Air–fuel ratio control method and device for internal combustion engine

Abstract

This air-fuel ratio control device includes a three-way catalyst (15), an upstream-side air-fuel ratio sensor (19), and a downstream-side air-fuel ratio sensor (20), and controls an air-fuel ratio such that an oxygen storage amount of the three-way catalyst (15) becomes a target oxygen storage amount. A controller (9) compares a downstream-side exhaust air-fuel ratio detected by the downstream-side air-fuel ratio sensor (19) with a lean slice level (LS) and a rich slice level (RS), and when a state farther to the lean side than the lean slice level (LS) continues for a predetermined time, the controller (9) adds a positive value to an air-fuel ratio detection characteristic of the upstream-side air-fuel ratio sensor (19) and updates a learning value. When a state farther to the rich side than the rich slice level (RS) continues for a predetermined time, a negative value is added, and a learning value is updated.

Description

Method and device for controlling air-fuel ratio of an internal combustion engine

[0001] The present invention relates to air-fuel ratio control for an internal combustion engine, which includes an upstream air-fuel ratio sensor and a downstream air-fuel ratio sensor, each of which is provided upstream and downstream of a three-way catalyst and which provide outputs according to the exhaust air-fuel ratio, and which controls the air-fuel ratio based on the upstream exhaust air-fuel ratio detected by the upstream air-fuel ratio sensor, so that the oxygen storage amount of the three-way catalyst becomes a target oxygen storage amount.

[0002] For example, a three-way catalyst, which is an exhaust purification catalyst, can oxidize CO and HC in exhaust gas and reduce NOx. However, to achieve both oxidation and reduction at a high level through catalytic action, the catalyst's ability to store and release oxygen, or its so-called oxygen storage capacity, is important. Therefore, a technique is known in which the oxygen storage capacity of a three-way catalyst is estimated and a target air-fuel ratio is controlled so that this oxygen storage capacity becomes an appropriate target oxygen storage capacity. The exhaust air-fuel ratio upstream of the three-way catalyst is detected as the actual air-fuel ratio by an upstream air-fuel ratio sensor, which is a so-called wide-range air-fuel ratio sensor that obtains an output according to the exhaust air-fuel ratio. Then, for example, the fuel injection amount is feedback-controlled so that this actual air-fuel ratio approaches the target air-fuel ratio.

[0003] In such control, the wide-range air-fuel ratio sensor used as the upstream air-fuel ratio sensor may have a deviation in the relationship between the actual air-fuel ratio and the sensor output (output air-fuel ratio), i.e., in the air-fuel ratio detection characteristics, due to deterioration over time, etc. Therefore, in order to correct this deviation, some means is used to detect the deviation and a learning correction is performed.

[0004] Patent Document 1 discloses a technology in which an oxygen sensor whose output changes suddenly at the stoichiometric air-fuel ratio is disposed downstream of a three-way catalyst, and when the oxygen sensor detects a rich air-fuel ratio while the upstream air-fuel ratio sensor indicates the stoichiometric air-fuel ratio, the correction value of the learning correction is increased by a predetermined amount, and conversely, when the oxygen sensor detects a lean air-fuel ratio while the upstream air-fuel ratio sensor indicates the stoichiometric air-fuel ratio, the correction value of the learning correction is decreased by a predetermined amount.

[0005] However, the rich / lean reversal of the exhaust air-fuel ratio downstream of the three-way catalyst occurs slowly due to the oxygen storage capacity of the three-way catalyst, and the rich or lean state continues for a certain period of time. In the configuration of Patent Document 1, the learning correction value continues to increase while the downstream exhaust air-fuel ratio is rich, and conversely, the learning correction value continues to decrease while the downstream exhaust air-fuel ratio is lean. Therefore, the air-fuel ratio detection characteristic of the upstream air-fuel ratio sensor is prone to overcorrection.

