JP2010019227A - Air-fuel ratio control device for internal combustion engine - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an air-fuel ratio control device for an internal combustion engine, for suppressing the worsening of exhaust emission even when disturbance occurs after learning determined to be converged is restricted. <P>SOLUTION: The internal combustion engine 10 has a plurality of cylinder groups (a first bank 11 and a second bank 12). In each cylinder group, an air-fuel ratio feedback control system is provided including a catalyst 18, a main air-fuel ratio sensor 20 and a sub oxygen sensor 22. When the learning of sub feedback control is determined to be converged, subsequent learning is restricted. When the learning is determined to be converged in each cylinder group, a predetermined index value for one cylinder group, calculated in accordance with the output of the sub oxygen sensor 22, is compared with that for the other. When a difference between the index value for one cylinder group and that for the other exceeds a predetermined value, the restriction of the learning is cancelled. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to an air-fuel ratio control apparatus for an internal combustion engine.

  Japanese Patent Laid-Open No. 2003-120381 includes an oxygen sensor that detects an exhaust oxygen concentration in the middle of an exhaust passage provided for each of a plurality of cylinder groups of a multi-cylinder internal combustion engine, and each cylinder is based on the output of the oxygen sensor. An air-fuel ratio control apparatus that performs group air-fuel ratio feedback control is disclosed. This apparatus performs learning control of the air-fuel ratio at the time of air-fuel ratio feedback control for each cylinder group, and on the condition that each learning result has converged, the learning result of any cylinder group and the oxygen sensor output of the cylinder group The air-fuel ratio feedback control based on the above is performed for each cylinder group.

  Further, when learning in air-fuel ratio feedback control has converged, a technique for restricting subsequent learning is conventionally known. According to this technique, the learning value can be prevented from fluctuating excessively, so that the center of the control air-fuel ratio can be stabilized. For this reason, excessive correction can be reliably prevented, and stable air-fuel ratio feedback control can be performed.

JP 2003-120381 A JP 2007-170344 A JP 2007-162626 A Japanese Patent Laid-Open No. 10-220267

  After learning converges and learning is limited, disturbances such as misfire may occur. Due to the influence of the disturbance, the output of the air-fuel ratio sensor arranged on the upstream side of the catalyst may shift. In such a case, it is desirable to correct the influence of the output deviation of the air-fuel ratio sensor by changing the learning value. However, since learning is limited, the followability of the learning value becomes slow. For this reason, the influence of the deviation of the air-fuel ratio sensor output cannot be corrected, and the exhaust emission tends to deteriorate.

  The present invention has been made to solve the above-described problems, and suppresses deterioration of exhaust emission even when disturbance occurs after learning is determined to have converged and learning is limited. An object of the present invention is to provide an air-fuel ratio control apparatus for an internal combustion engine that can perform the above-described operation.

In order to achieve the above object, a first invention is an air-fuel ratio control apparatus for an internal combustion engine having a plurality of cylinder groups and including an air-fuel ratio feedback control system for each of the cylinder groups.
The air-fuel ratio feedback control system for each cylinder group is:
An exhaust passage,
A catalyst disposed in the exhaust passage;
A main exhaust gas sensor disposed upstream of the catalyst;
A sub exhaust gas sensor disposed downstream of the catalyst;
Main feedback means for performing main feedback control of the air-fuel ratio based on the output of the main exhaust gas sensor;
Sub-feedback learning means including means for calculating a sub-feedback correction value based on the output of the sub-exhaust gas sensor, and means for calculating a learning value for correcting a stationary error component based on the sub-feedback correction value When,
Convergence determination means for determining convergence of learning by the sub-feedback learning means;
Learning limiting means for limiting subsequent learning when it is determined that the learning has converged;
Have
A comparing means for comparing a predetermined index value calculated based on an output of the sub exhaust gas sensor between the cylinder groups when it is determined that the learning has converged in each of the cylinder groups;
Restriction release means for releasing the restriction by the learning restriction means when the difference between the index values between the cylinder groups exceeds a predetermined value;
It is characterized by providing.

The second invention is the first invention, wherein
The index value is an integral value of a deviation between an output value of the sub exhaust gas sensor and a control target value thereof.

The third invention is the first invention, wherein
The index value is a time until the output of the sub exhaust gas sensor changes from a lean output to a rich output when the catalyst is saturated with oxygen, or when the catalyst is in an oxygen-deficient state. It is a time until the output of the exhaust gas sensor changes from rich output to lean output.

  According to the first invention, after it is determined that the learning value has deviated from the required value after the learning of the sub feedback control has converged, the learning limitation can be released. Thereby, the followability of learning can be improved, and the deviated learning value can be quickly brought close to the required value. For this reason, the control deviation of the air-fuel ratio can be corrected quickly, and deterioration of exhaust emission can be suppressed.

