し、定常時吸気圧ＰＭbse にこの大気圧補正係数Ｋpa１を乗算することにより、補正後吸気圧ＰＭｈを算出する。なお、上記大気圧補正係数Ｋpa１は大気圧ＰＡが高くなるほど大きくなって「１．０」に近づくようになる。従って、補正後吸気圧ＰＭｈは大気圧ＰＡが高くなるほど大きくなる。補正後吸気圧ＰＭｈの算出が行われた後、ステップＳ３０５に進む。
【００７８】
このステップＳ３０５の処理は後のステップＳ３０６，Ｓ３０７の処理と関係している。即ち、ステップＳ３０６の処理では上記補正後吸気圧ＰＭｈを徐変処理することにより徐変値ＰＭＳＭが算出され、ステップＳ３０７の処理では同徐変値ＰＭＳＭが第１の吸気圧記憶値ＰＭＳＭ１として記憶される。そして、上記ステップＳ３０５の処理においては、ＥＣＵ９２が、前回のステップＳ３０７の処理で記憶された第１の吸気圧記憶値ＰＭＳＭ１を前回の徐変値ＰＭＳＭi-1 として設定する。
【００７９】
このように徐変処理（Ｓ３０６）によって算出された徐変値ＰＭＳＭを一旦第１の吸気圧記憶値ＰＭＳＭ１として記憶（Ｓ３０７）するのは、後述するステップＳ３１０の処理で上記徐変値ＰＭＳＭを用いて別の処理を実行し、その処理によって徐変値ＰＭＳＭが変化してしまうためである。この場合でも、上記ステップＳ３０５の処理で第１の吸気圧記憶値ＰＭＳＭ１を前回の徐変値ＰＭＳＭi-1 とすることで、ステップＳ３０６の徐変処理を適切に行うことができるようになる。
【００８０】
上記ステップＳ３０５の処理が実行された後、ＥＣＵ９２は、ステップＳ３０６の処理として、下記の式（４）に基づき今回の徐変値ＰＭＳＭi を算出する。即ち、定常時の補正後吸気圧ＰＭｈから前回の徐変値ＰＭＳＭi-1 を減算して更に所定値ｎで除算し、その除算した値を前回の徐変値ＰＭＳＭiー1 に加算することで今回の徐変値ＰＭＳＭi が算出される。
【００８１】
【数４】
ＰＭＳＭi ＝ＰＭＳＭi-1 ＋（ＰＭｈ−ＰＭＳＭi-1 ）／ｎ …（４）
ここで、上記補正後吸気圧ＰＭｈの変化に対する徐変値ＰＭＳＭの推移傾向を図９に示す。同図においては補正後吸気圧ＰＭｈの推移を破線で示し、徐変値ＰＭＳＭの推移を太い実線で示す。また、マップ演算等により算出される上記補正後吸気圧ＰＭｈが破線で示すように推移するのに対し、実際の吸気圧ＰＭｒがどのように推移するかを二点鎖線で示す。
【００８２】
この図から明らかなように、例えばアクセル踏込量ＡＣＣＰが変化して上記補正後吸気圧ＰＭｈが破線で示すように変化したとき、その補正後吸気圧ＰＭｈの変化に対して徐変値ＰＭＳＭが太い実線で示すように緩やかに推移するようになる。補正後吸気圧ＰＭｈの変化に対して徐変値ＰＭＳＭがどれほど緩やかに推移するかは、上記式（４）における所定値ｎによって決定される。この所定値ｎは、予め実験等により設定された図示しないマップを参照して上記補正後吸気圧ＰＭｈとエンジン回転数ＮＥとに基づき算出される。
【００８３】
ステップＳ３０６の処理で徐変値ＰＭＳＭが算出され、ステップＳ３０７の処理で第１の吸気圧記憶値ＰＭＳＭ１の記憶が行われると、続いてステップＳ３０８に進む。ステップＳ３０８〜Ｓ３１１の処理は、現時点で吸気バルブ１９の閉弁時における徐変値ＰＭＳＭを予測して算出するためのものである。
【００８４】
ＥＣＵ９２は、ステップＳ３０８の処理として、現時点から吸気バルブ１９の閉弁時までに上記ステップＳ３０６の処理が行われる回数（徐変処理回数）Ｔ／Δｔを算出する。即ち、現時点から吸気バルブ１９の閉弁時までの時間Ｔを求め、その時間Ｔを燃料噴射量算出ルーチンの実行周期Δｔで除算することにより、上記徐変処理回数Ｔ／Δｔを算出する。
【００８５】
続いてＥＣＵ９２は、ステップＳ３０９の処理として現在記憶されている第１の吸気圧記憶値ＰＭＳＭ１、即ち最新の徐変値ＰＭＳＭを前回の徐変値ＰＭＳＭi-1 として設定する。更に、ＥＣＵ９２は、ステップＳ３１０の処理として、上記徐変処理回数Ｔ／Δｔ分だけ上記式（４）による徐変処理を実行し、Ｔ／Δｔ回の徐変処理後の徐変値ＰＭＳＭi 、即ち吸気バルブ１９の閉弁時の徐変値ＰＭＳＭi を算出する。その後、ＥＣＵ９２は、ステップＳ３１１の処理として、徐変値ＰＭＳＭi を第２の吸気圧記憶値ＰＭＳＭ２として記憶する。
【００８６】
今、図９に一点鎖線Ｌ１で示す時点にて上記ステップＳ３０６の処理が行われたとすると、その処理によって算出される今回の徐変値ＰＭＳＭi が第１の吸気圧記憶値ＰＭＳＭ１として記憶される。そして、続いてステップＳ３１０の処理が行われると、二点鎖線Ｌ２で示す吸気バルブ１９の閉弁時における徐変値ＰＭＳＭi が算出され、その徐変値ＰＭＳＭi がほぼ一点鎖線Ｌ１で示す時点にて第２の吸気圧記憶値ＰＭＳＭ２として記憶される。
【００８７】
このように第１及び第２の吸気圧記憶値ＰＭＳＭ１，ＰＭＳＭ２の記憶処理が行われた後には、それら記憶値ＰＭＳＭ１，ＰＭＳＭ２の差ΔＰ１（「ＰＭＳＭ２−ＰＭＳＭ１」）を用いて、吸気バルブ１９の閉弁時における吸気圧を予測して算出することができるようになる。即ち、現時点（一点鎖線Ｌ１）においてバキュームセンサ３６により検出される実際の吸気圧ＰＭｒに、上記第１及び第２の吸気圧記憶値ＰＭＳＭ１，ＰＭＳＭ２の差ΔＰ１を加算することで、吸気バルブ１９の閉弁時における吸気圧が得られるようになる。
【００８８】
ところで、バキュームセンサ３６の出力には吸気通路３２内を流れる空気の脈動による影響が生じるため、その影響を除去するために通常はバキュームセンサ３６の出力をＣＲフィルタ等によってフィルタ処理する。従って、上記吸気圧ＰＭｒは実際にはＣＲフィルタ等によるフィルタ処理の時定数分だけ適正値からずれることになり、そのずれの分だけ上記予測される吸気バルブ１９の閉弁時の吸気圧が不正確になる。
【００８９】
吸気圧算出ルーチンにおけるステップＳ３１２（図７）〜Ｓ３１５の処理は、上記吸気圧ＰＭｒのずれを考慮して第１の吸気圧記憶値ＰＭＳＭ１をフィルタ処理し、そのフィルタ出力ＰＭＳＭ１Ｓi を用いて吸気バルブ１９の閉弁時の吸気圧を正確に予測するためのものである。
【００９０】
ＥＣＵ９２は、上記ステップＳ３１１（図６）の処理を実行した後、ステップＳ３１２（図７）の処理として現在の燃焼方式が均質燃焼であるか否かを判断し、均質燃焼であればステップＳ３１３に進む。ＥＣＵ９２は、ステップＳ３１３の処理として、第１の吸気圧記憶値ＰＭＳＭ１を下記の式（５）に基づきフィルタ処理する。なお、式（５）において、ＰＭＳＭ１Ｓi は第１の吸気圧記憶値ＰＭＳＭ１のフィルタ出力であり、所定値ｍは当該フィルタ処理の時定数が上記ＣＲフィルタによるフィルタ処理の時定数と等しくなるように設定されるものである。
【００９１】
【数５】
ＰＭＳＭ１Ｓi
＝ＰＭＳＭ１Ｓi-1 ＋（ＰＭＳＭ１−ＰＭＳＭ１Ｓi-1 ）／ｍ …（５）
この式（５）に基づくフィルタ処理のフィルタ出力ＰＭＳＭ１Ｓi は、図９に太い実線で示すように徐変値ＰＭＳＭ（第１の吸気圧記憶値ＰＭＳＭ１）が変化したときには、図中に細い実線で示すように推移することとなる。
【００９２】
続いてＥＣＵ９２は、ステップＳ３１４の処理として、第２の吸気圧記憶値ＰＭＳＭ２から上記フィルタ出力ＰＭＳＭ１Ｓi を減算し、それらの差ΔＰ２を算出する。更に、ＥＣＵ９２は、ステップＳ３１５の処理として、実際の吸気圧ＰＭｒに上記差ΔＰ２加算し、その加算した値に更に体積効率ηｖを乗算した値を、吸気バルブ１９の閉弁時における吸気圧である予測吸気圧ＰＭＦＷＤとして算出する。なお、上記体積効率ηｖは前回の予測吸気圧ＰＭＦＷＤとエンジン回転数ＮＥとに基づきマップを参照して算出されるものである。こうして予測吸気圧ＰＭＦＷＤを算出した後、当該吸気圧算出ルーチンを一旦終了して燃料噴射量算出ルーチン（図４）に戻る。
【００９３】
従って、図９に一点鎖線Ｌ１で示す時点にて第１及び第２の吸気圧記憶値ＰＭＳＭ１，ＰＭＳＭ２の記憶処理が行われた場合、その時点での第１の吸気圧記憶値ＰＭＳＭ１のフィルタ出力ＰＭＳＭ１Ｓi が予測吸気圧ＰＭＦＷＤを算出に用いられる。即ち、一点鎖線Ｌ１で示す時点での第２の吸気圧記憶値ＰＭＳＭ２とフィルタ出力ＰＭＳＭ１Ｓi との差ΔＰ２を実際の吸気圧ＰＭｒに加算することで予測吸気圧ＰＭＦＷＤが算出される。
【００９４】
このように第１の吸気圧記憶値ＰＭＳＭ１に代えてフィルタ出力ＰＭＳＭ１Ｓi を用いて差ΔＰ２を算出し、その差ΔＰ２等から予測吸気圧ＰＭＦＷＤを求めることで、同吸気圧ＰＭｒにＣＲフィルタの時定数に応じたずれが生じても、その予測吸気圧ＰＭＦＷＤを正確な吸気バルブ１９の閉弁時の吸気圧として算出することができるようになる。
【００９５】
一方、上記ステップＳ３１２の処理において、現在の燃焼方式が均質燃焼でなく成層燃焼である旨判断されると、ステップＳ３１６に進む。ＥＣＵ９２は、ステップＳ３１６の処理として、第２の吸気圧記憶値ＰＭＳＭ２に体積効率ηｖを乗算した値を仮想吸気圧ＰＭｖとして算出する。なお、上記体積効率ηｖは前回の仮想吸気圧ＰＭｖとエンジン回転数ＮＥとに基づきマップを参照して算出されるものである。こうして仮想吸気圧ＰＭｖを算出した後、当該吸気圧算出ルーチンを一旦終了して燃料噴射量算出ルーチン（図４）に戻る。
【００９６】
上記算出される仮想吸気圧ＰＭｖは、現在の機関運転状態にて「均質燃焼」を実行したと仮定した場合（「仮想均質燃焼」）での吸気バルブ１９の閉弁時の吸気圧、即ち上記予測吸気圧ＰＭＦＷＤに対応した仮想値ということになる。均質燃焼運転時には、予測吸気圧ＰＭＦＷＤが実際の吸気圧ＰＭｒ等に基づき算出されるため、ステップＳ３１２〜Ｓ３１５の処理を行って同予測吸気圧ＰＭＦＷＤを正確に算出するようにしている。これに対し、成層燃焼運転時には、仮想吸気圧ＰＭｖが実際の吸気圧ＰＭｒに関係なく、第２の吸気圧記憶値ＰＭＳＭ２等にづき算出される。こうして算出される仮想吸気圧ＰＭｖは、ステップＳ３１６の処理によって正確な値として算出されるようになる。
【００９７】
次に、集約目標スロットル開度ＴＡｔの算出手順について図１０を参照して説明する。図１０は、集約目標スロットル開度ＴＡｔを算出するための集約目標スロットル開度ルーチンを示すフローチャートである。この集約目標スロットル開度算出ルーチンは、ＥＣＵ９２を通じて例えば所定時間毎の時間割り込みにて実行される。
【００９８】
集約目標スロットル開度算出ルーチンにおいては、ステップＳ４０５の処理で、基本スロットル開度ＴＡbse 、ＩＳＣ開度補正量ｆ（ｑｃａｌ）、及びその他の補正量Ａを用いて、下記の式（６）に基づき集約目標スロットル開度ＴＡｔが算出される。
【００９９】
【数６】
ＴＡｔ＝ＴＡbse ＋ｆ（ｑｃａｌ）＋Ａ …（６）
ところで、上記式（６）において、ＩＳＣ開度補正量ｆ（ｑｃａｌ）は、機関出力トルクを調整してアイドル回転数を制御するためのものである。このＩＳＣ開度補正量ｆ（ｑｃａｌ）は、ＩＳＣ補正量ｑｃａｌ及び変換係数Ｋａに基づき下記の式（７）によって算出される。
【０１００】
【数７】
ｆ（ｑｃａｌ）＝ｑｃａｌ＊Ｋａ …（７）
式（７）において、ＩＳＣ補正量ｑｃａｌは、アイドル回転数の調整量に対応した無次元のパラメータである。そして、変換係数Ｋａは、上記ＩＳＣ補正量ｑｃａｌをアイドル回転数の調整に必要とされる機関出力トルクの変化量、即ちスロットル開度の調整量に変換するためのものである。
【０１０１】
アイドル回転数の調整は、上記ＩＳＣ補正量ｑｃａｌを増減させることによって行われる。このＩＳＣ補正量ｑｃａｌの増減に伴いＩＳＣ開度補正量ｆ（ｑｃａｌ）が変化すると、集約目標スロットル開度ＴＡｔも変化することとなる。この集約目標スロットル開度ＴＡｔは、「均質燃焼」時の機関出力トルクに影響を及ぼすパラメータである目標スロットル開度ＴＡｓｊ、及び「成層燃焼」時の機関出力トルクに影響を及ぼすパラメータである最終燃料噴射量Ｑfin にそれぞれ関係している。そのため、集約目標スロットル開度ＴＡｔをＩＳＣ開度補正量ｆ（ｑｃａｌ）の増減により変化させることで、いずれの燃焼方式であっても機関出力トルクが変化してアイドル回転数が調整される。
【０１０２】
上記ＩＳＣ開度補正量ｆ（ｑｃａｌ）を算出するためのＩＳＣ補正量ｑｃａｌは、フィードバック補正項ｑｉ、計算用学習値ｑｇ、及び水温補正項ｑｔｈｗ等に基づき、下記の式（８）によって算出される。
【０１０３】
【数８】
ｑｃａｌ＝ｑｉ＋ｑｇ＋ｑｔｈｗ …（８）
式（８）において、フィードバック補正項ｑｉは、アイドル回転数を予め定められた目標値（例えば６００ｒｐｍ）に近づけるべく、アイドル運転時のエンジン回転数ＮＥに応じて所定の基準値（本実施形態では「０」）を中心に増減する値である。
【０１０４】
即ち、フィードバック補正項ｑｉは、アイドル回転数が目標値よりも低ければ大きくされる。その結果、ＩＳＣ補正量ｑｃａｌ及びＩＳＣ開度補正量ｆ（ｑｃａｌ）が大きくなり、ＩＳＣ開度補正量ｆ（ｑｃａｌ）等に基づき算出される集約目標スロットル開度ＴＡｔが開き側の値へと変化する。「均質燃焼」であれ、「成層燃焼」であれ、上記のように集約目標スロットル開度ＴＡｔが開き側の値に変化することにより、機関出力トルクが増大してアイドル回転数が目標値に向けて上昇する。
【０１０５】
「均質燃焼」時には、上記集約目標スロットル開度ＴＡｔが目標スロットル開度ＴＡｓｊに代入されるため、この目標スロットル開度ＴＡｓｊに基づきスロットル開度制御を行うことで予測吸気圧ＰＭＦＷＤ（吸入空気量）が増大する。その結果、予測吸気圧ＰＭＦＷＤ等に基づき決定される最終燃料噴射量Ｑfin が多くなって機関出力トルクが増大し、アイドル回転数が目標値に向かって上昇する。
【０１０６】
上記と異なり、「成層燃焼」時には、上記集約目標スロットル開度ＴＡｔ等に基づき、「仮想均質燃焼」時のスロットル開度である仮想スロットル開度ＴＡｖが算出される。更に、「仮想均質燃焼」時にスロットル開度を上記仮想スロットル開度ＴＡｖとしたときの吸気圧（吸入空気量）である仮想吸気圧ＰＭｖが上記仮想スロットル開度ＴＡｖ等に基づき算出される。そのため、ＩＳＣ開度補正量ｆ（ｑｃａｌ）により集約目標スロットル開度ＴＡｔが開き側の値に変化すると仮想吸気圧ＰＭｖが増大する。その結果、同仮想吸気圧ＰＭｖ等に応じて決定される最終燃料噴射量Ｑfin が多くなり、機関出力トルクが増大してアイドル回転数が目標値に向かって上昇する。