[0006] Japanese Patent Application Laid-Open No. 2003-138964

[0007] This invention provides an air-fuel ratio control method for an internal combustion engine, which is provided with an upstream air-fuel ratio sensor and a downstream air-fuel ratio sensor, each of which provides an output corresponding to the exhaust air-fuel ratio, upstream and downstream of a three-way catalyst in the exhaust passage of the internal combustion engine, and which controls the air-fuel ratio so that the oxygen storage amount of the three-way catalyst becomes a target oxygen storage amount based on the upstream exhaust air-fuel ratio detected by the upstream air-fuel ratio sensor, by setting a lean slice level indicating that the downstream exhaust air-fuel ratio detected by the downstream air-fuel ratio sensor is leaner and a rich slice level indicating that the downstream exhaust air-fuel ratio is richer, and when the downstream exhaust air-fuel ratio remains leaner than the lean slice level or richer than the rich slice level for a predetermined period of time, a learning correction is made to the air-fuel ratio detection characteristics of the upstream air-fuel ratio sensor.

[0008] For example, if the downstream exhaust air-fuel ratio downstream of the three-way catalyst remains leaner than the lean slice level for a certain period of time, it can be assumed that the actual exhaust air-fuel ratio upstream of the three-way catalyst (actual upstream exhaust air-fuel ratio) is leaner, regardless of the oxygen storage capacity of the three-way catalyst. Similarly, if the downstream exhaust air-fuel ratio remains richer than the rich slice level for a certain period of time, it can be assumed that the actual exhaust air-fuel ratio upstream of the three-way catalyst (actual upstream exhaust air-fuel ratio) is richer. Therefore, if the state of remaining leaner than the lean slice level or richer than the rich slice level continues for a predetermined period of time, it can be determined that the output air-fuel ratio of the upstream exhaust air-fuel ratio sensor is deviated from the actual exhaust air-fuel ratio, and a learning correction is performed on the air-fuel ratio detection characteristics of the upstream air-fuel ratio sensor.

[0009] In addition, the second invention sets a lean slice level indicating that the downstream exhaust air-fuel ratio detected by the downstream air-fuel ratio sensor is lean, and a rich slice level indicating that the downstream exhaust air-fuel ratio is lean, and when the downstream exhaust air-fuel ratio crosses the lean slice level to the lean side or crosses the rich slice level to the rich side and changes to the rich side a predetermined number of times in succession, a learning correction is made to the air-fuel ratio detection characteristics of the upstream air-fuel ratio sensor.

[0010] For example, if the output air-fuel ratio of the upstream exhaust air-fuel ratio relative to the actual exhaust air-fuel ratio deviates slightly toward the rich side, and the actual exhaust air-fuel ratio (actual upstream exhaust air-fuel ratio) becomes lean as a result of air-fuel ratio control, causing the downstream exhaust air-fuel ratio to exhibit a value near the lean slice level, a phenomenon occurs in which the lean slice level is crossed by a relatively small change in air-fuel ratio. If such lean changes occur multiple times in succession, it can be determined that the output air-fuel ratio of the upstream exhaust air-fuel ratio sensor relative to the actual exhaust air-fuel ratio is deviating, albeit by a relatively small amount. The same applies when the output air-fuel ratio of the upstream exhaust air-fuel ratio relative to the actual exhaust air-fuel ratio deviates toward the lean side. Therefore, if the downstream exhaust air-fuel ratio crosses the lean slice level and changes to the lean side, or if the downstream exhaust air-fuel ratio crosses the rich slice level and changes to the rich side, a predetermined number of times in succession, a learning correction is performed on the air-fuel ratio detection characteristics of the upstream air-fuel ratio sensor.

[0011] Thus, according to the present invention, deviations in the air-fuel ratio detection characteristics of the upstream air-fuel ratio sensor can be accurately grasped, and the learning correction of the air-fuel ratio detection characteristics can be performed without causing overcorrection.