  According to the second invention, by using the integrated value of the deviation between the output value of the sub exhaust gas sensor and the control target value as an index, it is possible to accurately determine whether or not the learned value is deviated from the required value. it can.

  According to the third invention, whether the learning value deviates from the required value by using, as an index, the time until the output of the sub exhaust gas sensor is reversed when the catalyst is in an oxygen saturated state or an oxygen deficient state. Whether or not can be determined with high accuracy. Even if the sub-feedback control is not being executed, it can be determined whether or not the learning value is deviated from the required value, so that the determination can be made early.

Embodiment 1 FIG.
[Description of system configuration]
FIG. 1 is a diagram for explaining a system configuration according to the first embodiment of the present invention. As shown in FIG. 1, the system of this embodiment includes an internal combustion engine 10 mounted as a power source in a vehicle, and an ECU (Electronic Control Unit) 50 that controls the internal combustion engine 10. The internal combustion engine 10 of the present embodiment has a first bank 11 and a second bank 12, and each bank is composed of three cylinders. That is, the internal combustion engine 10 is a V type 6 cylinder.

  In the present embodiment, an air-fuel ratio feedback control system is individually provided for each of the first bank 11 and the second bank 12. As shown in FIG. 1, the air-fuel ratio feedback control system of each of the first bank 11 and the second bank 12 has an exhaust manifold 14 that collects exhaust gases of cylinders that constitute the bank, and the collected exhaust gas flows. An exhaust passage 16, a catalyst 18 provided in the middle of the exhaust passage 16, a main air-fuel ratio sensor 20 disposed on the upstream side of the catalyst 18, and a sub oxygen sensor 22 disposed on the downstream side of the catalyst 18 are provided. is doing.

  The catalyst 18 has a function as a three-way catalyst that can favorably purify HC, CO, and NOx when the air-fuel ratio of the inflowing exhaust gas is near the stoichiometric air-fuel ratio. The main air-fuel ratio sensor 20 is a sensor that emits a substantially linear output with respect to the air-fuel ratio of the exhaust gas. The sub oxygen sensor 22 is a sensor that emits an output that changes suddenly depending on whether the exhaust gas is rich or lean with respect to the stoichiometric air-fuel ratio.

  In the following description, the air-fuel ratio feedback control system of the first bank 11 is called a first system, and the air-fuel ratio feedback control system of the second bank 12 is called a second system.

  Although omitted in FIG. 1, an intake passage is connected to each cylinder of the internal combustion engine 10. A fuel injection device 24 is provided in each of the first bank 11 and the second bank 12.

  The main air-fuel ratio sensor 20, the sub oxygen sensor 22, and the fuel injection device 24 described above are electrically connected to the ECU 50. In the present embodiment, the ECU 50 is shared by the first system and the second system. Although not shown in FIG. 1, the ECU 50 includes an ignition device, a throttle valve that adjusts the intake air amount, an air flow meter that detects the intake air amount, an accelerator position sensor that detects the accelerator pedal position, and a crankshaft Various sensors and actuators usually used for engine control, such as a crank angle sensor for detecting a rotation angle, are further connected.

(Air-fuel ratio feedback control)
The air-fuel ratio feedback control systems of the first system and the second system each execute air-fuel ratio feedback control as described below. The air-fuel ratio feedback control of the present embodiment is configured by combining main feedback control based on the output of the main air-fuel ratio sensor 20 and sub feedback control based on the output of the sub oxygen sensor 22. The ECU 50 calculates a corrected A / F output evabyf expressed by the following equation based on the output evafbse of the main air-fuel ratio sensor 20 and the output of the sub oxygen sensor 22, and the corrected A / F output evabyf is the target. A process of controlling the fuel injection amount so as to become a value corresponding to the air-fuel ratio is executed.

evabyf = evafbse + evafsfb + evafsfbg (1)
In the above formula (1), “evafbse” in the first term on the right side is the output voltage of the main air-fuel ratio sensor 20. Further, “evafsfb” in the second term on the right side is a sub feedback correction value calculated based on the output of the sub oxygen sensor 22. The third term “evafsfbg” on the right side is a learning value.

  In the present embodiment, the corrected A / F output evabyf is calculated according to the above equation (1), and further, main feedback control is performed to bring the corrected A / F output evabyf closer to a value corresponding to the target air-fuel ratio. The In this main feedback control, specifically, a process of converting the corrected A / F output evabyf into an air-fuel ratio, a process of calculating a deviation ΔA / F between the air-fuel ratio obtained as a result and the target air-fuel ratio, and A process of reflecting the deviation ΔA / F in the correction of the fuel injection amount with a predetermined gain is executed.