【０１０７】
また、上記フィードバック補正項ｑｉは、アイドル回転数が目標値よりも高ければ小さくされる。その結果、ＩＳＣ補正量ｑｃａｌ及びＩＳＣ開度補正量ｆ（ｑｃａｌ）が小さくなり、ＩＳＣ開度補正量ｆ（ｑｃａｌ）等に基づき算出される集約目標スロットル開度ＴＡｔが閉じ側の値へと変化する。「均質燃焼」であれ、「成層燃焼」であれ、上記のように集約目標スロットル開度ＴＡｔが閉じ側の値に変化することにより、機関出力トルクが減少してアイドル回転数が目標値に向けて下降する。
【０１０８】
「均質燃焼」時には、上記集約目標スロットル開度ＴＡｔが目標スロットル開度ＴＡｓｊに代入されるため、この目標スロットル開度ＴＡｓｊに基づきスロットル開度制御を行うことで予測吸気圧ＰＭＦＷＤ（吸入空気量）が減少する。その結果、予測吸気圧ＰＭＦＷＤ等に基づき決定される最終燃料噴射量Ｑfin が少なくなって機関出力トルクが減少し、アイドル回転数が目標値に向かって下降する。
【０１０９】
上記と異なり、「成層燃焼」時には、上記集約目標スロットル開度ＴＡｔ等に基づき仮想スロットル開度ＴＡｖが算出され、この仮想スロットル開度ＴＡｖ等に基づき仮想吸気圧ＰＭｖが算出される。そのため、ＩＳＣ開度補正量ｆ（ｑｃａｌ）により集約目標スロットル開度ＴＡｔが閉じ側の値に変化すると仮想吸気圧ＰＭｖが減少する。その結果、同仮想吸気圧ＰＭｖ等に応じて決定される最終燃料噴射量Ｑfin が少なくなり、機関出力トルクが減少してアイドル回転数が目標値に向かって下降する。
【０１１０】
上記式（８）中の計算用学習値ｑｇは、アイドル運転時に上記フィードバック補正項ｑｉが基準値「０」を含む所定範囲（本実施形態では「−γ＜ｑｉ＜γ」）内に収束するように、フィードバック補正項ｑｉに基づき基準値「０」を中心に増減する値である。
【０１１１】
即ち、計算用学習値ｑｇは、フィードバック補正項ｑｉが上記所定範囲よりも小さい側に外れていれば小さくされる。その結果、上記式（８）から明らかなように、アイドル回転数が目標値と一致する状態にＩＳＣ補正量ｑｃａｌを維持した条件のもとでは、計算用学習値ｑｇが小さくなるとフィードバック補正項ｑｉが大きくなり、同補正項ｑｉが上記所定範囲内へと収束するようになる。
【０１１２】
また、計算用学習値ｑｇは、フィードバック補正項ｑｉが上記所定範囲よりも大きい側に外れていれば大きくされる。その結果、上記式（８）から明らかなように、アイドル回転数が目標値と一致する状態にＩＳＣ補正量ｑｃａｌを維持した条件のもとでは、計算用ＩＳＣ補正量ｑｃａｌが大きくなるとフィードバック補正項ｑｉが小さくなり、同補正項ｑｉが上記所定範囲内へと収束するようになる。
【０１１３】
なお、上記式（８）中の水温補正項ｑｔｈｗは、エンジン１１の冷却水温ＴＨＷに応じて増減するものである。
このようにＩＳＣ補正量ｑｃａｌはフィードバック補正項ｑｉ、計算用学習値ｑｇ、及び水温補正項ｑｔｈｗ等から算出されるが、このうちの計算用学習値ｑｇは、燃焼方式毎に均質用学習値ｑｇｓｊと成層用学習値ｑｇｔｊとが用意されている。
【０１１４】
そして、「均質燃焼」時には均質用学習値ｑｇｓｊが計算用学習値ｑｇに代入され、同均質用学習値ｑｇｓｊの学習が行われる。即ち、「均質燃焼」でのアイドル運転時にフィードバック補正項ｑｉを所定範囲内に収束させるために上記計算用学習値ｑｇの増減が行われ、その増減後の計算用学習値ｑｇが均質用学習値ｑｇｓｊに代入される。この場合、フィードバック補正項ｑｉが所定範囲内に収束したときに均質用学習値ｑｇｓｊの学習が完了することとなる。
【０１１５】
エンジン１１の吸気系に経年変化等が生じて吸入空気量が適正値から外れると、「均質燃焼」でのアイドル運転時においては機関出力トルクが適正値からずれてアイドル回転数が目標値から外れることとなる。このアイドル回転数を目標値に近づけるべくフィードバック補正項ｑｉが増減され、更にフィードバック補正項ｑｉが所定範囲内に収束するよう計算用学習値ｑｇ（均質用学習値ｑｇｓｊ）が増減される。そして、フィードバック補正項ｑｉが所定範囲内に収束して均質用学習値ｑｇｓｊの学習が完了したとき、その均質用学習値ｑｇｓｊは上記吸入空気量の適正値からのずれ量に対応する値となる。
【０１１６】
従って、この均質用学習値ｑｇｓｊを以降のスロットル開度制御に反映することで、上記吸入空気量の適正値に対するずれを補償することが可能になる。均質用学習値ｑｇｓｊをスロットル開度制御に反映するには、同均質用学習値ｑｇｓｊを計算用学習値ｑｇとしてＩＳＣ補正量ｑｃａｌに反映する。このＩＳＣ補正量ｑｃａｌは、ＩＳＣ開度補正量ｆ（ｑｃａｌ）として集約目標スロットル開度ＴＡｔに反映される。「均質燃焼」時には、この集約目標スロットル開度ＴＡｔが代入される「均質燃焼」時の目標スロットル開度ＴＡｓｊに基づきスロットル開度制御が行われるため、このスロットル開度制御に上記均質用学習値ｑｇｓｊが反映される。
【０１１７】
また、「成層燃焼」時には成層用学習値ｑｇｔｊが計算用学習値ｑｇに代入され、同成層用学習値ｑｇｔｊの学習が行われる。即ち、「成層燃焼」でのアイドル運転時にフィードバック補正項ｑｉを所定範囲内に収束させるために上記計算用学習値ｑｇの増減が行われ、その増減後の計算用学習値ｑｇが成層用学習値ｑｇｔｊに代入される。この場合、フィードバック補正項ｑｉが所定範囲内に収束したときに成層用学習値ｑｇｔｊの学習が完了することとなる。
【０１１８】
エンジン１１の燃料系に経年変化等が生じて燃料噴射量が適正値から外れると、「成層燃焼」でのアイドル運転時においては機関出力トルクが適正値からずれてアイドル回転数が目標値からずれることとなる。このアイドル回転数を目標値に近づけるべくフィードバック補正項ｑｉが増減され、更にフィードバック補正項ｑｉが所定範囲内に収束するよう計算用学習値ｑｇ（成層用学習値ｑｇｔｊ）が増減される。そして、フィードバック補正項ｑｉが所定範囲内に収束して成層用学習値ｑｇｔｊの学習が完了したとき、その成層用学習値ｑｇｔｊは上記燃料噴射量の適正値からのずれ量に対応する値となる。
【０１１９】
従って、この成層用学習値ｑｇｔｊを燃料噴射制御に反映することで、上記燃料噴射量適正値に対するずれを補償することが可能になる。
ところで、成層用学習値ｑｇｔｊを燃料噴射制御に反映させるに際し、同成層用学習値ｑｇｔｊを計算用学習値ｑｇとしてＩＳＣ補正量ｑｃａｌに反映させると、このＩＳＣ補正量ｑｃａｌは、ＩＳＣ開度補正量ｆ（ｑｃａｌ）として集約目標スロットル開度ＴＡｔに反映されることとなり、これに伴い燃焼方式間での機関出力トルクに段差が生じてしまう。
【０１２０】
燃焼方式に応じて計算用学習値ｑｇとして均質用学習値ｑｇｓｊと成層用学習値ｑｇｔｊとを切り換えてＩＳＣ補正量ｑｃａｌに反映させた場合の均質用学習値ｑｇｓｊ及び成層用学習値ｑｇｔｊに対する集約目標スロットル開度ＴＡｔの関係を図１１に示す。なお、図１１（ａ）は、均質用学習値ｑｇｓｊ及び成層用学習値ｑｇｔｊを示すものであり、図１１（ｂ）の実線は、エンジン１１が定常状態であるときに同一機関運転状態のもとでの「均質燃焼」時において算出される集約目標スロットル開度ＴＡｔを示すものである。
【０１２１】
図１１（ａ）に示すように、均質用学習値ｑｇｓｊと成層用学習値ｑｇｔｊとは、吸入空気量と燃料噴射量といった異なる制御系に対応し、それぞれ吸入空気量の適正値からのずれ量、及び燃料噴射量の適正値からのずれ量に対応する値となることから、異なる値になることがある。即ち、吸入空気量の適正値に対するずれと燃料噴射量の適正値に対するずれとが互いに異なるものである場合、それらのずれを補償するための均質用学習値ｑｇｓｊ及び成層用学習値ｑｇｔｊも互いに異なる値になる。その結果、「均質燃焼」と「成層燃焼」とで集約目標スロットル開度ＴＡｔに反映されるＩＳＣ補正量ｑｃａｌも異なる値となり、図１１（ｂ）の実線に示すように、「均質燃焼」時において算出される集約目標スロットル開度ＴＡｔ及び「成層燃焼」時において算出される集約目標スロットル開度ＴＡｔも異なる値となる。
【０１２２】
ところで、集約目標スロットル開度ＴＡｔは、燃焼方式間で機関出力トルクを合致させるために、同一機関運転状態である条件下では図１１（ｂ）に破線で示されるように「均質燃焼」時と「成層燃焼」時とで等しい値とすべきである。
【０１２３】
そこで本実施形態では、集約目標スロットル開度ＴＡｔに対し、「成層燃焼」時に成層用学習値ｑｇｔｊを反映させる代わりに、「成層燃焼」時にも均質用学習値ｑｇｓｊを反映させる。これにより、「成層燃焼」時の集約目標スロットル開度ＴＡｔが図１１（ｂ）に破線で示す値となり、「均質燃焼」時において実線で示す値となる同「均質燃焼」時の集約目標スロットル開度ＴＡｔと合致する。この場合、集約目標スロットル開度ＴＡｔ等に基づき制御される「成層燃焼」時の燃料噴射量が均質用学習値ｑｇｓｊに対応したものとなり、このままでは燃料噴射量の適正値に対するずれを補償できない。そのため、上記式（３）に基づき「成層燃焼」時の最終燃料噴射量Ｑfin を補正することで、均質用学習値ｑｇｓｊと成層用学習値ｑｇｔｊとのずれ量に対応した燃料量を加味して「成層燃焼」時の燃料噴射量を制御する。
【０１２４】
このように集約目標スロットル開度ＴＡｔ及び燃料噴射量に対し、均質用学習値ｑｇｓｊ及び成層用学習値ｑｇｔｊを反映することで、吸入空気量や燃料噴射量の適正値に対するずれを補償しつつ、燃焼方式間で機関出力トルクを合わせることができるようになる。
【０１２５】
さて、説明を集約目標スロットル開度算出ルーチン（図１０）に戻す。
集約目標スロットル開度算出ルーチンにおいて、ＥＣＵ９２は、ステップＳ４０１の処理として、現在の燃焼方式に係わらず現在のアクセル踏込量ＡＣＣＰ及びエンジン回転数ＮＥに基づき「均質燃焼」時の基本スロットル開度ＴＡbse を算出する。その後、ステップＳ４０２に進む。
【０１２６】
ＥＣＵ９２は、ステップＳ４０２の処理として、計算用学習値ｑｇに均質用学習値ｑｇｓｊを代入した後、ステップＳ４０３に進む。
ＥＣＵ９２は、ステップＳ４０３の処理として上記式（８）によりＩＳＣ補正量ｑｃａｌを算出し、ステップＳ４０４の処理として上記式（７）によりＩＳＣ開度補正量ｆ（ｑｃａｌ）を算出する。続いて、ＥＣＵ９２は、ステップＳ４０５の処理として、上記式（６）により集約目標スロットル開度ＴＡｔを算出した後、この集約目標スロットル開度算出ルーチンを一旦終了する。
【０１２７】
次に、均質用学習値ｑｇｓｊ及び成層用学習値ｑｇｔｊの学習手順について図１２及び図１３を参照して説明する。図１２及び図１３は、均質用学習値ｑｇｓｊ及び成層用学習値ｑｇｔｊの学習を行うためのＩＳＣ学習ルーチンを示すフローチャートである。このＩＳＣ学習ルーチンは、ＥＣＵ９２を通じて例えば所定時間毎の時間割り込みにて実行される。
【０１２８】
ＩＳＣ学習ルーチンにおいては、ステップＳ５０１（図１２）の処理で学習値ｑｇｔｊ，ｑｇｓｊの増減（学習）が可能か否か、即ち学習条件が成立しているか否かが判断される。そして、ステップＳ５０２〜Ｓ５０４の処理で、燃焼方式に応じて計算用学習値ｑｇに成層用学習値ｑｇｔｊ、若しくは均質用学習値ｑｇｓｊが代入される。更に、Ｓ５０５〜Ｓ５０８の処理で上記計算用学習値ｑｇがフィードバック補正項ｑｉを所定範囲（本実施形態では「−γ＜ｑｉ＜γ」）内に収束させるべく増減される。また、ステップＳ５０９〜Ｓ５１７（図１３）では、フィードバック補正項ｑｉを所定範囲に収束させるための計算用学習値ｑｇの操作を行った後の同学習値ｑｇが、燃焼方式に応じて成層用学習値ｑｇｔｊ、若しくは均質用学習値ｑｇｓｊに代入される。これにより、燃焼方式毎にそれぞれ別々にＩＳＣ学習値を持つことが可能になる。
【０１２９】
さて、ＩＳＣ学習ルーチンにおいて、ＥＣＵ９２は、ステップＳ５０１の処理として、ＩＳＣ学習条件が成立しているか否か、即ち例えば以下に示す各種条件が全て成立しているか否かを判断する。
【０１３０】
・ＩＳＣフィードバック制御中であること
・エンジン回転数ＮＥの変動が小さいこと
・その他の外乱がないこと
そして、上記各条件の内のいずれか一つでも満たされていなければ、ＥＣＵ９２は、このＩＳＣ学習ルーチンを一旦終了する。上記各条件が全て満たされていれば、ステップＳ５０２に進む。
【０１３１】
ＥＣＵ９２は、ステップＳ５０２の処理として、成層燃焼運転中であるか否かを判断する。そして、成層燃焼運転中であればステップＳ５０３の処理として計算用学習値ｑｇに成層用学習値ｑｇｔｊを代入し、成層燃焼運転中でなく均質燃焼運転中であればステップＳ５０４の処理として計算用学習値ｑｇに均質用学習値ｑｇｓｊを代入する。こうして計算用学習値ｑｇを燃焼方式に対応したものとした後、ステップＳ５０５以降の処理を実行する。
【０１３２】
ＥＣＵ９２は、ステップＳ５０５の処理として、フィードバック補正項ｑｉが所定値γよりも大きいか否かを判断する。そして、「ｑｉ＞γ」であってフィードバック補正項ｑｉが所定範囲「−γ〜γ」に対し大きい側に外れている旨判断されると、ステップＳ５０６の処理として計算用学習値ｑｇに所定値ａを加算する。こうして計算用学習値ｑｇを大きくすると、アイドル回転数が目標値に一致する状態にＩＳＣ補正量ｑｃａｌを維持した条件のもとでは、フィードバック補正項ｑｉが所定範囲「−γ〜γ」内に向かって小さくなる。
【０１３３】
また、上記ステップＳ５０５の処理において「ｑｉ＞γ」でない旨判断されると、ステップＳ５０７に進んでフィードバック補正項ｑｉが所定値−γ未満であるか否かを判断する。そして、「ｑｉ＜−γ」であってフィードバック補正項ｑｉが所定範囲「−γ〜γ」に対し小さい側に外れている旨判断されると、ステップＳ５０８の処理として計算用学習値ｑｇから所定値ａを減算する。こうして計算用学習値ｑｇを小さくすると、アイドル回転数が目標値に一致する状態にＩＳＣ補正量ｑｃａｌを維持した条件のもとでは、フィードバック補正項ｑｉが所定範囲「−γ〜γ」内に向かって大きくなる。