[0012] 1 is an explanatory diagram showing the configuration of an internal combustion engine equipped with an air-fuel ratio control device of one embodiment; a flowchart showing the processing flow of learning correction of a first embodiment; a time chart showing the operation of the first embodiment when the downstream exhaust air-fuel ratio deviates to the lean side; a time chart showing the operation of the first embodiment when the downstream exhaust air-fuel ratio deviates to the rich side; a flowchart showing the processing flow of learning correction of a second embodiment; a time chart showing the operation of the second embodiment when the downstream exhaust air-fuel ratio deviates to the lean side; a time chart showing the operation of the second embodiment when the downstream exhaust air-fuel ratio deviates to the rich side;

[0013] An embodiment of the present invention will now be described in detail with reference to the drawings. FIG. 1 is an explanatory diagram showing a schematic configuration of an internal combustion engine 1 according to an embodiment of the present invention. The internal combustion engine 1 according to the embodiment is a four-stroke, spark-ignition internal combustion engine (a so-called gasoline engine), and each cylinder is provided with an intake valve 2, an exhaust valve 3, and an ignition plug 4. The illustrated example is configured as a port-injection internal combustion engine, and a fuel injection valve 5 is disposed so as to inject fuel toward an intake port 6. However, the internal combustion engine may also be configured as a direct-injection type, in which fuel is injected directly into a cylinder.

[0014] An electronically controlled throttle valve 10, the opening of which is controlled by a control signal from an engine controller 9, is disposed upstream of a collector section 8 of an intake passage 7 connected to the intake port 6 of each cylinder. An air flow meter 11, which detects the amount of intake air, is disposed upstream of the throttle valve 10, and an air cleaner 12 is disposed further upstream.

[0015] The exhaust ports 13 of each cylinder are joined together to form a single exhaust passage 14, which is provided with a three-way catalyst 15 as an exhaust purification catalyst for purifying the exhaust gas. The three-way catalyst 15 is, for example, a monolithic ceramic catalyst in which a catalytic layer containing catalytic metal is coated on the surface of a monolithic ceramic body having fine passages formed therein. The three-way catalyst 15 may further include a downstream catalyst (a so-called underfloor catalyst) arranged in series.

[0016] An upstream air-fuel ratio sensor 19 is disposed in the exhaust passage 14 on the inlet side of the three-way catalyst 15, i.e., upstream of the three-way catalyst 15, for detecting the exhaust air-fuel ratio of the exhaust gas emitted by the internal combustion engine 1 (in other words, the exhaust air-fuel ratio of the exhaust gas flowing into the three-way catalyst 15). This upstream air-fuel ratio sensor 19 is a so-called wide-range air-fuel ratio sensor that obtains an output corresponding to the exhaust air-fuel ratio. Furthermore, a downstream air-fuel ratio sensor 20 is disposed on the outlet side or downstream of the three-way catalyst 15, for detecting the exhaust air-fuel ratio of the exhaust gas flowing out from the three-way catalyst 15. Like the upstream air-fuel ratio sensor 19, the downstream air-fuel ratio sensor 20 is also a wide-range air-fuel ratio sensor that obtains an output corresponding to the exhaust air-fuel ratio.

[0017] Detection signals from the air-fuel ratio sensors 19, 20 and the air flow meter 11 are input to the engine controller 9. Detection signals from a number of sensors, such as a crank angle sensor 21 for detecting the engine speed, a water temperature sensor 22 for detecting the coolant temperature, and an accelerator position sensor 23 for detecting the amount of depression of the accelerator pedal operated by the driver, are also input to the engine controller 9. Based on these input signals, the engine controller 9 optimally controls the amount and timing of fuel injection by the fuel injection valve 5, the ignition timing by the spark plug 4, the opening of the throttle valve 10, etc.

[0018] As one of various controls of the internal combustion engine 1, the engine controller 9 performs air-fuel ratio control to maintain the oxygen storage capacity of the three-way catalyst 15 at a target oxygen storage capacity in order to optimize the exhaust gas purification performance of the three-way catalyst 15. In the air-fuel ratio control, the fuel injection amount is feedback-controlled (e.g., PID control) so that the exhaust air-fuel ratio detected by the upstream air-fuel ratio sensor 19 (hereinafter referred to as the upstream exhaust air-fuel ratio) conforms to the target air-fuel ratio. Here, the target air-fuel ratio is calculated so that the oxygen storage capacity of the three-way catalyst 15 estimated from the upstream exhaust air-fuel ratio matches the target oxygen storage capacity. For example, if the estimated oxygen storage capacity is greater than the target oxygen storage capacity, the target air-fuel ratio is set richer than the stoichiometric air-fuel ratio. Conversely, if the estimated oxygen storage capacity is smaller than the target oxygen storage capacity, the target air-fuel ratio is set leaner than the stoichiometric air-fuel ratio. Therefore, the oxygen storage capacity of the three-way catalyst 15 is basically maintained near the target oxygen storage capacity. The oxygen storage capacity can be expressed as a percentage, with the maximum oxygen storage capacity of the three-way catalyst 15 being 100%. The target oxygen storage capacity is set to an appropriate value, for example, around 50%.