  When the main air-fuel ratio sensor 20 exhibits ideal characteristics, the output evafbse and the air-fuel ratio upstream of the catalyst 18 (hereinafter referred to as “pre-catalyst air-fuel ratio”) have a unique relationship. In this case, if the main feedback is executed so that the output evafbse of the main air-fuel ratio sensor 20 becomes a value corresponding to the stoichiometric air-fuel ratio, the exhaust gas flowing into the catalyst 18 has an air-fuel ratio in the vicinity of the stoichiometric air-fuel ratio. Therefore, only the purified exhaust gas flows out downstream of the catalyst 18.

  However, in reality, the main air-fuel ratio sensor 20 does not necessarily have ideal output characteristics due to individual differences or aging of the main air-fuel ratio sensor 20 and the signal transmission system, or changes in the operating state of the internal combustion engine 10. It is not something that demonstrates. For this reason, rich or lean exhaust gas may blow through downstream of the catalyst 18 even under conditions where main feedback control is being executed. According to the sub oxygen sensor 22, it is possible to accurately detect such a situation. Therefore, when the sub oxygen sensor 22 detects that rich exhaust gas has blown downstream of the catalyst 18, it can be determined that the pre-catalyst air-fuel ratio has shifted to the rich side as a whole. In this case, if the output evafbse of the main air-fuel ratio sensor 20 is corrected so that the fuel injection amount is calculated to be smaller than the current amount, the pre-catalyst air-fuel ratio obtained as a result of the main feedback is brought close to the stoichiometric air-fuel ratio. Is possible.

  On the other hand, when the sub oxygen sensor 22 detects that the lean exhaust gas has blown downstream of the catalyst 18, it can be determined that the pre-catalyst air-fuel ratio has shifted to the lean side as a whole. In this case, if the output evafbse of the main air-fuel ratio sensor 20 is corrected so that the fuel injection amount is calculated to be larger than the current amount, the pre-catalyst air-fuel ratio obtained as a result of the main feedback is brought close to the stoichiometric air-fuel ratio. Is possible. The sub-feedback correction value evafsfb included in the above equation (1) is a correction value for realizing such a function. That is, the sub feedback control has a function that complements the main feedback control.

  Specifically, the ECU 50 calculates the sub feedback correction value evafsfb by performing a predetermined calculation on the deviation between the output of the sub oxygen sensor 22 and its target value (output value corresponding to the theoretical air-fuel ratio). More specifically, in the present embodiment, the sub feedback correction value evafsfb is calculated by PID control. That is, the sub feedback correction value evafsfb is calculated as the sum of a proportional term, an integral term, and a derivative term based on the deviation between the output of the sub oxygen sensor 22 and its target value.

  The integral term component included in the sub-feedback correction value evafsfb can be grasped as an error that is steadily inherent in the main feedback. The learning value evafsfbg in the above equation (1) is a value calculated by transferring the integral term component from the sub feedback correction value evafsfb at a predetermined update timing.

  According to such processing, the error component steadily included in the main feedback control is absorbed in the learning value evafsfbg, and only the fluctuation component of the error component included in the main feedback control is absorbed in the sub feedback correction value evafsfb. Can do. As learning progresses, the learning value evafsfbg converges to a value that appropriately reflects the steady error component described above, and takes a stable value.

  Such a learning value evafsfbg is stored in a standby RAM provided in the ECU 50. The standby RAM is a memory that retains stored information by continuing power supply even when the ignition switch is turned off. When learning has converged, the steady error component can be corrected immediately by using the learning value evafsfbg stored in the standby RAM. For this reason, the air-fuel ratio can be brought close to the target air-fuel ratio quickly after the internal combustion engine 10 is started or after the air-fuel ratio feedback control is resumed.

The convergence of learning can be determined by any of the following learning convergence determination conditions 1 to 3.
(Learning convergence judgment condition 1)
FIG. 2 is a diagram for explaining learning convergence determination conditions. FIG. 2A shows a waveform of the output of the sub oxygen sensor 22.

When the learning value evafsfbg is initialized, the error component that is steadily included cannot be sufficiently corrected, and for a while, the pre-catalyst air-fuel ratio is biased to either lean or rich. Will continue. Meanwhile, the output of the sub oxygen sensor 22 is biased to either the lean output or the rich output. In the example shown in FIG. 2 (A), with prior to time t _1, the output of the sub oxygen sensor 22 is biased to the lean output.