【０１３４】
このようにフィードバック補正項ｑｉを所定範囲「−γ〜γ」内に収束させるための計算用学習値ｑｇの増減を行った後、ステップＳ５０９（図１３）に進む。ＥＣＵ９２は、ステップＳ５０９の処理として、成層燃焼運転中か否かを判断する。そして、成層燃焼運転中であればステップＳ５１０の処理として成層用学習値ｑｇｔｊに現在の計算用学習値ｑｇを代入し、成層燃焼運転中でなく均質燃焼運転中であればステップＳ５１４の処理として均質用学習値ｑｇｓｊに現在の計算用学習値ｑｇを代入する。このように計算用学習値ｑｇの成層用学習値ｑｇｔｊ、若しくは均質用学習値ｑｇｓｊへの代入を行うことで、燃焼方式毎にそれぞれ別々にＩＳＣ学習値を持つことが可能になる。
【０１３５】
なお、成層用学習値ｑｇｔｊ、及び均質用学習値ｑｇｓｊはバックアップＲＡＭ９６に記憶される。そして、「成層燃焼」時にフィードバック補正項ｑｉが上記所定範囲内に収束したときに成層用学習値ｑｇｔｊの学習が完了し、「均質燃焼」時にフィードバック補正項ｑｉが上記所定範囲内に収束したときに均質用学習値ｑｇｓｊの学習が完了する。
【０１３６】
ＩＳＣ学習ルーチンにおいて、上記ステップＳ５１０の処理が行われた後には、ステップＳ５１１〜Ｓ５１３の処理が行われる。これらの処理は、成層用学習値ｑｇｔｊの学習が行われるとき、この成層用学習値ｑｇｔｊを所定条件のもとで均質用学習値ｑｇｓｊに反映し、均質用学習値ｑｇｓｊの学習を早期に完了させるためのものである。
【０１３７】
ＥＣＵ９２は、ステップＳ５１１の処理として、均質用学習値ｑｇｓｊの学習を早期に行う必要があるか否かの判断基準であるスピードアップ条件が成立しているか否かを判断する。こうした条件としては、
・バッテリの交換がなされたとき
・均質用学習値ｑｇｓｊや成層用学習値ｑｇｔｊが過度に基準値「０」から離れた値になるとき
等々があげられる。
【０１３８】
なお、上記のようなスピードアップ条件がいずれか一つでも成立したときは、均質用学習値ｑｇｓｊ及び成層用学習値ｑｇｔｊが初期値に戻されているときである。この初期値は、均質用学習値ｑｇｓｊ及び成層用学習値ｑｇｔｊの基準値「０」よりも大きい値に設定されている。そのため、上記条件の成立時には均質用学習値ｑｇｓｊ及び成層用学習値ｑｇｔｊの学習を素早く行い、早期に均質用学習値ｑｇｓｊ及び成層用学習値ｑｇｔｊを基準値「０」側へ向かって小さくする必要がある。
【０１３９】
ステップＳ５１１の処理において、スピードアップ条件が成立していない旨判断されると当該ＩＳＣ学習ルーチンを一旦終了し、同条件が成立している旨判断されるとステップＳ５１２に進む。ＥＣＵ９２は、ステップＳ５１２の処理として、このとき計算用学習値ｑｇ（成層用学習値ｑｇｔｊ）に所定値αを加算した値（「ｑｇ＋α」）に対し、現在の均質用学習値ｑｇｓｊが大きいか否かを判断する。
【０１４０】
そして、「ｑｇｓｊ＞ｑｇ＋α」でなく均質用学習値ｑｇｓｊが基準値「０」に対して過度に大きい値でない旨判断されると当該ＩＳＣ学習ルーチンを一旦終了し、「ｑｇｓｊ＞ｑｇ＋α」であって均質用学習値ｑｇｓｊが基準値「０」に対して過度に大きい値である旨判断されるとステップＳ５１３に進む。ＥＣＵ９２は、ステップＳ５１３の処理として、現在の均質用学習値ｑｇｓｊに上記「ｑｇ＋α」という値を代入した後、このＩＳＣ学習ルーチンを一旦終了する。このように成層用学習値ｑｇｔｊを均質用学習値ｑｇｓｊに反映させることで、同均質用学習値ｑｇｓｊの学習を早期に完了することができる。
【０１４１】
ＩＳＣ学習ルーチンにおいて、上記ステップＳ５１４の処理が行われた後には、ステップＳ５１５〜Ｓ５１７の処理が行われる。これらの処理は、均質用学習値ｑｇｓｊの学習が行われるとき、この均質用学習値ｑｇｓｊを所定条件のもとで成層用学習値ｑｇｔｊに反映し、成層用学習値ｑｇｔｊの学習を早期に完了させるためのものである。
【０１４２】
ＥＣＵ９２は、ステップＳ５１５の処理として、均質用学習値ｑｇｓｊの学習を早期に行う必要があるか否かの判断基準であるスピードアップ条件が成立しているか否かを判断する。こうした条件としては、ステップＳ５１１の条件と同じものがあげられる。
【０１４３】
ステップＳ５１５の処理において、スピードアップ条件が成立していない旨判断されると当該ＩＳＣ学習ルーチンを一旦終了し、同条件が成立している旨判断されるとステップＳ５１６に進む。ＥＣＵ９２は、ステップＳ５１６の処理として、このとき計算用学習値ｑｇ（均質用学習値ｑｇｓｊ）に所定値αを加算した値（「ｑｇ＋α」）に対し、現在の成層用学習値ｑｇｔｊが大きいか否かを判断する。
【０１４４】
そして、「ｑｇｔｊ＞ｑｇ＋α」でなく成層用学習値ｑｇｔｊが基準値「０」に対して過度に大きい値でない旨判断されると当該ＩＳＣ学習ルーチンを一旦終了し、「ｑｇｔｊ＞ｑｇ＋α」であって成層用学習値ｑｇｔｊが基準値「０」に対して過度に大きい値である旨判断されるとステップＳ５１７に進む。ＥＣＵ９２は、ステップＳ５１７の処理として、現在の成層用学習値ｑｇｔｊに上記「ｑｇ＋α」という値を代入した後、このＩＳＣ学習ルーチンを一旦終了する。このように均質用学習値ｑｇｓｊを成層用学習値ｑｇｔｊに反映させることで、同成層用学習値ｑｇｔｊの学習を早期に完了することができる。
【０１４５】
上記ＩＳＣ学習ルーチンにより学習される成層用学習値ｑｇｔｊ、及び均質用学習値ｑｇｓｊは、最終噴射量算出ルーチンにおけるステップＳ２０６（図５）の処理での上記式（３）に用いられる。
【０１４６】
即ち、式（３）において、最終燃料噴射量Ｑfin に加算される項である「｛（ｑｇｔｊ／ｑｇｓｊ）−１｝＊Ｃ＊（６００／ＮＥ）」にて、成層用学習値ｑｇｔｊ、及び均質用学習値ｑｇｓｊが用いられる。最終燃料噴射量Ｑfin は集約目標スロットル開度ＴＡｔ等に基づき算出されるが、この集約目標スロットル開度ＴＡｔの算出に用いられる計算用学習値ｑｇは、均質用学習値ｑｇｓｊに固定される。
【０１４７】
従って、成層燃焼運転時には「成層燃焼」中であるにもかかわらず、上記集約目標スロットル開度ＴＡｔ等に基づき算出される最終燃料噴射量Ｑfin が均質用学習値ｑｇｓｊに対応したものとなる。そのため、「成層燃焼」時には、上記「｛（ｑｇｔｊ／ｑｇｓｊ）−１｝＊Ｃ＊（６００／ＮＥ）」という項を最終燃料噴射量Ｑfin に加算することで、同最終燃料噴射量Ｑfin を適切なものへと補正する。即ち、上記「｛（ｑｇｔｊ／ｑｇｓｊ）−１｝＊Ｃ＊（６００／ＮＥ）」という項は、均質用学習値ｑｇｓｊと成層用学習値ｑｇｔｊとのずれ量に対応した燃料量の分であって、この項を加味して最終燃料噴射量Ｑfin を算出することにより、燃料噴射量制御に成層用学習値ｑｇｔｊが加味されて燃料噴射量の適正値に対するずれが補償される。
【０１４８】
以上詳述した処理が行われる本実施形態によれば、以下に示す効果が得られるようになる。
（１）「均質燃焼」時には、吸入空気量の適正値からのずれ量に対応する値として学習された均質用学習値ｑｇｓｊを加味して、「均質燃焼」時におけるスロットル開度制御の制御目標値である目標スロットル開度ＴＡｓｊ（集約目標スロットル開度ＴＡｔ）が算出される。そのため、上記目標スロットル開度ＴＡｓｊ等に基づきスロットル開度制御を行うことで、吸入空気量の適正値に対するずれを的確に補償することができる。
【０１４９】
また、「成層燃焼」時において、均質用学習値ｑｇｓｊを加味して「仮想均質燃焼」時の仮想スロットル開度ＴＡｖ及び仮想吸気圧ＰＭｖ（仮想吸入空気量）が算出される。そして、仮想吸気圧ＰＭｖ等に基づき基本燃料噴射量Ｑbse が算出され、この基本燃料噴射量Ｑbse 等に基づき成層用学習値ｑｇｔｊを加味して燃料噴射量の制御目標値である最終燃料噴射量Ｑfin が算出される。
【０１５０】
上記のように均質用学習値ｑｇｓｊの加味された仮想吸気圧ＰＭｖに基づき基本燃料噴射量Ｑbse を算出することで、「成層燃焼」時における燃料噴射量の制御目標値である上記最終燃料噴射量Ｑfin がこのとき「均質燃焼」を実行した場合に得られる機関出力トルクに関連付けられる。そして、この最終燃料噴射量Ｑfin に基づき「成層燃焼」時の燃料噴射量が制御されるため、「成層燃焼」時の機関出力トルクが、このとき「均質燃焼」を行った場合での機関出力トルクに的確に合致するようになる。
【０１５１】
更に、「成層燃焼」時には、最終燃料噴射量Ｑfin が基本燃料噴射量Ｑbse 等に基づき成層用学習値ｑｇｔｊを加味して算出されることで、同最終燃料噴射量Ｑfin に成層用学習値ｑｇｔｊが反映され、これにより燃料噴射量の適正値に対するずれを的確に補償することができる。
【０１５２】
従って、アイドル回転数を制御するための制御量である吸入空気量及び燃料噴射量の適正値に対するずれを補償しつつ、燃焼方式間で機関出力トルクを合わせることができる。こうした燃焼方式間での機関出力トルクが食い違う場合、これによるトルクショックを感じやすいのは自動車の減速中での燃焼方式の切換時などであるが、上記のように燃焼方式間で機関出力トルクを合わせることにより、自動車の減速中での燃焼方式切換時にトルクショックを感じることがなくなる。
【０１５３】
（２）「成層燃焼」時には、均質用学習値ｑｇｓｊが加味された仮想吸気圧ＰＭｖに基づき最終燃料噴射量Ｑfin （基本燃料噴射量Ｑbse ）が算出されるが、最終的には上記式（３）において「ｑｇｔｊ／ｑｇｓｊ」という項により、最終燃料噴射量Ｑfin が成層用学習値ｑｇｔｊと均質用学習値ｑｇｓｊとの両方を加味した値となる。そのため、上記成層用学習値ｑｇｔｊにより燃料噴射量の適正値に対するずれを補償する際、これが上記均質用学習値ｑｇｓｊを加味した仮想吸気圧ＰＭｖに起因して過剰なものになるのを抑制することができる。
【０１５４】
（３）均質用学習値ｑｇｓｊ及び成層用学習値ｑｇｔｊはアイドル運転中に学習されるため、エンジン回転数ＮＥが高くなるにつれて、これら学習値ｑｇｓｊ，ｑｇｔｊにより「成層燃焼」時の燃料噴射量の適正値に対するずれを補償する際、これが過剰になる。しかし、「成層燃焼」時には最終燃料噴射量Ｑfin が上記式（３）において「６００／ＮＥ」という項によりエンジン回転数ＮＥを加味した値となり、上記学習値ｑｇｓｊ，ｑｇｔｊにより燃料噴射量の適正値に対するずれを補償する際、これが過剰になるのを抑制することができる。
【０１５５】
（４）「成層燃焼」時に成層用学習値ｑｇｔｊが学習されるとき、スピードアップ条件が成立していれば、そのときの成層用学習値ｑｇｔｊ（計算用学習値ｑｇ）が均質用学習値ｑｇｓｊに反映される。また、「均質燃焼」時に均質用学習値ｑｇｓｊが学習されるとき、スピードアップ条件が成立していれば、そのときの均質用学習値ｑｇｓｊ（計算用学習値ｑｇ）が成層用学習値ｑｇｔｊに反映される。このように一方の学習値を他方の学習値に反映させることで、成層用学習値ｑｇｔｊ及び均質用学習値ｑｇｓｊが初期値に戻されたときなどに、それら学習値ｑｇｔｊ，ｑｇｓｊの学習を早期に完了することができる。
【０１５６】
なお、本実施形態は、例えば以下のように変更することもできる。
・本実施形態では、均質用学習値ｑｇｓｊと成層用学習値ｑｇｔｊとをスピードアップ条件の成立時に互いに反映し合い、それら学習値ｑｇｓｊ，ｑｇｔｊの学習を早期に完了することを可能としたが、必ずしも学習値ｑｇｓｊ，ｑｇｔｊを互いに反映し合う必要はない。均質用学習値ｑｇｓｊと成層用学習値ｑｇｔｊとを互いに反映しないようにするならば、ＩＳＣ学習ルーチン（図１３）におけるステップＳ５１１〜Ｓ５１３の処理やステップＳ５１５〜Ｓ５１７の処理を省略することができる。
【図面の簡単な説明】
【図１】本実施形態の制御装置が適用されるエンジン全体を示す断面図。
【図２】同制御装置の電気的構成を示すブロック図。
【図３】目標スロットル開度ＴＡｓｊの変化に対する位相進み補償後スロットル開度ＴＡｈ、及び実際のスロットル開度ＴＡｒの推移を示すタイムチャート。
【図４】最終燃料噴射量Ｑfin の算出手順を示すフローチャート。
【図５】吸気温補正係数Ｋtha を算出する際に参照されるマップ。
【図６】予測吸気圧ＰＭＦＷＤ及び仮想吸気圧ＰＭｖの算出手順を示すフローチャート。
【図７】予測吸気圧ＰＭＦＷＤ及び仮想吸気圧ＰＭｖの算出手順を示すフローチャート。
【図８】大気圧補正係数Ｋpa１を算出する際に参照されるマップ。
【図９】補正後吸気圧ＰＭｈ、徐変値ＰＭＳＭ、フィルタ出力ＰＭＳＭ１Ｓi 、及び実際の吸気圧ＰＭｒの推移を示すタイムチャート。
【図１０】集約目標スロットル開度ＴＡｔの算出手順を示すフローチャート。
【図１１】均質用学習値ｑｇｓｊ及び成層用学習値ｑｇｔｊに対する集約目標スロットル開度ＴＡｔの関係を示す図。
【図１２】均質用学習値ｑｇｓｊ及び成層用学習値ｑｇｔｊの学習手順を示すフローチャート。
【図１３】均質用学習値ｑｇｓｊ及び成層用学習値ｑｇｔｊの学習手順を示すフローチャート。
【符号の説明】
１１…エンジン、１１ｂ…水温センサ、１４ｃ…クランクポジションセンサ、２１ｂ…カムポジションセンサ、２３…スロットルバルブ、２４…スロットル用モータ、２５…アクセルペダル、２６…アクセルポジションセンサ、３６…バキュームセンサ、３７…吸気温センサ、４０…燃料噴射弁、４４…スロットルポジションセンサ、９２…電子制御ユニット（ＥＣＵ）。[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a control device for an internal combustion engine in which a combustion method is switched according to an engine operating state.