[0019] Furthermore, in order to improve the accuracy of air-fuel ratio control, the engine controller 9 performs learning correction of the air-fuel ratio detection characteristics of the upstream air-fuel ratio sensor 19. In other words, the relationship between the actual air-fuel ratio and the sensor output, i.e., the air-fuel ratio detection characteristics, of the upstream air-fuel ratio sensor 19 may deviate due to initial individual variations as well as deterioration over time, etc. Therefore, learning correction of the air-fuel ratio detection characteristics using the downstream air-fuel ratio sensor 20 is performed to correct this deviation.

[0020] In other words, if the output characteristic of the upstream air-fuel ratio sensor 19 relative to the actual air-fuel ratio shifts to the rich side or the lean side, the upstream exhaust air-fuel ratio on the upstream side of the three-way catalyst 15 will shift conversely to the lean side or the rich side as a result of air-fuel ratio control. The three-way catalyst 15 has oxygen storage capacity, but when the oxygen storage amount is close to saturation, if the upstream exhaust air-fuel ratio becomes lean, the downstream exhaust air-fuel ratio will also become lean. Conversely, when the oxygen storage amount is close to zero, if the upstream exhaust air-fuel ratio becomes rich, the downstream exhaust air-fuel ratio will also become rich.

[0021] In the first embodiment, a lean slice level LS and a rich slice level RS are set for the downstream exhaust air-fuel ratio detected by the downstream air-fuel ratio sensor 20. The lean slice level LS is slightly leaner than the air-fuel ratio equivalent to the stoichiometric air-fuel ratio, indicating that the oxygen storage amount of the three-way catalyst 15 is greater than the target oxygen storage amount and the downstream exhaust air-fuel ratio is biased toward the lean side. The rich slice level RS is slightly richer than the air-fuel ratio equivalent to the stoichiometric air-fuel ratio, indicating that the oxygen storage amount of the three-way catalyst 15 is less than the target oxygen storage amount and the downstream exhaust air-fuel ratio is biased toward the rich side.

[0022] In the first embodiment, if the downstream exhaust air-fuel ratio remains leaner than the lean slice level LS for a predetermined period of time, a learning correction is made to the air-fuel ratio detection characteristics of the upstream air-fuel ratio sensor 19 (i.e., the output value is shifted toward the lean side). Similarly, if the downstream exhaust air-fuel ratio remains richer than the rich slice level RS for a predetermined period of time, a learning correction is made to the air-fuel ratio detection characteristics of the upstream air-fuel ratio sensor 19 (i.e., the output value is shifted toward the rich side). Here, the "predetermined period of time" corresponds to a duration during which the downstream exhaust air-fuel ratio detected by the downstream air-fuel ratio sensor 20 can be considered to substantially indicate the upstream exhaust air-fuel ratio. In other words, if the upstream exhaust air-fuel ratio is leaner on average (averaged over time) due to a deviation in the air-fuel ratio detection characteristics of the upstream air-fuel ratio sensor 19, the oxygen storage capacity of the three-way catalyst 15 approaches saturation, and the downstream exhaust air-fuel ratio begins to indicate a lean value. If this state continues for a certain period of time, it can be assumed that the value of the downstream exhaust air-fuel ratio indicated by the downstream air-fuel ratio sensor 20 is substantially equal to the upstream exhaust air-fuel ratio. The same applies when the upstream exhaust air-fuel ratio is biased toward the rich side on average (averaged over time) due to a deviation in the air-fuel ratio detection characteristics of the upstream air-fuel ratio sensor 19. Therefore, the deviation of the downstream exhaust air-fuel ratio from the stoichiometric air-fuel ratio in this state corresponds to a deviation in the air-fuel ratio detection characteristics of the upstream air-fuel ratio sensor 19. Therefore, it is possible to make an appropriate learning correction (update the learning value) corresponding to the magnitude of the deviation in the air-fuel ratio detection characteristics.