At time t ₁ before the state of FIG. 2 (A), the fuel injection amount is gradually corrected in the increasing direction by the main and sub feedback control, eventually, the upstream air-fuel ratio becomes richer than the stoichiometric air-fuel ratio. When rich exhaust gas flows into the catalyst 18, the oxygen stored in the catalyst 18 is consumed. When the oxygen in the catalyst 18 is depleted, rich exhaust gas flows out downstream of the catalyst 18. As a result, the output of the sub oxygen sensor 22 is inverted from lean to rich (time t ₁ ).

From time t _1, the fuel injection amount will be corrected in the reduction direction by the main and sub feedback control. Eventually, the pre-catalyst air-fuel ratio becomes leaner than the stoichiometric air-fuel ratio. When lean exhaust gas flows into the catalyst 18, oxygen is occluded in the catalyst 18. When oxygen is saturated in the catalyst 18, lean exhaust gas flows out downstream of the catalyst 18. As a result, the output of the sub oxygen sensor 22 is inverted from rich to lean (time t ₂ ).

Thereafter, the output of the sub oxygen sensor 22 is repeatedly inverted between the rich output and the lean output (time t ₃ , t ₄ , t ₅ ...). As a result, the learning value evafsfbg approaches an appropriate value and eventually converges. Therefore, an appropriate number of determinations is set in advance, and when it is confirmed that the number of output reversals of the sub oxygen sensor 22 has reached the number of determinations, it can be determined that learning has converged. Such a determination condition is the learning convergence determination condition 1.

(Learning convergence judgment condition 2)
FIG. 2B shows a waveform of an integrated value (hereinafter referred to as “deviation integrated value”) of the deviation between the output of the sub oxygen sensor 22 and its target value. The target value of the output of the sub oxygen sensor 22 is an output corresponding to the theoretical air-fuel ratio. As the deviation integrated value, for example, a smoothed value subjected to a smoothing process (smoothing process) using a low-pass filter or the like may be used. In the following description, including the case where the annealing value is used, it is simply referred to as “deviation integral value”.

  As shown in FIG. 2B, the integrated deviation value of the sub oxygen sensor 22 converges to a value within a certain range as learning progresses. Therefore, when the appropriate determination range and determination time are set in advance and it is confirmed that the deviation integrated value of the sub oxygen sensor 22 is within the determination range, the learning has converged. Can be determined. Such a determination condition is the learning convergence determination condition 2.

(Learning convergence judgment condition 3)
FIG. 3 is a diagram for explaining learning convergence determination condition 3. Here, FIG. 3A shows a waveform of the output of the sub oxygen sensor 22 similar to FIG. 3B represents a change in the integral term component in the sub feedback correction value evafsfb, and FIG. 3C represents a change in the learning value evafsfbg.

  As shown in FIGS. 3B and 3C, the ECU 50 updates the learning value evafsfbg by transferring the integral term component in the sub feedback correction value evafsfb to the learning value evafsfbg at a predetermined update period. . The broken line in FIG. 3B represents the waveform of the integral term component of the sub feedback correction value evafsfb when the transfer to the learning value evafsfbg is not performed.

  Further, the ECU 50 sequentially calculates a value obtained by integrating the update amount of the learning value evafsfbg over a predetermined integration time (hereinafter referred to as “learning value update amount integrated value”). FIG. 3D shows a change in the learning value update amount integrated value. FIG. 3D shows an example in which the length of the integration time is set so that the learning value update amount integrated value is calculated as the sum of the update amounts for the past three times.

  When the learning value evafsfbg converges to an appropriate value and the fluctuation thereof decreases, the learning value update amount integrated value converges to a value near 0 as shown in FIG. Therefore, when an appropriate integration time and an appropriate determination range near 0 are set in advance and it is confirmed that the learning value update amount integrated value over the integration time has reached the determination range, the learning is performed. It can be determined that it has converged. Such a determination condition is the learning convergence determination condition 3.

  The ECU 50 determines whether the learning has converged for each of the first bank 11 and the second bank 12 using at least one of the learning convergence determination conditions 1 to 3 described above. When it is determined that the learning has converged for each of the first bank 11 and the second bank 12, the ECU 50 executes a process of restricting subsequent learning.

  Specific processing for restricting learning includes, for example, processing for changing the sub feedback gain used for calculating the sub feedback correction value evafsfb to a value smaller than that before learning convergence. By restricting learning after the convergence of learning, it is possible to prevent the learning value from fluctuating excessively, so that the center of the control air-fuel ratio can be stabilized. Therefore, it is possible to reliably prevent the correction from being performed excessively, and to perform stable air-fuel ratio feedback control.