[0002]
[Prior art]
In recent years, in an internal combustion engine for automobiles, an internal combustion engine of a type in which a combustion method is switched according to an engine operating state has been proposed and put into practical use in order to achieve both improvement in fuel consumption and sufficient engine output. Has been. Examples of this type of internal combustion engine include those described in JP-A-10-169490.
[0003]
The internal combustion engine described in this publication burns a homogeneous air-fuel mixture in which the fuel is evenly mixed with air in the state of the stoichiometric air-fuel ratio during predetermined engine operation such as at high rotation and high load where high output is required. “Stratified combustion” is performed to obtain sufficient engine output.
[0004]
In addition, when the engine speed is low and the load is not very high, the fuel-air mixture with a high fuel concentration is distributed around the spark plug so that the average air-fuel ratio of the whole air-fuel mixture is much leaner than the stoichiometric air-fuel ratio. "Strategic combustion" is performed to obtain ignition of the qi. When performing such “stratified combustion”, the throttle valve of the internal combustion engine is controlled to open more than when performing “homogenous combustion” in order to make the air-fuel ratio of the air-fuel mixture leaner than the stoichiometric air-fuel ratio. Engine pumping loss is reduced and fuel consumption is improved.
[0005]
By switching the combustion method of the internal combustion engine between “homogeneous combustion” and “stratified combustion” according to the engine operating state as described above, fuel efficiency can be improved and sufficient engine output can be obtained. become.
[0006]
By the way, in an internal combustion engine, idle speed control (ISC) for controlling the idling speed is executed. In an internal combustion engine in which the combustion method is switched, a control system for controlling the operation of the engine is used. Of these, the idling speed is controlled by adjusting the engine output torque by controlling a different control system for each combustion method. In the internal combustion engine described in the above publication, when the idling speed is “homogeneous combustion”, it is adjusted by correcting the throttle opening, and when it is “stratified combustion”, it is adjusted by correcting the fuel injection amount.
[0007]
That is, at the time of “homogeneous combustion”, the throttle opening is corrected based on a feedback correction term that increases or decreases around a predetermined reference value (for example, “0”) according to the actual idle speed. When the amount of air sucked into the combustion chamber is adjusted by correcting the throttle opening, the fuel injection amount determined based on the intake air amount is changed, and the engine output torque is adjusted. Approaches the target value.
[0008]
On the other hand, at the time of “stratified combustion”, the fuel injection amount is directly corrected based on the feedback correction term and the engine output torque is adjusted, whereby the idle speed is brought close to the target value. The reason why the engine speed is controlled by the fuel injection amount at the time of “stratified combustion” is not determined uniquely based on the intake air amount at the time of “stratified combustion”. This is because the engine output torque is unlikely to change based on the adjustment of the intake air amount by the correction, and the idle speed is difficult to control.
[0009]
In an internal combustion engine, the intake air amount and the fuel injection amount deviate from appropriate values due to aging of the intake system and the fuel system, etc., and the engine output torque becomes inappropriate and the idle speed is reduced. May deviate from the target value. In order to suppress the deviation of the intake air amount and the fuel injection amount from the appropriate values, so-called ISC learning control is performed. With this learning control, the engine output torque is appropriate to suppress the deviation of the idle speed from the target value. Adjusted to the value.
[0010]
That is, during idle operation in “homogeneous combustion”, a learning value for homogeneity used for correcting the throttle opening (intake air amount) is set so that the feedback correction term converges within a predetermined range including the reference value. The Then, the learning of the homogenous learning value is completed when the feedback correction term converges within the predetermined range. The homogeneity learning value after the completion of learning is a value corresponding to the amount of change in the throttle opening that can compensate for the deviation of the intake air amount from the appropriate value. Therefore, by correcting the throttle opening at the time of “homogeneous combustion” with the learning value for homogeneity after the completion of the learning and reflecting the learning value for homogeneity in the subsequent throttle opening, there is a deviation from the appropriate value of the intake air amount. Compensated. As a result, the engine output torque becomes an appropriate value, and the deviation of the idle speed from the target value is suppressed, and the combustion state in the internal combustion engine deteriorates during idle operation or the like due to the deviation of the intake air amount from the appropriate value. Can be suppressed.
[0011]
Further, during the idling operation in “stratified combustion”, the learning value for stratification used for correcting the fuel injection amount is set so that the feedback correction term converges to a value within a predetermined range including the reference value. The learning of the stratified learning value is completed when the feedback correction term converges within the predetermined range. The learning value for stratification after the learning is completed becomes a value corresponding to the change amount of the fuel injection amount that can compensate for the deviation of the fuel injection amount from the appropriate value. Therefore, by correcting the fuel injection amount at the time of “stratified combustion” with the learning value for stratification after the completion of the learning and reflecting the learning value for stratification in the subsequent fuel injection amount, the deviation of the fuel injection amount from the appropriate value is changed. Compensated. As a result, the engine output torque becomes an appropriate value, and the deviation of the idling engine speed from the target value is suppressed, and the combustion state in the internal combustion engine deteriorates during idle operation or the like due to the deviation of the fuel injection amount from the appropriate value. Can be suppressed.
[0012]
[Problems to be solved by the invention]
By learning the learning value for homogeneity and the learning value for stratification as described above, it is possible to compensate for the difference in intake air amount during "homogeneous combustion" and to compensate for the difference in fuel injection amount during "stratified combustion" However, it is difficult to match the engine output torque between the combustion methods. Usually, in order to match the engine output torque between the combustion systems, it is necessary to match the engine output torque during the homogeneous combustion operation and the engine output torque during the stratified combustion operation under the same engine operation state.
[0013]
However, if the learning value for homogeneity and the learning value for stratification are switched according to the combustion method, this will lead to the generation of a step in the engine output torque when the combustion method is switched. This is because the homogeneity learning value and the stratification learning value correspond to different control systems such as the intake system and the fuel system, and compensate for each deviation of the control amount of the control system from the appropriate value. It is.
[0014]
That is, when the deviation from the appropriate value of the intake air amount is different from the deviation from the appropriate value of the fuel injection amount, the learning value for stratification and the learning value for homogeneity for compensating for the deviation are also different. Further, the adjustment amount of the engine output torque accompanying the correction by the learning value for homogeneity and the learning value for stratification also becomes different from each other, and a step of the engine output torque is generated when the combustion method is switched.
[0015]
The present invention has been made in view of such circumstances, and its purpose is to accurately compensate for deviations in the control values of various control systems that control the idle speed, such as the intake air amount and the fuel injection amount, from appropriate values. However, an object of the present invention is to provide a control device for an internal combustion engine that can match engine output torque between combustion systems.
[0016]
[Means for Solving the Problems]
In the following, means for achieving the above object and its effects are described.
In order to achieve the above object, according to the first aspect of the present invention, the combustion system is switched between homogeneous combustion and stratified combustion according to the engine operating state, and the intake air amount control is performed when the idle speed is adjusted during homogeneous combustion. Applied to an internal combustion engine that is controlled by fuel injection amount control during stratified combustion operation, and learns a value corresponding to a deviation from an appropriate value of the intake air amount during idle operation in homogeneous combustion as a learning value for homogeneity, In an internal combustion engine control device that learns a value corresponding to a deviation from an appropriate value of the fuel injection amount during idle operation in stratified combustion as a stratified learning value, the same intake is performed based on the control target value of the intake air amount during homogeneous combustion operation The air amount is controlled, and the control target value of the intake air amount is calculated by taking the homogeneity learning value into account, thereby obtaining an appropriate value of the intake air amount. Intake air amount control means for reflecting the learning value for homogeneity in the control of the intake air amount to compensate for the deviation, and the intake air amount when it is assumed that the homogeneous combustion operation is executed in the engine operation state during the stratified combustion operation Virtual intake air amount calculating means for calculating the virtual intake air amount in consideration of the learning value for homogeneity, and controlling the fuel injection amount based on a control target value of the fuel injection amount during stratified combustion operation, By calculating the control target value of the fuel injection amount in consideration of the learning value for stratification based on the virtual intake air amount, the fuel injection amount can be controlled to compensate for the deviation of the fuel injection amount from the appropriate value. Fuel injection amount control means reflecting the learning value for stratification.
[0017]
According to this configuration, the homogeneity learning value is reflected in the control of the intake air amount during the homogeneous combustion operation, thereby compensating for the deviation of the intake air amount from the appropriate value. In addition, during stratified combustion operation, a control target value for the fuel injection amount is calculated based on the virtual intake air amount calculated by taking the homogeneity learning value into consideration, and this control target value is obtained when homogeneous combustion is executed at this time. It will be related to the engine output torque. Since the fuel injection amount is controlled based on this control target value, the engine output torque during the stratified charge combustion operation properly matches the engine output torque during the homogeneous combustion operation. Further, the control of the fuel injection amount during the stratified combustion operation reflects not only the virtual intake air amount but also the stratified learning value, thereby compensating for the deviation of the fuel injection amount from the appropriate value. Accordingly, the engine output torque can be matched between the combustion methods while compensating for the deviation of the intake air amount and the fuel injection amount, which are control amounts for controlling the idle speed, from the appropriate values.
[0018]
According to a second aspect of the present invention, in the first aspect of the invention, when the fuel injection amount control means calculates the control target value of the fuel injection amount based on the virtual intake air amount, the fuel injection amount control means adds to the learning value for stratification. Thus, the learning value for homogeneity was taken into account.
[0019]
For the fuel injection amount control during the stratified combustion operation, a virtual intake air amount is used in consideration of the learning value for homogeneity. When calculating the control target value of the fuel injection amount during the stratified charge combustion operation based on the virtual intake air amount, the homogeneity learning value is added to the stratified learning value in addition to the stratified learning value. When compensating for the deviation of the fuel injection amount from the appropriate value, it can be suppressed that this becomes excessive due to the virtual intake air amount in consideration of the homogeneity learning value.
[0020]
According to a third aspect of the invention, in the first or second aspect of the invention, when the fuel injection amount control means calculates the control target value of the fuel injection amount based on the virtual intake air amount, the engine speed is further set. To be added.
[0021]
Since the learning value for homogeneity and the learning value for stratification are learned during idle operation, as the engine speed increases, when the deviation from the appropriate value of the fuel injection amount during stratified combustion operation is compensated by the same learning value, Become excessive. However, when the control target value of the fuel injection amount is calculated based on the virtual intake air amount, according to the same configuration in which the engine speed is further taken into account, the learning value can be obtained from the appropriate value of the fuel injection amount during stratified combustion operation. When compensating for the deviation, it can be suppressed that this becomes excessive.
[0022]
According to a fourth aspect of the present invention, when the virtual intake air amount calculating means is assumed to execute the homogeneous combustion operation in the engine operation state during the stratified combustion operation in the first aspect of the present invention. The virtual throttle opening that is the opening of the throttle valve of the engine is calculated in consideration of the homogeneity learning value, and the virtual intake air amount is calculated from the virtual throttle opening.
[0023]
According to this configuration, since the learning value for homogeneity is taken into account when calculating the virtual throttle opening used for calculating the virtual intake air amount, the virtual intake air amount can be set to an appropriate value. By controlling the fuel injection amount during the stratified combustion operation based on the control target value of the fuel injection amount calculated based on this virtual intake air amount, the engine output torque during the stratified combustion operation is changed to the engine output torque during the homogeneous combustion operation. More appropriately.
[0028]
Claim 5 In the described invention, claims 1 to 4 In the invention described in any of the above, Either homogeneous combustion or stratified combustion When the learning value corresponding to the combustion method is learned during idling operation of this combustion method, the learning value is set under a predetermined condition. Of homogeneous combustion and stratified combustion The other Corresponding to the combustion method The learning value reflecting means for reflecting the learning value is further provided.
[0029]
According to this configuration, since the learning values corresponding to the respective combustion methods reflect each other, learning of these learning values can be completed early.
[0030]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, an embodiment in which the present invention is applied to an in-line four-cylinder direct injection gasoline engine for automobiles will be described with reference to FIGS.
[0031]
As shown in FIG. 1, the engine 11 includes a total of four pistons 12 (only one is shown in FIG. 1) provided for reciprocating movement in the cylinder block 11a for each cylinder. Each piston 12 has a recess 12 a necessary for executing “stratified combustion”, which will be described later, at its head, and is connected to a crankshaft 14, which is an output shaft, via a connecting rod 13. The reciprocating movement of the piston 12 is converted into rotation of the crankshaft 14 by the connecting rod 13.
[0032]
A signal rotor 14 a is attached to the crankshaft 14. A plurality of protrusions 14b are provided on the outer periphery of the signal rotor 14a at equal angles with the axis of the crankshaft 14 as the center. A crank position sensor 14c is provided on the side of the signal rotor 14a. Then, when the crankshaft 14 rotates and each projection 14b of the signal rotor 14a sequentially passes the side of the crank position sensor 14c, the sensor 14c detects a pulse shape corresponding to the passage of each projection 14b. A signal is output.
[0033]
The cylinder block 11 a is provided with a water temperature sensor 11 b for detecting the cooling water temperature of the engine 11. A cylinder head 15 is provided at the upper end of the cylinder block 11 a, and a combustion chamber 16 is provided between the cylinder head 15 and the piston 12. An intake passage 32 and an exhaust passage 33 are connected to the combustion chamber 16. The combustion chamber 16 and the intake passage 32 are communicated and blocked by the opening / closing operation of the intake valve 19, and the combustion chamber 16 and the exhaust passage 33 are communicated and blocked by the opening / closing operation of the exhaust valve 20.
[0034]
On the other hand, an intake camshaft 21 and an exhaust camshaft 22 for opening and closing the intake valve 19 and the exhaust valve 20 are rotatably supported by the cylinder head 15. The intake and exhaust camshafts 21 and 22 are connected to the crankshaft 14 via timing belts and gears (both not shown), and the rotation of the crankshaft 14 is transmitted by the belts and gears. . When the intake camshaft 21 rotates, the intake valve 19 opens and closes, and when the exhaust camshaft 22 rotates, the exhaust valve 20 opens and closes.
[0035]
In the cylinder head 15, a cam position sensor 21 b that detects a protrusion 21 a provided on the outer peripheral surface of the shaft 21 and outputs a detection signal is provided on the side of the intake camshaft 21. When the intake camshaft 21 rotates, the projection 21a of the shaft 21 passes the side of the cam position sensor 21b. In this state, detection signals are output from the cam position sensor 21b at predetermined intervals corresponding to the passage of the protrusion 21a.