[0023] FIG. 2 is a flowchart showing the flow of the learning correction process in the first embodiment. For simplicity, the flowchart in FIG. 2 shows both the rich-side learning correction and the lean-side learning correction together. In the first step 1, it is determined whether a learning permission condition is met. The learning permission condition includes the downstream air-fuel ratio sensor 20 being activated, air-fuel ratio feedback control being in progress, and no learning prohibition request being made. If learning is permitted, the process proceeds to step 2, where it is determined whether the downstream exhaust air-fuel ratio detected by the downstream air-fuel ratio sensor 20 is leaner than the lean slice level LS or richer than the rich slice level RS. If the answer is NO, i.e., if the downstream exhaust air-fuel ratio is between the lean slice level LS and the rich slice level RS (this range will be referred to as the "stoichiometric range" for convenience), the learning value is not updated.

[0024] If the determination in step 2 is YES, the process proceeds to step 3, where it is determined whether the downstream exhaust air-fuel ratio has been leaner than the lean slice level LS or richer than the rich slice level RS for a predetermined time. If NO, the processes of steps 2 and 3 are repeated. If the determination in step 3 is YES, the process proceeds to step 4, where the learned value is updated. Specifically, a learned value corresponding to the degree of deviation of the downstream exhaust air-fuel ratio from the stoichiometric air-fuel ratio (e.g., the median value between the lean slice level LS and the rich slice level RS) is calculated, and the calculated learned value is added to the existing learned value to update the learned value. For example, if the larger the learned value, the more the detected air-fuel ratio of the upstream air-fuel ratio sensor 19 changes to the leaner side, the learned value added when the state leaner than the lean slice level LS has continued for a predetermined time is a positive value, and the learned value added when the state richer than the rich slice level RS has continued for a predetermined time is a negative value. The learned value is determined based on the magnitude of deviation of the downstream exhaust air-fuel ratio from the stoichiometric air-fuel ratio after a predetermined time has elapsed, using a table that assigns the magnitude of deviation from the stoichiometric air-fuel ratio to a corresponding learned value. Note that the learned value may be determined using the magnitude of deviation from the lean slice level LS or the rich slice level RS as a parameter, rather than the magnitude of deviation from the stoichiometric air-fuel ratio.

[0025] 3 is a time chart showing the operation of the learning correction in the first embodiment, particularly showing the operation when the downstream exhaust air-fuel ratio deviates to the lean side. From top to bottom, the diagram shows (a) a learning prohibition determination flag indicating that the learning correction is prohibited, (b) a learning permission determination flag indicating that the learning correction is permitted, (c) a lean determination flag indicating that the downstream exhaust air-fuel ratio is leaner than the lean slice level LS, (e) a learning update determination flag indicating that it is time to update the learning value, (f) the output of the downstream air-fuel ratio sensor 20, i.e., the downstream exhaust air-fuel ratio, (g) a learning timer (learning counter) that measures the above-mentioned predetermined time, and (h) a learning value.

[0026] In the example of Fig. 3, the learning prohibition determination flag is turned off at time t1, and learning correction is permitted at time t3. In the illustrated example, the downstream exhaust air-fuel ratio becomes leaner than the lean slice level LS at time t2, before time t3, but the learning timer starts at time t3 when the learning permission determination flag is turned on. At time t4, when the downstream exhaust air-fuel ratio remains leaner than the lean slice level LS for a predetermined period of time, the learning update determination flag is turned on, and the learning value is updated as shown in column (h). For example, the learning value is updated by adding a positive learning value of a magnitude calculated by multiplying the difference between the downstream exhaust air-fuel ratio at time t4 and the stoichiometric air-fuel ratio by a coefficient to the existing learning value.