However, after learning converges and learning is limited, if the ignition plug becomes abnormal in any cylinder and misfire occurs, the controllability of the air-fuel ratio may deteriorate. When misfire occurs, a lot of H ₂ is generated in the exhaust gas. The main air-fuel ratio sensor 20 has a characteristic that when the amount of H ₂ in the exhaust gas increases, the output shifts in a richer direction than the actual air-fuel ratio. When the output of the main air-fuel ratio sensor 20 shifts in the rich direction, the main feedback control tries to change the air-fuel ratio in the lean direction. As a result, the actual air-fuel ratio is likely to deviate more leanly than the target air-fuel ratio. For this reason, when a misfire occurs, the air-fuel ratio tends to shift to a leaner side than the target air-fuel ratio. For this reason, exhaust emission tends to deteriorate.

On the other hand, the output of the sub oxygen sensor 22 is not easily affected by misfire. This is because even if H ₂ increases due to the occurrence of misfire, the H ₂ disappears due to the reaction at the catalyst 18, and therefore hardly reaches the sub oxygen sensor 22. For this reason, even if misfire occurs, learning by sub-feedback control acts in a normal direction. Therefore, when the air-fuel ratio shifts to a leaner side than the target air-fuel ratio, the learned value evafsfbg changes to a value that returns the air-fuel ratio to the target air-fuel ratio. However, since learning is limited after learning has converged, the follow-up of the learning value evafsfbg is slow. For this reason, the learned value evafsfbg continues to take a value that deviates from a required value (hereinafter simply referred to as “required value”) required to return the air-fuel ratio to the vicinity of the target air-fuel ratio. As a result, the state where the air-fuel ratio is shifted to the lean side from the target air-fuel ratio cannot be corrected sufficiently.

The same situation also occurs when an abnormality occurs in the main air-fuel ratio sensor 20, or when excessive fuel occurs in some cylinders. That is, when a crack is generated in the element of the main air-fuel ratio sensor 20 and exhaust gas is mixed into the atmospheric chamber from the crack, the output of the main air-fuel ratio sensor 20 is offset as compared with the normal time. For this reason, control deviation of the main feedback control occurs, and the learning value evafsfbg deviates from the required value. Further, in some cylinders, when excessive fuel is generated due to a change in the characteristics of the fuel injector, a large amount of H ₂ is generated in the exhaust gas as in the case of misfire. For this reason, as in the case of misfire, control deviation of the main feedback control occurs, and the learned value evafsfbg deviates from the required value.

  When the learning value evafsfbg deviates from the required value due to the above-described misfire, disturbance of the main air-fuel ratio sensor 20, excessive fuel, or the like, it is desirable to cancel the learning restriction. If the restriction on learning is removed, the follow-up performance becomes faster, so that the learning value evafsfbg can be quickly brought close to the required value. For this reason, the deviation between the target air-fuel ratio and the actual air-fuel ratio can be corrected quickly.

  When the learning value evafsfbg matches the required value in both the first system and the second system, the air-fuel ratio is well controlled in the vicinity of the target air-fuel ratio in both the first system and the second system. Is done. In such a case, the output of the sub-oxygen sensor 22 of the first system and the output of the sub-oxygen sensor 22 of the second system are repeatedly inverted in the same manner between the rich output and the lean output. As a result, the deviation integrated value of the output of the first system sub oxygen sensor 22 and the deviation integrated value of the output of the second system sub oxygen sensor 22 converge to substantially the same value.

  On the other hand, if the learned value evafsfbg deviates from the required value in at least one of the first system and the second system, the air-fuel ratio of the system deviates from the target air-fuel ratio, so the output of the sub oxygen sensor 22 Tends to be rich or lean. In such a case, the deviation integrated value of the output of the first system sub oxygen sensor 22 and the deviation integrated value of the output of the second system sub oxygen sensor 22 do not converge to the same value.

  Therefore, if the deviation integrated value of the output of the sub-oxygen sensor 22 of the first system and the deviation integrated value of the output of the sub-oxygen sensor 22 of the second system are compared, the learning value evafsfbg of at least one system is calculated from the required value. It is possible to accurately determine whether or not there is a divergence. Therefore, in the present embodiment, it is determined whether or not the learning value evafsfbg deviates from the required value by such a method, and if it is determined that there is a deviation, learning in the air-fuel ratio feedback control system of both systems is determined. The restriction was lifted. By canceling the restriction of learning, the followability of the learning value evafsfbg is improved, so that the deviating learning value evafsfbg can be quickly brought close to the required value. Therefore, since the deviation between the actual air-fuel ratio and the target air-fuel ratio can be quickly eliminated, the deterioration of exhaust emission can be reliably suppressed.