[0036]
A throttle valve 23 for adjusting the intake air amount of the engine 11 is provided in the upstream portion of the intake passage 32. The opening degree of the throttle valve 23 is adjusted by drivingly controlling the throttle motor 24 based on the depression amount (accelerator depression amount) of the accelerator pedal 25 detected by the accelerator position sensor 26. By adjusting the opening degree of the throttle valve 23, the intake air amount of the engine 11 is adjusted. The opening of the throttle valve 23 is detected by a throttle position sensor 44.
[0037]
In addition, a vacuum sensor 36 for detecting the pressure in the passage 32 is provided in a portion of the intake passage 32 located on the downstream side of the throttle valve 23. The vacuum sensor 36 outputs a detection signal corresponding to the detected pressure in the intake passage 32. Further, an intake air temperature sensor 37 for detecting the temperature of the air (intake air) passing through the passage 32 is provided in a portion of the intake passage 32 positioned on the upstream side of the throttle valve 23. The intake air temperature sensor 37 outputs a detection signal corresponding to the detected intake air temperature (intake air temperature).
[0038]
Further, the cylinder head 15 includes a fuel injection valve 40 that injects and supplies fuel into the combustion chamber 16, and an ignition plug 41 that ignites an air-fuel mixture composed of fuel and air filled in the combustion chamber 16. Is provided. When the fuel is injected from the fuel injection valve 40 into the combustion chamber 16, the fuel is mixed with the air sucked into the combustion chamber 16 via the intake passage 32, and the air and fuel are mixed in the combustion chamber 16. An air-fuel mixture is formed. Further, the air-fuel mixture in the combustion chamber 16 is ignited by the spark plug 41 and combusted, and the air-fuel mixture after combustion is sent to the exhaust passage 33 as exhaust gas.
[0039]
Next, the electrical configuration of the control device for the engine 11 in the present embodiment will be described with reference to FIG.
The control device includes an electronic control unit (hereinafter referred to as “ECU”) 92 for controlling the operating state of the engine 11 such as fuel injection amount control, fuel injection timing control, and throttle opening control. The ECU 92 is configured as a calculation logic operation circuit including a ROM 93, a CPU 94, a RAM 95, a backup RAM 96, and the like.
[0040]
Here, the ROM 93 is a memory in which various control programs and maps to be referred to when executing these various control programs are stored. The CPU 94 performs arithmetic processing based on the various control programs and maps stored in the ROM 93. Execute. The RAM 95 is a memory for temporarily storing calculation results in the CPU 94 and data input from each sensor. The backup RAM 96 is a non-volatile memory for storing the stored data when the engine 11 is stopped. is there. The ROM 93, CPU 94, RAM 95, and backup RAM 96 are connected to each other via a bus 97 and are connected to an external input circuit 98 and an external output circuit 99.
[0041]
A water temperature sensor 11b, a crank position sensor 14c, a cam position sensor 21b, an accelerator position sensor 26, a vacuum sensor 36, an intake air temperature sensor 37, a throttle position sensor 44, and the like are connected to the external input circuit 98. On the other hand, the throttle motor 24, the fuel injection valve 40, and the like are connected to the external output circuit 99.
[0042]
The ECU 92 configured as described above switches the combustion method between “homogeneous combustion” and “stratified combustion” according to the operating state of the engine 11.
That is, the ECU 92 obtains the engine speed NE based on the detection signal from the crank position sensor 14c. Further, the ECU 92 calculates the basic fuel injection amount Qbse, which is a value corresponding to the engine load, based on a parameter related to the intake air amount of the engine 11 and the like. As such parameters, when the current combustion method is “homogeneous combustion”, the intake pressure PM of the engine 11 directly related to the intake air amount is adopted, and when the current combustion method is “stratified combustion”. An accelerator depression amount ACCP or the like indirectly related to the intake air amount is employed. The intake pressure PM is obtained based on the detection signal from the vacuum sensor 36, and the accelerator depression amount ACCP is obtained based on the detection signal from the accelerator position sensor 26.
[0043]
In accordance with the basic fuel injection amount Qbse (engine load) and the engine speed NE, the ECU 92 is in a state in which the current operation state is to execute any one of “stratified combustion” and “homogeneous combustion”. It is determined whether or not there is a combustion method corresponding to the determination. That is, “homogeneous combustion” is performed when the operating state of the engine 11 is in the high rotation / high load region, and “stratified combustion” is performed when the engine 11 is in the low rotation / low load region. In this way, the combustion method is changed because the engine output is increased by setting the air-fuel ratio of the air-fuel mixture to the rich side during high rotation and high load where high output is required, and low rotation and low load that does not require very high output This is because sometimes the air-fuel ratio is set to a lean value to improve fuel efficiency.
[0044]
When the combustion method of the engine 11 is “homogeneous combustion”, the ECU 92 drives and controls the fuel injection valve 40 to correspond to the final fuel injection amount Qfin obtained from the basic fuel injection amount Qbse during the intake stroke of the engine 11. An amount of fuel is injected into the combustion chamber 16. In the air-fuel mixture formed in the combustion chamber 16 based on such fuel injection, the air-fuel ratio becomes richer than the stoichiometric air-fuel ratio or the stoichiometric air-fuel ratio. Further, the ECU 92 obtains the actual throttle opening degree TAr based on the detection signal from the throttle position sensor 44. Then, the throttle motor 24 is driven and controlled so that the actual throttle opening TAr approaches the target throttle opening TAsj at the time of “homogeneous combustion”, so that the throttle opening of the engine 11 is suitable for “homogeneous combustion”.
[0045]
When the combustion method of the engine 11 is “stratified combustion”, the ECU 92 controls the fuel injection valve 40 to correspond to the final fuel injection amount Qfin obtained from the basic fuel injection amount Qbse during the compression stroke of the engine 11. An amount of fuel is injected and supplied. In the air-fuel mixture formed in the combustion chamber 16 by such fuel injection, the air-fuel ratio is set to a leaner value than the air-fuel ratio at the time of “homogeneous combustion”. Further, the ECU 92 drives and controls the throttle motor so that the actual throttle opening TAr approaches the target throttle opening TAtj at the time of “stratified combustion”, and the throttle opening of the engine 11 is suitable for “stratified combustion”. To do.
[0046]
During the “stratified combustion”, the fuel injected from the fuel injection valve 40 during the compression stroke of the engine 11 enters the recess 12 a (FIG. 1) provided in the head of the piston 12 and is ignited by the movement of the piston 12. Collected around the plug 41. By collecting the fuel around the spark plug 41 in this way, even if the average air-fuel ratio of the entire air-fuel mixture in the combustion chamber 16 is made leaner than during “homogeneous combustion”, the air-fuel ratio of the air-fuel mixture around the plug 41 is reduced. A good air-fuel mixture is ignited that is suitable for ignition. Further, in order to make the average air-fuel ratio of the entire air-fuel mixture in the combustion chamber 16 leaner than that in “homogeneous combustion”, the throttle opening is controlled to the open side and the intake air amount is increased. The pumping loss of the engine 11 is reduced.
[0047]
By the way, in the engine 11, when the engine output torque is adjusted to a required value, among the various control systems for controlling the operation of the engine 11, a different control system is used for each combustion method, and the engine output torque is adjusted. Make adjustments. Examples of the operation control of the engine 11 that requires adjustment of the engine output torque include so-called idle speed control (ISC) for controlling the idle speed of the engine 11.
[0048]
In the ISC, during the idling operation in “homogeneous combustion”, the amount of intake air is changed by adjusting the opening of the throttle valve 23 (throttle opening). When the intake air amount is adjusted in this way, the fuel injection amount determined in accordance with the intake air amount (intake pressure) changes, whereby the engine output torque is adjusted and the idle speed is controlled. Further, during idle operation in “stratified combustion”, the engine output torque is adjusted by directly adjusting the fuel injection amount to control the idle speed. This is because the fuel injection amount is not determined based on parameters (intake pressure, etc.) that are directly related to the intake air amount during “stratified combustion”. This is because the engine output torque does not easily change even if the operation is performed.
[0049]
Next, throttle opening control and fuel injection control during “homogeneous combustion” will be described.
As the target throttle opening degree TAsj at the time of the “homogeneous combustion”, an aggregated target throttle opening degree TAt described later is substituted. The aggregate target throttle opening degree TAt is equal to the “homogeneous combustion” in the case where it is assumed that “homogeneous combustion” is executed at the accelerator depression amount ACCP and the engine speed NE at that time, regardless of the currently executed combustion method. It is calculated as a suitable throttle opening. Therefore, when “homogeneous combustion” is being performed, the aggregated target throttle opening degree TAt is used as the target throttle opening degree TAsj suitable for the “homogeneous combustion”. When the target throttle opening degree TAsj at the time of “homogeneous combustion” is calculated in this way, the ECU 92 controls the throttle motor 24 based on the target throttle opening degree TAsj and the actual throttle opening degree TAr. 23 opening degree control is performed.
[0050]
That is, the ECU 92 calculates the post-phase advance compensated throttle opening degree TAh based on the following equation (1).
[0051]
[Expression 1]

As can be seen from the equation (1), the post-phase advance compensated throttle opening degree TAh is obtained by differentiating the actual throttle opening degree TAr with respect to time t and further multiplying by a predetermined coefficient Kd. It is a value calculated by adding to TAr. The phase advance-compensated throttle opening degree TAh calculated in this way becomes a value closer to the target throttle opening degree TAsj than the actual throttle opening degree TAr while the target throttle opening degree TAsj is changing.
[0052]
The ECU 92 calculates a difference e2 between the target throttle opening degree TAsj and the phase advance compensated throttle opening degree TAh. The ECU 92 drives and controls the throttle motor 24 so that the difference e2 approaches “0”, that is, the phase opening compensated throttle opening TAh approaches the target throttle opening TAsj.
[0053]
Here, FIG. 3 shows how the phase advance compensated throttle opening TAh and the actual throttle opening TAr change when the target throttle opening TAsj changes with time.
[0054]
When the target throttle opening degree TAsj changes as shown by a two-dot chain line in FIG. 3, the throttle opening degree TAh after phase advance compensation changes accordingly in the vicinity of the target throttle opening degree TAsj as shown by a thin solid line. When the throttle motor 24 is controlled such that the difference e2 between the phase advance-compensated throttle opening degree TAh and the target throttle opening degree TAsj that changes in this way approaches “0”, the actual throttle opening degree TAr becomes the target throttle opening degree. It changes with a predetermined response delay as shown by a thick solid line with respect to the change of the degree TAsj. The reason why the actual throttle opening degree TAr is delayed in response is to prevent the throttle opening degree TAr from overshooting.
[0055]
When the throttle opening control during “homogeneous combustion” is performed in this way, the actual intake pressure PMr (intake air amount) in the engine 11 corresponds to the opening of the throttle valve 23. The ECU 92 calculates the predicted intake pressure PMFWD from the actual throttle opening TAr, the actual intake pressure PMr, and the like. The predicted intake pressure PMFWD is a value obtained by predicting the intake pressure when the intake valve 19 is closed, and is calculated by an intake pressure calculation routine described later.
[0056]
The ECU 92 calculates a basic fuel injection amount Qbse based on the predicted intake pressure PMFWD and the engine speed NE. Then, the ECU 92 controls the fuel injection valve 40 to inject and supply the amount of fuel corresponding to the final fuel injection amount Qfin obtained from the basic fuel injection amount Qbse to the combustion chamber 16 during the intake stroke. By such fuel injection, “homogeneous combustion” is executed, and the engine output torque is controlled to a required value.
[0057]
Therefore, during the homogeneous combustion operation of the engine 11, the engine output torque is controlled by adjusting the throttle opening based on the target throttle opening TAsj to which the aggregate target throttle opening TAt is substituted.
[0058]
Next, fuel injection control during “stratified combustion” will be described.
The above-mentioned aggregate target throttle opening degree TAt is the same as the “homogeneous combustion” (hereinafter referred to as “the homogeneous combustion”) assuming that the “homogeneous combustion” is executed at the accelerator depression amount ACCP and the engine speed NE at the time of “stratified combustion”. It is calculated as a target throttle opening suitable for “virtual homogeneous combustion”. At the time of “stratified combustion”, the actual throttle opening TAr that should be obtained by performing the throttle opening control based on the aggregate target throttle opening TAt in the “virtual homogeneous combustion” is set to the aggregate target throttle opening TAt. Based on this, the virtual throttle opening degree TAv is calculated.
[0059]
That is, as shown in FIG. 3, the transition of the target throttle opening degree TAsj (aggregate target throttle opening degree TAt) at the time of “homogeneous combustion” and the transition of the throttle opening degree TAh after phase advance compensation are substantially equal. Assume that “TAh = TAsj (TAh = TAt)”. Under this assumption, the actual throttle opening degree TAr is calculated from the target throttle opening degree TAsj (aggregated target throttle opening degree TAt) by the procedure reverse to the above formula (1), and the throttle opening degree TAr is calculated as the virtual throttle opening degree. Degree TAv.
[0060]
Further, the virtual intake pressure PMv, which is the intake pressure (predicted intake pressure PMFWD) at the time of “homogeneous combustion (virtual homogeneous combustion)” in a state where the throttle opening of the engine 11 becomes the virtual throttle opening TAv, is determined as the virtual throttle. Calculated based on the opening degree TAv and the like. Based on the virtual intake pressure PMv (virtual intake air amount), the engine speed NE, and the like, the basic fuel injection amount Qbse at the time of “stratified combustion” is calculated.
[0061]
The ECU 92 controls the fuel injection valve 40 to inject and supply the fuel corresponding to the final fuel injection amount Qfin obtained from the basic fuel injection amount Qbse to the combustion chamber 16 during the compression stroke. The final fuel injection amount Qfin is a control target value of the fuel injection amount in the engine 11. By this fuel injection control based on the final fuel injection amount Qfin, “stratified combustion” is executed, and the engine output torque is controlled to a required value.
[0062]
Accordingly, during the stratified charge combustion operation of the engine 11, the virtual throttle opening degree TAv and the virtual intake pressure PMv corresponding to “virtual homogeneous combustion” are calculated based on the aggregate target throttle opening degree TAt, and calculated from this virtual intake pressure PMv and the like. The engine output torque is controlled by adjusting the fuel injection amount based on the final fuel injection amount Qfin (basic fuel injection amount Qbse). For this reason, at the time of “stratified combustion”, the aggregate target throttle opening degree TAt is calculated by an aggregate target throttle opening calculation routine described later as a value corresponding to the engine output torque required at the time of “stratified combustion”. Is done.
[0063]
As described above, during the stratified combustion operation, the final fuel injection amount Qfin is calculated according to the virtual throttle opening degree TAv and the virtual intake pressure PMv corresponding to “virtual homogeneous combustion”, and the final fuel injection amount Qfin is actually calculated at this time. This is related to the engine output torque obtained when "homogeneous combustion" is executed. Since the fuel injection amount is controlled based on the final fuel injection amount Qfin, the engine output torque during the stratified combustion operation corresponding to the final fuel injection amount Qfin is the engine when the homogeneous combustion operation is performed at this time. Matches the output torque. Thereby, when the combustion method is switched between “stratified combustion” and “homogeneous combustion”, it is possible to suppress the occurrence of a step in the engine output torque.
[0064]
Next, the procedure for calculating the final fuel injection amount Qfin will be described with reference to FIG. FIG. 4 is a flowchart showing a fuel injection amount calculation routine for calculating the final fuel injection amount Qfin based on the virtual intake pressure PMv and the predicted intake pressure PMFWD during the stratified charge combustion operation and the homogeneous combustion operation. The fuel injection amount calculation routine is executed by interruption at predetermined time intervals through the ECU 92.
[0065]
In the fuel injection amount calculation routine, the process of step S201 is for calculating the virtual intake pressure PMv or the predicted intake pressure PMFWD. After the process of step S201 is executed, the ECU 92 calculates the basic fuel injection amount Qbse by the following equation (2) using the virtual intake pressure PMv or the predicted intake pressure PMFWD as the intake pressure PM as the process of step S202. To do. That is, the basic fuel injection amount Qbse is calculated by multiplying the intake pressure PM by the intake air temperature correction coefficient Ktha and the constant K.