[0027] As a result of such learning correction, the air-fuel ratio output of the upstream air-fuel ratio sensor 19 approaches the actual air-fuel ratio, so that the amount of oxygen stored in the three-way catalyst 15 under air-fuel ratio feedback control approaches the target oxygen storage amount, and the downstream exhaust air-fuel ratio usually falls within the stoichiometric range.

[0028] It should be noted that if the downstream exhaust air-fuel ratio becomes leaner than the lean slice level LS after the learning permission determination flag is turned on, the learning timer starts at that point.

[0029] 4 is a time chart illustrating the operation when the downstream exhaust air-fuel ratio deviates to the rich side. Column (d) shows a rich determination flag, instead of the lean determination flag (c) in FIG. 3, which indicates that the downstream exhaust air-fuel ratio is richer than the rich slice level RS.

[0030] In the example of Fig. 4, the learning prohibition determination flag is turned off at time t1, and learning correction is permitted at time t3. In the illustrated example, the downstream exhaust air-fuel ratio becomes richer than the rich slice level RS at time t2, before time t3, but the learning timer starts at time t3 when the learning permission determination flag is turned on. At time t4, when the downstream exhaust air-fuel ratio remains richer than the rich slice level RS for a predetermined period of time, the learning update determination flag is turned on, and the learning value is updated as shown in column (h). For example, the learning value is updated by adding a negative learning value of a magnitude calculated by multiplying the difference between the downstream exhaust air-fuel ratio at time t4 and the stoichiometric air-fuel ratio by a coefficient to the existing learning value.

[0031] Thus, according to the learning correction of the first embodiment, when the air-fuel ratio detection characteristic of the upstream air-fuel ratio sensor 19 deviates to an extent that it exceeds the lean slice level LS or the rich slice level RS, a relatively large learning value is applied at one time, so that the deviation can be corrected at high speed.

[0032] Next, a second embodiment of the present invention will be described. In the second embodiment, as in the first embodiment, a lean slice level LS and a rich slice level RS are set for the downstream exhaust air-fuel ratio detected by the downstream air-fuel ratio sensor 20. In the second embodiment, the learning value of the upstream air-fuel ratio sensor 19 is updated when the downstream exhaust air-fuel ratio crosses the lean slice level LS and changes from the stoichiometric range to the lean side, or crosses the rich slice level RS and changes from the stoichiometric range to the rich side, each occurring twice in succession.

[0033] FIG. 5 is a flowchart showing the processing flow of the second embodiment. For simplicity of explanation, the flowchart in FIG. 5 shows both the rich-side learning correction and the lean-side learning correction together. In the first step 11, as in step 1 in FIG. 2, it is determined whether the learning permission condition is met. If learning is permitted, the process proceeds to step 12, where it is determined whether the downstream exhaust air-fuel ratio detected by the downstream air-fuel ratio sensor 20 has changed from the stoichiometric range to the lean side across the lean slice level LS or changed to the rich side across the rich slice level RS. If NO, i.e., the downstream exhaust air-fuel ratio remains in the stoichiometric range, the determination in step 12 is repeated.

[0034] If the determination in step 12 is YES, that is, if the downstream exhaust air-fuel ratio has changed from the stoichiometric range to the lean side across the lean slice level LS or changed to the rich side across the rich slice level RS, the process proceeds to step 13, where it is repeatedly determined whether the downstream exhaust air-fuel ratio has returned to the stoichiometric range. In other words, after the first lean or rich change, the process waits for the downstream exhaust air-fuel ratio to return to the stoichiometric range.

[0035] When the downstream exhaust air-fuel ratio falls within the stoichiometric range and the determination in step 13 is YES, the process proceeds from step 13 to step 14, where it is determined whether a second lean or rich change has occurred in the same direction as in step 12. If the determination in step 14 is NO, the process proceeds to step 15, where it is determined whether a rich or lean change has occurred in the opposite direction to that in step 12. If the determination in step 15 is NO, this means that the downstream exhaust air-fuel ratio remains within the stoichiometric range, and the process returns to the determination in step 14.