[Specific Processing in Embodiment 1]
FIG. 4 is a flowchart of a routine executed by the ECU 50 in the present embodiment in order to realize the above function. According to the routine shown in FIG. 4, first, in the air-fuel ratio feedback control systems of both the first system and the second system, whether the learning of the sub feedback control has converged and the sub feedback control is being executed. It is determined whether or not (step 100). Specifically, whether or not the learning completion flag XSFBGEND indicating the convergence of learning of both systems is turned on and the flag XSFB indicating that the sub-feedback control of both systems is being executed is turned on. It is confirmed.

  When learning of the sub-feedback control of both systems has not converged, it is not necessary to execute the following processing. In the present embodiment, the deviation between the learning value evafsfbg and the required value is determined by comparing the deviation integrated values of the outputs of the sub-oxygen sensors 22 of both systems. For this reason, as a precondition for the determination, it is a condition that the sub-feedback control of both systems is being executed. Therefore, if one or both of the two conditions are not satisfied in step 100, the processing of this routine is terminated as it is.

  On the other hand, if both of the above two conditions are satisfied in step 100, then, the deviation integrated value SUMVO1 of the output of the first system sub oxygen sensor 22 and the second system sub oxygen sensor 22 are next. The deviation integrated value SUMVO2 of the output is read and it is determined whether or not the absolute value of the difference between the two is greater than or equal to a predetermined value α (step 102).

  If the difference between the deviation integrated values of the sub-oxygen sensors 22 of both systems is smaller than the predetermined value α in step 102, it can be determined that the learned value evafsfbg has not deviated from the required value in any system. In this case, since it can be determined that good air-fuel ratio feedback control is being executed in any system, there is no need for relearning. Therefore, in this case, the processing of this routine is terminated as it is.

  On the other hand, when the difference between the deviation integrated values of the sub-oxygen sensors 22 in both systems is equal to or greater than the predetermined value α in step 102, it is determined that the learning value evafsfbg has deviated from the required value in at least one system. it can. In this case, the learning completion flag XSFBGEND is turned off to execute relearning (step 104). When the learning completion flag XSFBGEND is turned off, it is considered that learning has not converged in both systems, so that the restriction on learning is canceled in both systems. As a result, the followability of the learning value evafsfbg is improved, so that the learning value evafsfbg that is deviating can be quickly brought close to the required value. Therefore, the deviation of the air-fuel ratio can be corrected and the air-fuel ratio can be controlled in the vicinity of the target air-fuel ratio. As a result, deterioration of exhaust emission can be avoided.

  In this embodiment, it is determined whether or not the learning value evafsfbg deviates from the required value by comparing the deviation integrated values of the outputs of the sub oxygen sensors 22 of both systems. If it is an index value indicating the state of, it may be determined by comparing other values. For example, the learning value evafsfbg of both systems may be compared to determine the deviation from the required value.

  In the present embodiment, the internal combustion engine 10 has been described as having a V-type 6 cylinder. However, the number of cylinders and the cylinder arrangement of the internal combustion engine in the present invention are not limited to this, and air-fuel ratio feedback control is individually performed. What is necessary is just to be provided with the some cylinder group which performs.

  In the first embodiment described above, the first bank 11 and the second bank 12 are the “cylinder group” in the first invention, and the main air-fuel ratio sensor 20 is the “main exhaust gas sensor” in the first invention. The sub oxygen sensor 22 corresponds to the “sub exhaust gas sensor” in the first invention, and the deviation integrated value of the output of the sub oxygen sensor 22 corresponds to the “predetermined index value” in the first invention. Further, the ECU 50 performs main feedback control based on the output of the main air-fuel ratio sensor 20, so that the “main feedback means” in the first aspect of the invention is based on the output of the sub oxygen sensor 22 and the sub feedback correction value evafsfb and By calculating the learning value evafsfbg, the “sub-feedback learning means” in the first invention determines the convergence of learning in accordance with any one of the learning convergence conditions 1 to 3, thereby determining “convergence determination” in the first invention. When the “means” determines that learning has converged, the “learning limiting means” in the first aspect of the present invention performs the process of changing the sub-feedback gain to a value smaller than that before learning convergence. By executing the process, the “comparison means” in the first invention is the step described above. Is "restriction release means" of the invention by executing the processing of 104, are realized respectively.

Embodiment 2. FIG.
Next, the second embodiment of the present invention will be described with reference to FIG. 5 to FIG. 7. The difference from the first embodiment will be mainly described, and the description of the same matters will be described. Simplify or omit.

  This embodiment is the same as Embodiment 1 described above except that the determination method for determining whether or not the learned value evafsfbg deviates from the required value. FIGS. 5 and 6 are diagrams showing changes in the outputs of the sub-oxygen sensors 22 of the first system and the second system from the start of the internal combustion engine 10.