[0066]
[Expression 2]
Qbse = PM * Ktha * K (2)
Note that, when calculating the virtual intake pressure PMv and the predicted intake pressure PMFWD, the volume efficiency ηv described later is used, but the intake air temperature correction coefficient Ktha in the above equation (2) is the volume efficiency ηv due to the change in the intake air temperature THA. This is to compensate for changes. The ECU 92 obtains the intake air temperature THA based on the detection signal from the intake air temperature sensor 37 and calculates the intake air temperature correction coefficient Ktha with reference to the map of FIG. 5 based on the intake air temperature THA. The intake air temperature correction coefficient Ktha calculated in this way decreases as the intake air temperature THA increases and approaches “1.0”. Therefore, the corrected basic fuel injection amount Qbse increases as the intake air temperature THA decreases.
[0067]
After the process of step S202 is executed, the process proceeds to step S203. The processing in steps S203 and S204 is for calculating the final fuel injection amount Qfin from the basic fuel injection amount Qbse and the like.
[0068]
The ECU 92 calculates a mode correction coefficient Kmode as the process of step S203. This mode correction coefficient Kmode is a correction coefficient for compensating for the difference in required fuel injection amount due to the difference in combustion efficiency between “homogeneous combustion” and “stratified combustion”, and the ECU 92 depends on the current combustion method. A mode correction coefficient Kmode is calculated. This mode correction coefficient Kmode is set to “Kmode = 1.0” at the time of “homogeneous combustion” in which the combustion efficiency is lower than “stratified combustion”. The reason why the combustion efficiency is lower in “homogeneous combustion” than in “stratified combustion” is that pump loss and cooling loss are larger in “homogeneous combustion” than in “stratified combustion”.
[0069]
When the processing of step S203 is executed as described above and the mode correction coefficient Kmode is calculated, the ECU 92 calculates the final fuel injection quantity Qfin by multiplying the basic fuel injection quantity Qbse by the mode correction coefficient Kmode in the following step S204. .
[0070]
By using the mode correction coefficient Kmode to calculate the final fuel injection amount Qfin as described above, the final fuel injection amount Qfin is adjusted based on the difference in combustion efficiency for each combustion method. The final fuel injection amount Qfin is adjusted to the decrease side with respect to the "homogeneous combustion". By performing the fuel injection control based on the final fuel injection amount Qfin calculated taking into account the difference in combustion efficiency for each combustion method, the engine output torque control based on the fuel injection amount control is performed regardless of which combustion method is executed. Accuracy will be improved.
[0071]
The ECU 92 determines whether or not the stratified combustion operation is being executed as the processing of the subsequent step S205. Then, if the stratified charge combustion operation is not being performed, the fuel injection amount calculation routine is temporarily ended. If the stratified charge combustion operation is being performed, the process of step S206 is executed, and then the fuel injection amount calculation routine is temporarily ended. The process of step S206 is to compensate for the deviation of the fuel injection amount from the appropriate value due to aging of the fuel system during “stratified combustion”.
[0072]
The ECU 92 corrects the final fuel injection amount Qfin based on the following equation (3) as a process of step S206.
[0073]
[Equation 3]

In the expression (3), the term of the fuel injection amount at the time of “stratified combustion” of the term added to the final fuel injection amount Qfin, that is, “{(qgtj / qgsj) −1} * C * (600 / NE)”. The deviation from the appropriate value is compensated. This compensation will be described in detail later. In equation (3), the stratified learning value qgtj is learned as a value corresponding to the amount of deviation of the fuel injection amount from the appropriate value by the ISC learning routine described later. The homogeneity learning value qgsj is learned by the ISC learning routine as a value corresponding to the amount of deviation from the appropriate value of the intake air amount due to aging of the intake system. Further, in the expression (3), the value “C * (600 / NE)” determined by the weighting coefficient C and the engine speed NE indicates that the correction of the final fuel injection amount Qfin based on the expression (3) becomes excessive. This is to suppress and optimize the correction.
[0074]
Next, the process of step S201 in the fuel injection amount calculation routine will be described in detail with reference to FIGS. 6 and 7 are flowcharts showing an intake pressure calculation routine for calculating the predicted intake pressure PMFWD and the virtual intake pressure PMv. This intake pressure calculation routine is executed through the ECU 92 every time the process proceeds to step S201 of the fuel injection amount calculation routine.
[0075]
In the intake pressure calculation routine, the ECU 92 determines whether or not the stratified charge combustion operation is being performed as a process of step S301 (FIG. 6). If the stratified charge combustion operation is being performed, as a process of step S302, based on the aggregate target throttle opening degree TAt and the actual throttle opening degree TAr, the virtual throttle opening degree that is the throttle opening degree TAt at the time of “virtual homogeneous combustion”. TAv is calculated. Thereafter, the process proceeds to step S303. If it is determined in step S301 that the stratified charge combustion operation is not being performed, the process proceeds directly to step S303.
[0076]
In step S303, the ECU 92 calculates the steady-state intake pressure PMbse based on the current actual throttle opening TAr or virtual throttle opening TAv and the engine speed NE. This steady-state intake pressure PMbse is the intake pressure when the engine 11 is in a steady operation in the state of the throttle openings TAr and TAv and the engine speed NE. The steady-state intake pressure PMbse is calculated based on the actual throttle opening degree TAr and the engine speed NE during the homogeneous combustion operation, and is calculated based on the virtual throttle opening degree TAv and the engine speed NE during the stratified combustion operation.
[0077]
In step S304, the ECU 92 calculates the atmospheric pressure correction coefficient Kpa1 based on the atmospheric pressure PA by referring to the map of FIG. 8, and multiplies the steady-state intake pressure PMbse by the atmospheric pressure correction coefficient Kpa1 to correct the atmospheric pressure correction coefficient Kpa1. The rear intake pressure PMh is calculated. The atmospheric pressure correction coefficient Kpa1 increases as the atmospheric pressure PA increases and approaches “1.0”. Therefore, the corrected intake pressure PMh increases as the atmospheric pressure PA increases. After the corrected intake pressure PMh is calculated, the process proceeds to step S305.
[0078]
The processing in step S305 is related to the processing in later steps S306 and S307. That is, in step S306, the gradual change value PMSM is calculated by gradually changing the corrected intake pressure PMh, and in step S307, the gradual change value PMSM is stored as the first intake pressure memory value PMSM1. The In the process of step S305, the ECU 92 sets the first intake pressure memory value PMSM1 stored in the previous process of step S307 as the previous gradual change value PMSMi-1.
[0079]
The gradual change value PMSM calculated by the gradual change process (S306) is temporarily stored as the first intake pressure memory value PMSM1 (S307) by using the gradual change value PMSM in the process of step S310 described later. This is because another process is executed and the gradual change value PMSM is changed by the process. Even in this case, the gradual change process of step S306 can be appropriately performed by setting the first intake pressure storage value PMSM1 to the previous gradual change value PMSMi-1 in the process of step S305.
[0080]
After the process of step S305 is executed, the ECU 92 calculates the current gradual change value PMSMi based on the following equation (4) as the process of step S306. That is, by subtracting the previous gradual change value PMSMi-1 from the corrected intake pressure PMh at normal time and further dividing by the predetermined value n, this divided value is added to the previous gradual change value PMSMi-1. The gradual change value PMSMi is calculated.
[0081]
[Expression 4]
PMSMi = PMSMi-1 + (PMh-PMSMi-1) / n (4)
Here, the transition tendency of the gradually changing value PMSM with respect to the change of the corrected intake pressure PMh is shown in FIG. In the figure, the transition of the corrected intake pressure PMh is indicated by a broken line, and the transition of the gradually changing value PMSM is indicated by a thick solid line. Further, while the corrected intake pressure PMh calculated by map calculation or the like changes as indicated by a broken line, how the actual intake pressure PMr changes is indicated by a two-dot chain line.
[0082]
As is apparent from this figure, for example, when the accelerator depression amount ACCP changes and the corrected intake pressure PMh changes as indicated by a broken line, the gradually changing value PMSM is thicker than the corrected intake pressure PMh. As shown by the solid line, it changes gradually. How gently the gradual change value PMSM changes with respect to the change in the corrected intake pressure PMh is determined by the predetermined value n in the above equation (4). The predetermined value n is calculated based on the corrected intake pressure PMh and the engine speed NE with reference to a map (not shown) set in advance by experiments or the like.
[0083]
When the gradual change value PMSM is calculated in the process of step S306 and the first intake pressure memory value PMSM1 is stored in the process of step S307, the process proceeds to step S308. The processes in steps S308 to S311 are for predicting and calculating the gradually changing value PMSM when the intake valve 19 is closed at the present time.
[0084]
The ECU 92 calculates the number of times that the process of step S306 is performed (the number of times of gradual change process) T / Δt from the current time until the intake valve 19 is closed as the process of step S308. That is, the time T from the current time to the closing time of the intake valve 19 is obtained, and the time T is divided by the execution period Δt of the fuel injection amount calculation routine to calculate the gradual change processing count T / Δt.
[0085]
Subsequently, the ECU 92 sets the first intake pressure memory value PMSM1 currently stored as the process of step S309, that is, the latest gradual change value PMSM as the previous gradual change value PMSMi-1. Further, the ECU 92 executes the gradual change process according to the above equation (4) for the number of times of the gradual change process T / Δt as the process of step S310. A gradual change value PMSMi when the intake valve 19 is closed is calculated. Thereafter, the ECU 92 stores the gradual change value PMSMi as the second intake pressure memory value PMSM2 as the process of step S311.
[0086]
Now, assuming that the process of step S306 is performed at the time indicated by the alternate long and short dash line L1 in FIG. 9, the current gradual change value PMSMi calculated by the process is stored as the first intake pressure memory value PMSM1. Then, when the process of step S310 is subsequently performed, a gradually changing value PMSMi when the intake valve 19 is closed indicated by a two-dot chain line L2 is calculated, and at the time when the gradually changing value PMSMMi is substantially indicated by a one-dot chain line L1. It is stored as the second intake pressure memory value PMSM2.
[0087]
After the first and second intake pressure memory values PMSM1 and PMSM2 are thus stored, the difference ΔP1 (“PMSM2−PMSM1”) between the stored values PMSM1 and PMSM2 is used to determine the intake valve 19's value. The intake pressure when the valve is closed can be predicted and calculated. That is, the difference ΔP1 between the first and second intake pressure memory values PMSM1 and PMSM2 is added to the actual intake pressure PMr detected by the vacuum sensor 36 at the present time (the one-dot chain line L1). The intake pressure when the valve is closed can be obtained.
[0088]
Incidentally, since the output of the vacuum sensor 36 is affected by the pulsation of the air flowing in the intake passage 32, the output of the vacuum sensor 36 is usually filtered by a CR filter or the like in order to remove the influence. Therefore, the intake pressure PMr actually deviates from an appropriate value by the time constant of the filter processing by the CR filter or the like, and the estimated intake pressure when the intake valve 19 is closed is inconsistent by the deviation. Become accurate.
[0089]
In the processing of steps S312 (FIG. 7) to S315 in the intake pressure calculation routine, the first intake pressure memory value PMSM1 is filtered in consideration of the deviation of the intake pressure PMr, and the intake valve 19 is used by using the filter output PMSM1Si. This is for accurately predicting the intake pressure when the valve is closed.
[0090]
After executing the process of step S311 (FIG. 6), the ECU 92 determines whether or not the current combustion method is homogeneous combustion as the process of step S312 (FIG. 7). If it is homogeneous combustion, the process proceeds to step S313. move on. The ECU 92 filters the first intake pressure memory value PMSM1 based on the following equation (5) as the process of step S313. In Equation (5), PMSM1Si is the filter output of the first intake pressure memory value PMSM1, and the predetermined value m is set so that the time constant of the filter processing is equal to the time constant of the filter processing by the CR filter. It is what is done.
[0091]
[Equation 5]
PMSM1Si
= PMSM1Si-1 + (PMSM1-PMSM1Si-1) / m (5)
The filter output PMSM1Si of the filter processing based on the equation (5) is shown by a thin solid line in the figure when the gradually changing value PMSM (first intake pressure memory value PMSM1) changes as shown by a thick solid line in FIG. It will change as follows.
[0092]
Subsequently, as a process of step S314, the ECU 92 subtracts the filter output PMSM1Si from the second intake pressure memory value PMSM2 to calculate a difference ΔP2. Further, as a process of step S315, the ECU 92 adds the difference ΔP2 to the actual intake pressure PMr, and further multiplies the added value by the volume efficiency ηv to obtain the intake pressure when the intake valve 19 is closed. Calculated as the predicted intake pressure PMFWD. The volume efficiency ηv is calculated with reference to a map based on the previous predicted intake pressure PMFWD and the engine speed NE. After calculating the predicted intake pressure PMFWD in this way, the intake pressure calculation routine is temporarily terminated and the routine returns to the fuel injection amount calculation routine (FIG. 4).
[0093]
Accordingly, when the storage processing of the first and second intake pressure stored values PMSM1, PMSM2 is performed at the time indicated by the one-dot chain line L1 in FIG. 9, the filter output of the first intake pressure stored value PMSM1 at that time PMSM1Si is used to calculate the predicted intake pressure PMFWD. That is, the predicted intake pressure PMFWD is calculated by adding the difference ΔP2 between the second intake pressure stored value PMSM2 and the filter output PMSM1Si at the time indicated by the one-dot chain line L1 to the actual intake pressure PMr.
[0094]
In this way, the difference ΔP2 is calculated using the filter output PMSM1Si instead of the first intake pressure memory value PMSM1, and the predicted intake pressure PMFWD is obtained from the difference ΔP2 or the like, so that the time constant of the CR filter is obtained as the intake pressure PMr. Even if a deviation occurs according to the estimated intake pressure PMFWD, the predicted intake pressure PMFWD can be accurately calculated as the intake pressure when the intake valve 19 is closed.
[0095]
On the other hand, if it is determined in the process of step S312 that the current combustion method is stratified combustion instead of homogeneous combustion, the process proceeds to step S316. As a process of step S316, the ECU 92 calculates a value obtained by multiplying the second intake pressure memory value PMSM2 by the volume efficiency ηv as the virtual intake pressure PMv. The volume efficiency ηv is calculated with reference to a map based on the previous virtual intake pressure PMv and the engine speed NE. After calculating the virtual intake pressure PMv in this way, the intake pressure calculation routine is temporarily terminated and the routine returns to the fuel injection amount calculation routine (FIG. 4).
[0096]
The calculated virtual intake pressure PMv is the intake pressure when the intake valve 19 is closed when it is assumed that “homogeneous combustion” is executed in the current engine operating state (“virtual homogeneous combustion”), that is, This is a virtual value corresponding to the predicted intake pressure PMFWD. During the homogeneous combustion operation, the predicted intake pressure PMFWD is calculated based on the actual intake pressure PMr and the like, so the processing of steps S312 to S315 is performed to accurately calculate the predicted intake pressure PMFWD. On the other hand, during the stratified combustion operation, the virtual intake pressure PMv is calculated based on the second intake pressure memory value PMSM2, etc., regardless of the actual intake pressure PMr. The virtual intake pressure PMv calculated in this way is calculated as an accurate value by the process of step S316.
[0097]
Next, a procedure for calculating the aggregate target throttle opening degree TAt will be described with reference to FIG. FIG. 10 is a flowchart showing an aggregate target throttle opening routine for calculating the aggregate target throttle opening TAt. This aggregated target throttle opening calculation routine is executed through the ECU 92 by, for example, a time interruption every predetermined time.
[0098]
In the aggregated target throttle opening calculation routine, the basic throttle opening TAbse, the ISC opening correction amount f (qcal), and other correction amounts A are used in the process of step S405, based on the following equation (6). The aggregate target throttle opening degree TAt is calculated.
[0099]
[Formula 6]
TAt = TAbse + f (qcal) + A (6)
By the way, in the above equation (6), the ISC opening correction amount f (qcal) is for adjusting the engine output torque to control the idling speed. The ISC opening correction amount f (qcal) is calculated by the following equation (7) based on the ISC correction amount qcal and the conversion coefficient Ka.