[0036] On the other hand, if the determination in step 15 is YES, the process returns to the determination in step 12. In other words, in this case, the history of the first lean or rich change is canceled.

[0037] If the determination in step 14 is YES, that is, if it is determined that a second lean or rich change in the same direction as in step 12 has occurred, this means that two consecutive lean or rich changes have occurred, so the process proceeds from step 14 to step 16, where the learned value is updated. Here, the learned value is updated by adding a fixed learned value (positive or negative value) corresponding to the difference between the lean slice level LS or rich slice level RS and the stoichiometric air-fuel ratio to the existing learned value.

[0038] 6 is a time chart showing the operation of the learning correction in the second embodiment, and shows the operation when the downstream exhaust air-fuel ratio deviates to the lean side. In particular, it shows an example in which the history of the first lean change is canceled by a subsequent rich change.

[0039] From top to bottom, the diagram shows: (a) a learning prohibition judgment flag indicating that learning correction is prohibited; (b) a learning permission judgment flag indicating that learning correction is permitted; (c) a lean judgment flag indicating that the downstream exhaust air-fuel ratio is leaner than the lean slice level LS; (d) a rich judgment flag indicating that the downstream exhaust air-fuel ratio is richer than the rich slice level RS; (e) a learning update judgment flag indicating that it is time to update the learning value; (f) the output of the downstream air-fuel ratio sensor 20, i.e., the downstream exhaust air-fuel ratio; (g) a learning timer (learning counter) that measures the duration of the lean judgment flag and the rich judgment flag; (j) an initialization flag that cancels the history of the first lean change or rich change; and (h) a learning value.

[0040] 6, at time t11, the downstream exhaust air-fuel ratio becomes leaner than the lean slice level LS. This is the first lean change, but then at time t12, it returns to the stoichiometric range, and at time t13, the downstream exhaust air-fuel ratio changes from the stoichiometric range to the richer side than the rich slice level RS. This rich change cancels the count of the first lean change.

[0041] Thereafter, at time t14, the downstream exhaust air-fuel ratio enters the stoichiometric range, and at time t15, it becomes leaner than the lean slice level LS. This is the first lean change. Then, after temporarily returning to the stoichiometric range at time t16, a second lean change occurs at time t17, causing the learning update determination flag to be turned on, and the learning value is updated as shown in column (h). As described above, the learning value is updated by adding a positive learning value of a certain magnitude to the existing learning value.

[0042] For example, there may be a case where the deviation of the air-fuel ratio detection characteristic of the upstream air-fuel ratio sensor 19 is relatively small, and the downstream exhaust air-fuel ratio, which indicates a value approximate to the upstream exhaust air-fuel ratio, remains near the lean slice level LS on average (averaged over time). In such a case, some factor may cause consecutive lean changes that cross the lean slice level LS from the stoichiometric range to the lean side. Therefore, by updating the learning value when two consecutive lean changes occur, it is possible to accurately learn and correct the relatively small deviation of the air-fuel ratio detection characteristic.

[0043] 7 is a time chart illustrating the operation of the second embodiment when the downstream exhaust air-fuel ratio deviates to the rich side. In particular, it shows an example in which the history of the first rich change is canceled by a lean change during the process.

[0044] 7, at time t21, the downstream exhaust air-fuel ratio becomes richer than the rich slice level RS. This is the first rich change, but then at time t22, it returns to the stoichiometric range, and then at time t23, the downstream exhaust air-fuel ratio changes from the stoichiometric range to the leaner side than the lean slice level LS. This lean change cancels the count of the first rich change.

[0045] Thereafter, at time t24, the downstream exhaust air-fuel ratio enters the stoichiometric range, and at time t25, it becomes richer than the rich slice level RS. This is a new first rich change. Then, after temporarily returning to the stoichiometric range at time t26, a second rich change occurs at time t27, causing the learning update determination flag to be turned on, and the learning value is updated as shown in column (h). As described above, the learning value is updated by adding a negative learning value of a certain magnitude to the existing learning value.

[0046] According to the learning correction of the second embodiment, it is possible to learn and correct a relatively small deviation in the air-fuel ratio detection characteristic of the upstream air-fuel ratio sensor 19 compared to the first embodiment.