  When the internal combustion engine 10 is started, the first system 18 and the second system catalyst 18 are both in an oxygen saturated state. For a while after the start, the oxygen stored in the catalyst 18 is released, so that the output of the sub oxygen sensor 22 is maintained at a lean output. Then, when the oxygen stored in the catalyst 18 is consumed and the oxygen deficiency is reached, the output of the sub oxygen sensor 22 is inverted to the rich output. Thereafter, the output of the sub oxygen sensor 22 is repeatedly inverted between the rich output and the lean output.

  When the learning value evafsfbg substantially matches the required value in both the first system and the second system, the air-fuel ratio is accurately maintained near the target air-fuel ratio in any system. For this reason, the amount of oxygen occluded in the first system and second system catalyst 18 after the start decreases at substantially the same pace. Further, as described above, since the first system and the second system catalyst 18 at the time of start-up are both in an oxygen saturation state, the amount of oxygen stored in the first system catalyst 18 at the time of start-up, The amount of oxygen stored in the system catalyst 18 coincides with the amount of oxygen. Accordingly, after the start-up, the time until the oxygen of the first system catalyst 18 is depleted is substantially equal to the time until the oxygen of the second system catalyst 18 is depleted. Therefore, as shown in FIG. 5, after the start, the time until the output of the first system sub oxygen sensor 22 is inverted to the rich output and the time until the output of the second system sub oxygen sensor 22 is inverted to the rich output. Is substantially equal to this time.

  On the other hand, when the learned value evafsfbg deviates from the required value in at least one of the first system and the second system, the deviation of the air-fuel ratio from the target air-fuel ratio becomes large. For this reason, after starting, the pace at which the amount of oxygen stored in the catalyst 18 decreases does not match between the first system and the second system. As a result, after the start-up, the time until the oxygen of the first system catalyst 18 is depleted does not match the time until the oxygen of the second system catalyst 18 is depleted. Therefore, as shown in FIG. 6, after the start, the time T1 until the output of the first system sub oxygen sensor 22 is inverted to the rich output and the output of the second system sub oxygen sensor 22 are inverted to the rich output. The difference from the time T2 until is increased.

  Thus, after the internal combustion engine 10 is started, the time T1 until the output of the first system sub oxygen sensor 22 is inverted to the rich output and the output of the second system sub oxygen sensor 22 is inverted to the rich output. If it is compared with the time T2, it is possible to accurately determine whether or not the learning value evafsfbg of at least one system deviates from the required value. Therefore, in the present embodiment, it is determined whether or not the learning value evafsfbg deviates from the required value by such a method, and if it is determined that there is a deviation, learning in the air-fuel ratio feedback control system of both systems is determined. The restriction was lifted. Thereby, the same effect as Embodiment 1 is acquired.

  This embodiment also has the following advantages. After starting, it may take some time before the execution of the sub-feedback control is started. In the present embodiment, it is possible to determine whether or not the learned value evafsfbg deviates from the required value even when the sub feedback control is not executed. For this reason, according to the present embodiment, there is an advantage that it can be determined earlier whether or not the learning value evafsfbg deviates from the required value.

[Specific Processing in Second Embodiment]
FIG. 7 is a flowchart of a routine executed by the ECU 50 in the present embodiment in order to realize the above function. According to the routine shown in FIG. 7, first, in the air-fuel ratio feedback control systems of both the first system and the second system, it is determined whether or not the learning of the sub feedback control has converged (step 110). Specifically, it is confirmed whether or not a learning completion flag XSFBGEND representing the convergence of learning of both systems is turned on.

  If it is determined in step 110 that the learning of the sub-feedback control of both systems has not converged, it is not necessary to execute the following processing. Therefore, in this case, the processing of this routine is terminated as it is.

  In this embodiment, after the internal combustion engine 10 is started, the ECU 50 performs the time T1 until the output of the first system sub-oxygen sensor 22 is reversed to the rich output, and the output of the second system sub-oxygen sensor 22 is the rich output. The time T2 until it is reversed to each time is counted. If it is determined in step 110 that the learning of the sub-feedback control of both systems has converged, their inversion times T1 and T2 are read, and the absolute value of the difference between them is equal to or greater than the predetermined value β. Is determined (step 112).

  If the difference between the inversion times T1 and T2 of both systems is smaller than the predetermined value β in step 112, it can be determined that the learning value evafsfbg has not deviated from the required value in any system. In this case, since it can be determined that good air-fuel ratio feedback control is being executed in any system, there is no need for relearning. Therefore, in this case, the processing of this routine is terminated as it is.