[0100]
[Expression 7]
f (qcal) = qcal * Ka (7)
In Expression (7), the ISC correction amount qcal is a dimensionless parameter corresponding to the adjustment amount of the idle rotation speed. The conversion coefficient Ka is for converting the ISC correction amount qcal into a change amount of the engine output torque required for adjusting the idle speed, that is, an adjustment amount of the throttle opening.
[0101]
The idle speed is adjusted by increasing or decreasing the ISC correction amount qcal. When the ISC opening correction amount f (qcal) changes with the increase / decrease of the ISC correction amount qcal, the aggregate target throttle opening degree TAt also changes. The aggregate target throttle opening degree TAt is a target throttle opening degree TAsj that is a parameter that affects the engine output torque during “homogeneous combustion” and a final fuel that is a parameter that affects the engine output torque during “stratified combustion”. Each is related to the injection quantity Qfin. Therefore, by changing the aggregate target throttle opening degree TAt by increasing / decreasing the ISC opening correction amount f (qcal), the engine output torque changes and the idling speed is adjusted regardless of the combustion method.
[0102]
The ISC correction amount qcal for calculating the ISC opening correction amount f (qcal) is calculated by the following equation (8) based on the feedback correction term qi, the calculation learning value qg, the water temperature correction term qthw, and the like. The
[0103]
[Equation 8]
qcal = qi + qg + qthw (8)
In the equation (8), the feedback correction term qi is a predetermined reference value (in this embodiment, according to the engine speed NE during idle operation) in order to bring the idle speed close to a predetermined target value (for example, 600 rpm). “0”).
[0104]
That is, the feedback correction term qi is increased if the idle speed is lower than the target value. As a result, the ISC correction amount qcal and the ISC opening correction amount f (qcal) increase, and the aggregate target throttle opening amount TAt calculated based on the ISC opening correction amount f (qcal) changes to the open side value. To do. Regardless of whether it is “homogeneous combustion” or “stratified combustion”, the aggregate target throttle opening degree TAt changes to the open side value as described above, so that the engine output torque increases and the idling engine speed becomes the target value. Rise.
[0105]
At the time of “homogeneous combustion”, the aggregate target throttle opening degree TAt is substituted for the target throttle opening degree TAsj, and therefore, the throttle opening degree control is performed based on the target throttle opening degree TAsj, thereby predicting the intake pressure PMFWD (intake air amount). Will increase. As a result, the final fuel injection amount Qfin determined based on the predicted intake pressure PMFWD or the like increases, the engine output torque increases, and the idle speed increases toward the target value.
[0106]
Unlike the above, at the time of “stratified combustion”, the virtual throttle opening degree TAv that is the throttle opening degree at the time of “virtual homogeneous combustion” is calculated based on the aggregate target throttle opening degree TAt and the like. Further, a virtual intake pressure PMv, which is an intake pressure (intake air amount) when the throttle opening is set to the virtual throttle opening TAv during “virtual homogeneous combustion”, is calculated based on the virtual throttle opening TAv and the like. Therefore, the virtual intake pressure PMv increases when the aggregate target throttle opening degree TAt changes to the open side value by the ISC opening correction amount f (qcal). As a result, the final fuel injection amount Qfin determined according to the virtual intake pressure PMv or the like increases, the engine output torque increases, and the idle speed increases toward the target value.
[0107]
The feedback correction term qi is made smaller if the idle speed is higher than the target value. As a result, the ISC correction amount qcal and the ISC opening correction amount f (qcal) are reduced, and the aggregate target throttle opening amount TAt calculated based on the ISC opening correction amount f (qcal) or the like changes to a close side value. To do. Regardless of whether it is “homogeneous combustion” or “stratified combustion”, the aggregate target throttle opening degree TAt changes to the closed value as described above, so that the engine output torque decreases and the idling engine speed becomes the target value. And descend.
[0108]
At the time of “homogeneous combustion”, the aggregated target throttle opening degree TAt is substituted for the target throttle opening degree TAsj. Therefore, by performing throttle opening degree control based on the target throttle opening degree TAsj, the predicted intake pressure PMFWD (intake air amount) Decrease. As a result, the final fuel injection amount Qfin determined based on the predicted intake pressure PMFWD or the like is decreased, the engine output torque is decreased, and the idling engine speed is decreased toward the target value.
[0109]
Unlike the above, at the time of “stratified combustion”, the virtual throttle opening degree TAv is calculated based on the aggregate target throttle opening degree TAt and the like, and the virtual intake pressure PMv is calculated based on the virtual throttle opening degree TAv and the like. Therefore, the virtual intake pressure PMv decreases when the aggregate target throttle opening degree TAt changes to the close side value by the ISC opening correction amount f (qcal). As a result, the final fuel injection amount Qfin determined according to the virtual intake pressure PMv or the like is reduced, the engine output torque is reduced, and the idling engine speed is lowered toward the target value.
[0110]
The learning value for calculation qg in the equation (8) converges within a predetermined range (in this embodiment, “−γ <qi <γ”) in which the feedback correction term qi includes the reference value “0” during idling. Thus, the value is increased or decreased around the reference value “0” based on the feedback correction term qi.
[0111]
That is, the calculation learning value qg is reduced if the feedback correction term qi deviates from the predetermined range. As a result, as is apparent from the above equation (8), under the condition that the ISC correction amount qcal is maintained in a state where the idle rotation speed matches the target value, the feedback correction term qi decreases as the calculation learning value qg decreases. Increases and the correction term qi converges within the predetermined range.
[0112]
The calculation learning value qg is increased if the feedback correction term qi is out of the predetermined range. As a result, as apparent from the above equation (8), under the condition that the ISC correction amount qcal is maintained in a state where the idling engine speed matches the target value, the feedback correction term is increased when the calculation ISC correction amount qcal is increased. qi becomes small, and the correction term qi converges within the predetermined range.
[0113]
The water temperature correction term qthw in the above equation (8) increases or decreases according to the cooling water temperature THW of the engine 11.
As described above, the ISC correction amount qcal is calculated from the feedback correction term qi, the calculation learning value qg, the water temperature correction term qthw, and the like. The calculation learning value qg is a learning value qgsj for homogeneity for each combustion method. And a stratified learning value qgtj are prepared.
[0114]
At the time of “homogeneous combustion”, the learning value for homogeneity qgsj is substituted into the learning value for calculation qg, and the learning value for homogeneity qgsj is learned. That is, during the idling operation in “homogeneous combustion”, the calculation learning value qg is increased / decreased to converge the feedback correction term qi within a predetermined range, and the calculated learning value qg after the increase / decrease is the homogeneity learning value. Assigned to qgsj. In this case, learning of the homogenous learning value qgsj is completed when the feedback correction term qi converges within a predetermined range.
[0115]
If the intake air amount deviates from an appropriate value due to a secular change or the like in the intake system of the engine 11, the engine output torque deviates from the appropriate value during idle operation in "homogeneous combustion", and the idling speed deviates from the target value. It will be. The feedback correction term qi is increased or decreased to bring the idle speed closer to the target value, and the calculation learning value qg (homogeneous learning value qgsj) is increased or decreased so that the feedback correction term qi converges within a predetermined range. When the feedback correction term qi converges within a predetermined range and learning of the homogenous learning value qgsj is completed, the homogenous learning value qgsj becomes a value corresponding to the deviation amount of the intake air amount from the appropriate value. .
[0116]
Therefore, by reflecting this homogeneity learning value qgsj in the subsequent throttle opening control, it is possible to compensate for the deviation of the intake air amount from the appropriate value. In order to reflect the homogeneity learning value qgsj in the throttle opening control, the homogeneity learning value qgsj is reflected in the ISC correction amount qcal as the calculation learning value qg. The ISC correction amount qcal is reflected in the aggregate target throttle opening degree TAt as the ISC opening correction amount f (qcal). At the time of “homogeneous combustion”, the throttle opening degree control is performed based on the target throttle opening degree TAsj at the time of “homogeneous combustion” to which this aggregated target throttle opening degree TAt is substituted. qgsj is reflected.
[0117]
Further, at the time of “stratified combustion”, the stratified learning value qgtj is substituted into the calculation learned value qg, and learning of the stratified learning value qgtj is performed. That is, during the idling operation in “stratified combustion”, the calculation learning value qg is increased or decreased to converge the feedback correction term qi within a predetermined range, and the calculated learning value qg is used as the stratified learning value. Assigned to qgtj. In this case, learning of the stratified learning value qgtj is completed when the feedback correction term qi converges within a predetermined range.
[0118]
When the fuel system of the engine 11 changes over time and the fuel injection amount deviates from an appropriate value, the engine output torque deviates from the appropriate value and the idling speed deviates from the target value during idling operation in “stratified combustion”. It will be. The feedback correction term qi is increased or decreased to bring the idle speed closer to the target value, and the calculation learning value qg (stratification learning value qgtj) is increased or decreased so that the feedback correction term qi converges within a predetermined range. When the feedback correction term qi converges within a predetermined range and learning of the stratification learning value qgtj is completed, the stratification learning value qgtj becomes a value corresponding to the deviation amount from the appropriate value of the fuel injection amount. .
[0119]
Accordingly, by reflecting the stratified learning value qgtj in the fuel injection control, it is possible to compensate for the deviation from the fuel injection amount appropriate value.
By the way, when the stratified learning value qgtj is reflected in the fuel injection control, and the stratified learning value qgtj is reflected as the calculation learning value qg in the ISC correction amount qcal, the ISC correction amount qcal is calculated as the ISC opening correction amount. As f (qcal) is reflected in the aggregate target throttle opening degree TAt, a step is generated in the engine output torque between the combustion methods.
[0120]
Aggregation target for homogeneity learning value qgsj and stratification learning value qgtj when the learning value qgsj for homogeneity and the learning value qgtj for stratification are switched as the learning value for calculation qg according to the combustion method and reflected in the ISC correction amount qcal The relationship of the throttle opening degree TAt is shown in FIG. FIG. 11A shows the homogeneity learning value qgsj and the stratification learning value qgtj. The solid line in FIG. 11B shows the same engine operating state when the engine 11 is in a steady state. The integrated target throttle opening degree TAt calculated at the time of “homogeneous combustion” in FIG.
[0121]
As shown in FIG. 11 (a), the learning value qgsj for homogeneity and the learning value qgtj for stratification correspond to different control systems such as the intake air amount and the fuel injection amount, and the deviation amounts from the appropriate values of the intake air amount, respectively. , And the value corresponding to the amount of deviation from the appropriate value of the fuel injection amount, it may be a different value. That is, when the deviation from the appropriate value of the intake air amount and the deviation from the appropriate value of the fuel injection amount are different from each other, the homogeneity learning value qgsj and the stratification learning value qgtj for compensating for the deviation are also different from each other. Value. As a result, the ISC correction amount qcal reflected in the aggregate target throttle opening degree TAt differs between “homogeneous combustion” and “stratified combustion”. As shown by the solid line in FIG. The aggregated target throttle opening degree TAt calculated in step 1 and the aggregated target throttle opening degree TAt calculated in “stratified combustion” are also different values.
[0122]
By the way, the aggregate target throttle opening degree TAt is equal to that at the time of “homogeneous combustion” as shown by the broken line in FIG. 11B under the same engine operating condition in order to match the engine output torque between the combustion methods. The value should be equal to that in “stratified combustion”.
[0123]
Therefore, in the present embodiment, instead of reflecting the stratified learning value qgtj at the time of “stratified combustion”, the learned value qgsj for homogeneity is reflected at the time of “stratified combustion” with respect to the aggregate target throttle opening degree TAt. As a result, the aggregate target throttle opening degree TAt at the time of “stratified combustion” becomes a value indicated by a broken line in FIG. 11B, and becomes the value indicated by a solid line at the time of “homogeneous combustion”. It matches the opening degree TAt. In this case, the fuel injection amount at the time of “stratified combustion” controlled based on the aggregate target throttle opening degree TAt and the like corresponds to the learning value qgsj for homogeneity, and the deviation of the fuel injection amount from the appropriate value cannot be compensated as it is. Therefore, by correcting the final fuel injection amount Qfin at the time of “stratified combustion” based on the above equation (3), the fuel amount corresponding to the deviation amount between the homogenized learning value qgsj and the stratified learning value qgtj is added. Controls the fuel injection amount during “stratified combustion”.
[0124]
In this way, by reflecting the homogeneity learning value qgsj and the stratification learning value qgtj with respect to the aggregate target throttle opening TAt and the fuel injection amount, while compensating for the deviation of the intake air amount and the fuel injection amount from the appropriate values, The engine output torque can be matched between the combustion methods.
[0125]
Now, the description returns to the aggregate target throttle opening calculation routine (FIG. 10).
In the aggregate target throttle opening degree calculation routine, the ECU 92 determines the basic throttle opening degree TAbse at the time of “homogeneous combustion” based on the current accelerator depression amount ACCP and the engine speed NE regardless of the current combustion method as the process of step S401. calculate. Thereafter, the process proceeds to step S402.
[0126]
In step S402, the ECU 92 assigns the homogeneity learning value qgsj to the calculation learning value qg, and then proceeds to step S403.
The ECU 92 calculates the ISC correction amount qcal by the above equation (8) as the processing of step S403, and calculates the ISC opening correction amount f (qcal) by the above equation (7) as the processing of step S404. Subsequently, as a process of step S405, the ECU 92 calculates the aggregate target throttle opening degree TAt by the above equation (6), and then ends the aggregate target throttle opening degree calculation routine.
[0127]
Next, the learning procedure of the homogeneity learning value qgsj and the stratification learning value qgtj will be described with reference to FIGS. 12 and 13 are flowcharts showing an ISC learning routine for learning the homogeneity learning value qgsj and the stratification learning value qgtj. This ISC learning routine is executed through the ECU 92 by, for example, a time interruption every predetermined time.
[0128]
In the ISC learning routine, it is determined whether or not the learning values qgtj and qgsj can be increased or decreased (learning) in the process of step S501 (FIG. 12), that is, whether or not the learning condition is satisfied. In steps S502 to S504, the learning value for stratification qgtj or the learning value for homogeneity qgsj is substituted for the learning value for calculation qg according to the combustion method. Further, in the processes of S505 to S508, the calculation learning value qg is increased or decreased to converge the feedback correction term qi within a predetermined range (“−γ <qi <γ” in the present embodiment). In steps S509 to S517 (FIG. 13), the learning value qg after the calculation learning value qg for operating the feedback correction term qi to converge to a predetermined range is used as the learning value for stratification according to the combustion method. The value is substituted into the value qgtj or the homogeneity learning value qgsj. This makes it possible to have an ISC learning value separately for each combustion method.
[0129]
In the ISC learning routine, the ECU 92 determines whether or not an ISC learning condition is satisfied, that is, for example, whether or not all of the following various conditions are satisfied, as processing of step S501.
[0130]
-ISC feedback control in progress
-Small fluctuations in engine speed NE
・ No other disturbance
If any one of the above conditions is not satisfied, the ECU 92 once ends this ISC learning routine. If all the above conditions are satisfied, the process proceeds to step S502.
[0131]
The ECU 92 determines whether or not the stratified combustion operation is being performed as a process of step S502. If the stratified charge combustion operation is being performed, the stratified learning value qgtj is substituted into the calculation learning value qg as the process of step S503. If the stratified charge combustion operation is not being performed but the homogeneous combustion operation is being performed, the calculation learning is performed as step S504. The homogeneity learning value qgsj is substituted for the value qg. After making the learning value qg for calculation correspond to the combustion method in this way, the processing after step S505 is executed.
[0132]
In step S505, the ECU 92 determines whether the feedback correction term qi is larger than a predetermined value γ. Then, when it is determined that “qi> γ” and the feedback correction term qi is out of the predetermined range “−γ to γ”, the calculation learning value qg is set to a predetermined value as processing in step S506. Add a. When the learning value for calculation qg is increased in this way, the feedback correction term qi is set within the predetermined range “−γ to γ” under the condition that the ISC correction amount qcal is maintained in a state where the idle speed matches the target value. Become smaller.