  On the other hand, if the difference between the inversion times T1 and T2 between the two systems is greater than or equal to the predetermined value β in step 112, it can be determined that the learning value evafsfbg is deviated from the required value in at least one system. In this case, the learning completion flag XSFBGEND is turned off to execute relearning (step 114). When the learning completion flag XSFBGEND is turned off, it is considered that learning has not converged in both systems, so that the restriction on learning is canceled in both systems. As a result, the followability of the learned value evafsfbg is improved, so that the learned value evafsfbg that has deviated can be quickly brought close to the required value. Therefore, the deviation of the air-fuel ratio can be corrected and the air-fuel ratio can be controlled in the vicinity of the target air-fuel ratio, so that deterioration of exhaust emission can be avoided.

  In the second embodiment described above, the inversion times T1 and T2 correspond to the “predetermined index value” in the first invention.

  In the present embodiment, it is determined whether or not the learning value evafsfbg deviates from the required value based on the time until the outputs of the sub oxygen sensors 22 of both systems are reversed to the rich output after the internal combustion engine 10 is started. However, even at times other than such time, the same determination should be made based on the time from when the oxygen storage states of the catalysts 18 of both systems match until the output of the sub oxygen sensor 22 is reversed. Can do.

  For example, when the fuel cut of the internal combustion engine 10 is executed, air containing no fuel flows through the catalyst 18 of both systems, so that both the catalysts 18 are saturated with oxygen. Therefore, after returning from the fuel cut, whether the learning value evafsfbg deviates from the required value is similarly determined based on the time until the outputs of the sub oxygen sensors 22 of both systems are reversed from the lean output to the rich output. can do. Further, when the fuel increase is executed at high load or the like, the rich air-fuel ratio exhaust gas flows through the catalyst 18 of both systems, so that both the catalysts 18 are in an oxygen-deficient state. Therefore, it is similarly determined whether or not the learning value evafsfbg deviates from the required value based on the time until the output of the sub oxygen sensor 22 of both systems is reversed from the rich output to the lean output after the fuel increase is completed. be able to.

It is a figure for demonstrating the system configuration | structure of Embodiment 1 of this invention. It is a figure for demonstrating learning convergence determination conditions. It is a figure for demonstrating learning convergence determination conditions. It is a flowchart of the routine performed in Embodiment 1 of the present invention. It is a figure which shows the change of the output of the sub-oxygen sensor of the 1st system | strain and the 2nd system | strain since the start of an internal combustion engine. It is a figure which shows the change of the output of the sub-oxygen sensor of the 1st system | strain and the 2nd system | strain since the start of an internal combustion engine. It is a flowchart of the routine performed in Embodiment 2 of this invention.

Explanation of symbols

10 internal combustion engine 11 first bank 12 second bank 14 exhaust manifold 16 exhaust passage 18 catalyst 20 main air-fuel ratio sensor 22 sub oxygen sensor 24 fuel injection device 50 ECU

Claims (3)

An air-fuel ratio control apparatus for an internal combustion engine having a plurality of cylinder groups and including an air-fuel ratio feedback control system for each cylinder group,
The air-fuel ratio feedback control system for each cylinder group is:
An exhaust passage,
A catalyst disposed in the exhaust passage;
A main exhaust gas sensor disposed upstream of the catalyst;
A sub exhaust gas sensor disposed downstream of the catalyst;
Main feedback means for performing main feedback control of the air-fuel ratio based on the output of the main exhaust gas sensor;
Sub-feedback learning means including means for calculating a sub-feedback correction value based on the output of the sub-exhaust gas sensor, and means for calculating a learning value for correcting a stationary error component based on the sub-feedback correction value When,
Convergence determination means for determining convergence of learning by the sub-feedback learning means;
Learning limiting means for limiting subsequent learning when it is determined that the learning has converged;
Have
A comparing means for comparing a predetermined index value calculated based on an output of the sub exhaust gas sensor between the cylinder groups when it is determined that the learning has converged in each of the cylinder groups;
Restriction release means for releasing the restriction by the learning restriction means when the difference between the index values between the cylinder groups exceeds a predetermined value;
An air-fuel ratio control apparatus for an internal combustion engine, comprising:   2. The air-fuel ratio control apparatus for an internal combustion engine according to claim 1, wherein the index value is an integral value of a deviation between an output value of the sub exhaust gas sensor and a control target value thereof.   The index value is a time until the output of the sub exhaust gas sensor changes from a lean output to a rich output when the catalyst is saturated with oxygen, or when the catalyst is in an oxygen-deficient state. 2. The air-fuel ratio control apparatus for an internal combustion engine according to claim 1, wherein the time until the output of the exhaust gas sensor changes from a rich output to a lean output.

Images