[0133]
If it is determined in step S505 that “qi> γ” is not established, the process proceeds to step S507 to determine whether the feedback correction term qi is less than a predetermined value −γ. When it is determined that “qi <−γ” and the feedback correction term qi is out of the predetermined range “−γ to γ”, it is determined from the calculation learning value qg as processing in step S508. Subtract the value a. When the calculation learning value qg is reduced in this way, the feedback correction term qi moves toward the predetermined range “−γ to γ” under the condition that the ISC correction amount qcal is maintained in a state where the idle speed matches the target value. Become bigger.
[0134]
After the increase or decrease of the calculation learning value qg for converging the feedback correction term qi within the predetermined range “−γ to γ” as described above, the process proceeds to step S509 (FIG. 13). The ECU 92 determines whether or not the stratified combustion operation is being performed as the process of step S509. If the stratified charge combustion operation is being performed, the current calculation learned value qg is substituted into the stratified charge learning value qgtj as a process in step S510. The current learning value for calculation qg is substituted for the learning value for qgsj. Thus, by substituting the learning value for calculation qg into the learning value for stratification qgtj or the learning value for homogeneity qgsj, it becomes possible to have an ISC learning value for each combustion method.
[0135]
The stratification learning value qgtj and the homogeneity learning value qgsj are stored in the backup RAM 96. When the feedback correction term qi converges within the predetermined range during “stratified combustion”, learning of the stratified learning value qgtj is completed, and when the feedback correction term qi converges within the predetermined range during “homogeneous combustion” The learning of the homogenous learning value qgsj is completed.
[0136]
In the ISC learning routine, after the process of step S510 is performed, the processes of steps S511 to S513 are performed. In these processes, when learning of the stratification learning value qgtj is performed, the learning value for stratification qgtj is reflected on the learning value for homogeneity qgsj under a predetermined condition, and the learning of the learning value for homogeneity qgsj is completed early. It is for making it happen.
[0137]
In step S511, the ECU 92 determines whether or not a speed-up condition, which is a criterion for determining whether or not the learning of the homogenized learning value qgsj needs to be performed at an early stage, is satisfied. These conditions include:
・ When the battery is replaced
When the learning value qgsj for homogeneity and the learning value qgtj for stratification become values that are excessively separated from the reference value “0”
And so on.
[0138]
Note that when any one of the above speed-up conditions is satisfied, the homogeneity learning value qgsj and the stratification learning value qgtj are returned to the initial values. This initial value is set to a value larger than the reference value “0” of the homogeneity learning value qgsj and the stratification learning value qgtj. Therefore, when the above conditions are satisfied, it is necessary to quickly learn the learning value qgsj for homogeneity and the learning value qgtj for stratification, and reduce the learning value qgsj for homogeneity and the learning value qgtj for stratification to the reference value “0” side at an early stage. There is.
[0139]
If it is determined in step S511 that the speed-up condition is not satisfied, the ISC learning routine is temporarily terminated. If it is determined that the condition is satisfied, the process proceeds to step S512. In step S512, the ECU 92 determines whether or not the current homogeneity learning value qgsj is larger than the value obtained by adding the predetermined value α to the calculation learning value qg (stratification learning value qgtj) (“qg + α”). Determine whether.
[0140]
When it is determined that “qgsj> qg + α” is not satisfied and the homogeneity learning value qgsj is not an excessively large value with respect to the reference value “0”, the ISC learning routine is temporarily terminated, and “qgsj> qg + α” is satisfied. If it is determined that the homogeneity learning value qgsj is an excessively large value with respect to the reference value “0”, the process proceeds to step S513. In step S513, the ECU 92 assigns the value “qg + α” to the current learning value qgsj for homogeneity, and then ends this ISC learning routine. In this way, by reflecting the stratification learning value qgtj in the homogeneity learning value qgsj, learning of the homogeneity learning value qgsj can be completed early.
[0141]
In the ISC learning routine, after the processing in step S514 is performed, the processing in steps S515 to S517 is performed. In these processes, when the learning value qgsj for homogeneity is learned, the learning value qgsj for homogeneity is reflected on the learning value qgtj for stratification under a predetermined condition, and the learning of the stratification learning value qgtj is completed early. It is for making it happen.
[0142]
In step S515, the ECU 92 determines whether a speed-up condition that is a criterion for determining whether the learning of the homogenized learning value qgsj needs to be performed at an early stage is satisfied. Examples of such conditions include the same conditions as in step S511.
[0143]
If it is determined in step S515 that the speed-up condition is not satisfied, the ISC learning routine is temporarily terminated. If it is determined that the condition is satisfied, the process proceeds to step S516. In step S516, the ECU 92 determines whether or not the current learning value for stratification qgtj is larger than the value (“qg + α”) obtained by adding the predetermined value α to the learning value for calculation qg (learning value for homogeneity qgsj). Determine whether.
[0144]
When it is determined that the learning value for stratification qgtj is not an excessively large value with respect to the reference value “0” instead of “qgtj> qg + α”, the ISC learning routine is temporarily ended, and “qgtj> qg + α” If it is determined that the learning value qgtj for stratification is an excessively large value with respect to the reference value “0”, the process proceeds to step S517. In step S517, the ECU 92 assigns the value “qg + α” to the current stratification learning value qgtj, and then ends this ISC learning routine. In this way, by reflecting the homogeneity learning value qgsj in the stratification learning value qgtj, learning of the stratification learning value qgtj can be completed early.
[0145]
The stratification learning value qgtj and the homogeneity learning value qgsj learned by the ISC learning routine are used in the above equation (3) in the process of step S206 (FIG. 5) in the final injection amount calculation routine.
[0146]
That is, in the expression (3), the learning value for stratification qgtj and the homogeneity are obtained by “{(qgtj / qgsj) −1} * C * (600 / NE)” which is a term added to the final fuel injection amount Qfin. The learning value qgsj for use is used. The final fuel injection amount Qfin is calculated on the basis of the aggregate target throttle opening degree TAt and the like, but the calculation learning value qg used for calculating the aggregate target throttle opening degree TAt is fixed to the homogeneity learning value qgsj.
[0147]
Accordingly, the final fuel injection amount Qfin calculated based on the aggregate target throttle opening degree TAt and the like corresponds to the learning value qgsj for homogeneity even though “stratified combustion” is being performed during the stratified combustion operation. Therefore, at the time of “stratified charge combustion”, the term “{(qgtj / qgsj) −1} * C * (600 / NE)” is added to the final fuel injection amount Qfin so that the final fuel injection amount Qfin is appropriately set. Correct to something. That is, the term “{(qgtj / qgsj) −1} * C * (600 / NE)” is the amount of fuel corresponding to the amount of deviation between the learning value for homogeneity qgsj and the learning value for stratification qgtj. Thus, by calculating the final fuel injection amount Qfin taking this term into account, the learning value for stratification qgtj is added to the fuel injection amount control, and the deviation of the fuel injection amount from the appropriate value is compensated.
[0148]
According to the present embodiment in which the processing detailed above is performed, the following effects can be obtained.
(1) At the time of “homogeneous combustion”, the control target of the throttle opening control at the time of “homogeneous combustion” by adding the learning value qgsj for homogeneity learned as a value corresponding to the deviation amount from the appropriate value of the intake air amount A target throttle opening degree TAsj (aggregated target throttle opening degree TAt), which is a value, is calculated. Therefore, by performing throttle opening control based on the target throttle opening TAsj and the like, it is possible to accurately compensate for the deviation of the intake air amount from the appropriate value.
[0149]
Further, at the time of “stratified combustion”, the virtual throttle opening degree TAv and the virtual intake pressure PMv (virtual intake air amount) at the time of “virtual homogeneous combustion” are calculated in consideration of the homogeneity learning value qgsj. Then, the basic fuel injection amount Qbse is calculated based on the virtual intake pressure PMv and the like, and the final fuel injection amount Qfin which is the control target value of the fuel injection amount by adding the stratified learning value qgtj based on the basic fuel injection amount Qbse and the like. Is calculated.
[0150]
By calculating the basic fuel injection amount Qbse based on the virtual intake pressure PMv to which the homogeneity learning value qgsj is added as described above, the final fuel injection amount which is the control target value of the fuel injection amount at the time of “stratified combustion” This is related to the engine output torque obtained when Qfin performs “homogenous combustion” at this time. Since the fuel injection amount at the time of “stratified combustion” is controlled based on the final fuel injection amount Qfin, the engine output torque at the time of “stratified combustion” is the engine output when “homogeneous combustion” is performed at this time. It will match the torque accurately.
[0151]
Further, at the time of “stratified combustion”, the final fuel injection amount Qfin is calculated by adding the stratified learning value qgtj based on the basic fuel injection amount Qbse and the like, so that the stratified learning value qgtj is added to the final fuel injection amount Qfin. As a result, the deviation of the fuel injection amount from the appropriate value can be compensated accurately.
[0152]
Therefore, it is possible to match the engine output torque between the combustion methods while compensating for the deviation of the intake air amount and the fuel injection amount, which are control amounts for controlling the idle speed, from the appropriate values. If the engine output torque differs between the combustion methods, it is easy to feel a torque shock due to this when the combustion method is switched during deceleration of the vehicle. However, as described above, the engine output torque is different between the combustion methods. By combining them, a torque shock is not felt when the combustion system is switched during deceleration of the automobile.
[0153]
(2) At the time of “stratified combustion”, the final fuel injection amount Qfin (basic fuel injection amount Qbse) is calculated based on the virtual intake pressure PMv to which the homogeneity learning value qgsj is added. ), The term “qgtj / qgsj” makes the final fuel injection amount Qfin take into account both the stratification learning value qgtj and the homogeneity learning value qgsj. Therefore, when the deviation from the appropriate value of the fuel injection amount is compensated by the learning value for stratification qgtj, it is suppressed that this becomes excessive due to the virtual intake pressure PMv in consideration of the learning value for homogeneity qgsj. Can do.
[0154]
(3) Since the learning value qgsj for homogeneity and the learning value qgtj for stratification are learned during idle operation, the fuel injection amount at the time of “stratified combustion” is determined by these learning values qgsj and qgtj as the engine speed NE increases. This is excessive when compensating for deviations from the proper values. However, at the time of “stratified combustion”, the final fuel injection amount Qfin takes into account the engine speed NE by the term “600 / NE” in the above equation (3), and the appropriate value of the fuel injection amount by the learning values qgsj and qgtj. It is possible to suppress the excess when compensating for the deviation from the above.
[0155]
(4) When the stratified learning value qgtj is learned at the time of “stratified combustion”, if the speed-up condition is satisfied, the stratified learning value qgtj (the learning value for calculation qg) at that time is the learned value for homogeneity qgsj. It is reflected in. Further, when the homogeneity learning value qgsj is learned during “homogeneous combustion”, if the speed-up condition is satisfied, the homogeneity learning value qgsj (calculation learning value qg) at that time becomes the stratification learning value qgtj. Reflected. By reflecting one learning value in the other learning value in this way, when the stratification learning value qgtj and the homogeneity learning value qgsj are returned to the initial values, the learning of these learning values qgtj and qgsj is performed early. Can be completed.
[0156]
In addition, this embodiment can also be changed as follows, for example.
In the present embodiment, the homogeneity learning value qgsj and the stratification learning value qgtj are mutually reflected when the speed-up condition is established, and learning of these learning values qgsj and qgtj can be completed early. The learning values qgsj and qgtj do not necessarily have to reflect each other. If the homogeneity learning value qgsj and the stratification learning value qgtj are not reflected from each other, the processing of steps S511 to S513 and the processing of steps S515 to S517 in the ISC learning routine (FIG. 13) can be omitted.
[Brief description of the drawings]
FIG. 1 is a cross-sectional view showing an entire engine to which a control device of an embodiment is applied.
FIG. 2 is a block diagram showing an electrical configuration of the control device.
FIG. 3 is a time chart showing changes in phase opening compensated throttle opening TAh and actual throttle opening TAr with respect to a change in target throttle opening TAsj.
FIG. 4 is a flowchart showing a procedure for calculating a final fuel injection amount Qfin.
FIG. 5 is a map referred to when calculating an intake air temperature correction coefficient Ktha.
FIG. 6 is a flowchart showing a procedure for calculating a predicted intake pressure PMFWD and a virtual intake pressure PMv.
FIG. 7 is a flowchart showing a procedure for calculating a predicted intake pressure PMFWD and a virtual intake pressure PMv.
FIG. 8 is a map that is referred to when an atmospheric pressure correction coefficient Kpa1 is calculated.
FIG. 9 is a time chart showing changes in corrected intake pressure PMh, gradually changing value PMSM, filter output PMSM1Si, and actual intake pressure PMr;
FIG. 10 is a flowchart showing a procedure for calculating an aggregate target throttle opening degree TAt.
FIG. 11 is a diagram showing the relationship of the aggregate target throttle opening degree TAt to the homogeneity learning value qgsj and the stratification learning value qgtj.
FIG. 12 is a flowchart showing a learning procedure for a homogeneity learning value qgsj and a stratification learning value qgtj.
FIG. 13 is a flowchart showing a learning procedure for the homogeneity learning value qgsj and the stratification learning value qgtj.
[Explanation of symbols]
DESCRIPTION OF SYMBOLS 11 ... Engine, 11b ... Water temperature sensor, 14c ... Crank position sensor, 21b ... Cam position sensor, 23 ... Throttle valve, 24 ... Motor for throttle, 25 ... Accelerator pedal, 26 ... Accelerator position sensor, 36 ... Vacuum sensor, 37 ... Intake temperature sensor, 40 ... fuel injection valve, 44 ... throttle position sensor, 92 ... electronic control unit (ECU).

Claims (5)

The combustion mode is switched between homogeneous combustion and stratified combustion according to the engine operating state, and the idle speed is adjusted by intake air amount control during homogeneous combustion, and by fuel injection amount control during stratified combustion operation. Applied to internal combustion engines, learns the value corresponding to the deviation from the appropriate value of the intake air amount during idle operation in homogeneous combustion as a learning value for homogeneity, and from the appropriate value of fuel injection amount during idle operation in stratified combustion In a control device for an internal combustion engine that learns a value corresponding to a deviation as a learning value for stratification,
Control of the intake air amount based on the control target value of the intake air amount during the homogeneous combustion operation, and by calculating the control target value of the intake air amount in consideration of the learning value for homogeneity, An intake air amount control means that reflects the learning value for homogeneity in the control of the intake air amount so as to compensate for a deviation from an appropriate value of the intake air amount;
A virtual intake air amount calculating means for calculating a virtual intake air amount that is an intake air amount when it is assumed that the homogeneous combustion operation is executed in the engine operation state at the time of the stratified combustion operation in consideration of the learning value for homogeneity;
The fuel injection amount is controlled based on the control target value of the fuel injection amount during the stratified combustion operation, and the control target value of the fuel injection amount is added to the stratified learning value based on the virtual intake air amount. A fuel injection amount control means for reflecting the learning value for stratification in the control of the fuel injection amount so as to compensate for a deviation from an appropriate value of the fuel injection amount by calculating
A control device for an internal combustion engine, comprising: 2. The internal combustion engine according to claim 1, wherein the fuel injection amount control means takes into account the homogeneity learning value in addition to the stratification learning value when calculating a control target value of the fuel injection amount based on the virtual intake air amount. Control device. The control device for an internal combustion engine according to claim 1 or 2, wherein the fuel injection amount control means further considers the engine speed when calculating a control target value of the fuel injection amount based on the virtual intake air amount. The virtual intake air amount calculation means takes into account the virtual throttle opening, which is the opening of the throttle valve of the engine when the homogeneous combustion operation is executed in the engine operation state during the stratified combustion operation, with the learning value for homogeneity. The control device for an internal combustion engine according to claim 1, wherein the virtual intake air amount is calculated from the virtual throttle opening. The control apparatus for an internal combustion engine according to any one of claims 1 to 4,
When learning of the learning value corresponding to the combustion method is performed during idling operation in any one of the homogeneous combustion and the stratified combustion method, the learned value is converted into the homogeneous combustion and the stratified combustion under a predetermined condition. The learning value reflecting means for reflecting in the learning value corresponding to the other combustion method is further provided.
A control device for an internal combustion engine.

Images