[0047] Although the learning correction of the first embodiment and the learning correction of the second embodiment have been described separately above, these two different types of learning correction can be used simultaneously. For example, when the air-fuel ratio detection characteristic of the upstream air-fuel ratio sensor 19 is shifted to the rich side, if the downstream exhaust air-fuel ratio remains leaner than the lean slice level LS for a predetermined period of time as in the first embodiment, the learning value is updated as described in the first embodiment (this is referred to as the first learning correction). For example, the learning timer in (g) of FIGS. 6 and 7 measures the duration of this lean state. On the other hand, if the downstream exhaust air-fuel ratio does not remain leaner than the lean slice level LS for a predetermined period of time but crosses the lean slice level LS from the stoichiometric range multiple times (e.g., twice) in succession, the learning value is updated as described in the second embodiment (this is referred to as the second learning correction).

[0048] In the first learning correction, a learning value corresponding to the difference between the stoichiometric air-fuel ratio and the downstream exhaust air-fuel ratio is added, while in the second learning correction, a learning value of a fixed magnitude corresponding to the difference between the lean slice level LS and the stoichiometric air-fuel ratio is added. Therefore, the learning value obtained by the first learning correction is generally greater than the learning value obtained by the second learning correction. Therefore, if the air-fuel ratio detection characteristic of the upstream air-fuel ratio sensor 19 deviates relatively significantly due to some factor, the first learning correction corrects the air-fuel ratio detection characteristic relatively significantly. If a relatively small deviation in the air-fuel ratio detection characteristic remains, the second learning correction corrects the air-fuel ratio detection characteristic by a relatively small magnitude. This allows for responsive learning correction of deviations in the air-fuel ratio detection characteristic while also achieving fine learning correction.

[0049] While one embodiment of the present invention has been described above, the present invention is not limited to the above embodiment and various modifications are possible. For example, in the second embodiment, the learning correction is performed after two lean or rich changes, but the learning correction may be performed after three or more lean or rich changes.

[0050] In a preferred embodiment, the lean slice level LS and rich slice level RS related to the downstream exhaust air-fuel ratio can be set to an air-fuel ratio level equal to a threshold value for resetting the estimated value in the estimation calculation of the oxygen storage amount. That is, the estimated value of the oxygen storage amount is reset to its respective initial value based on the lean determination flag and rich determination flag shown in Figures 3 and 4. However, the lean slice level LS and rich slice level RS in the present invention may be set to a level different from the air-fuel ratio level provided for resetting the estimated value of the oxygen storage amount.

Claims

1. An air-fuel ratio control method for an internal combustion engine, which is provided with an upstream air-fuel ratio sensor and a downstream air-fuel ratio sensor, each of which provides an output according to the exhaust air-fuel ratio, upstream and downstream of a three-way catalyst in the exhaust passage of the internal combustion engine, and which controls the air-fuel ratio based on the upstream exhaust air-fuel ratio detected by the upstream air-fuel ratio sensor so that the oxygen storage amount of the three-way catalyst becomes a target oxygen storage amount, the method comprising: setting, with respect to the downstream exhaust air-fuel ratio detected by the downstream air-fuel ratio sensor, a lean slice level indicating that the downstream exhaust air-fuel ratio is leaner and a rich slice level indicating that the downstream exhaust air-fuel ratio is richer; and performing learning correction of the air-fuel ratio detection characteristics of the upstream air-fuel ratio sensor when the downstream exhaust air-fuel ratio remains leaner than the lean slice level or richer than the rich slice level for a predetermined period of time.

2. The air-fuel ratio control method for an internal combustion engine according to claim 1, wherein the learning correction is performed by adding a learning value whose magnitude corresponds to the degree of deviation of the downstream exhaust air-fuel ratio from the stoichiometric air-fuel ratio.

3. An air-fuel ratio control method for an internal combustion engine, which is provided with an upstream air-fuel ratio sensor and a downstream air-fuel ratio sensor, each of which provides an output according to the exhaust air-fuel ratio, upstream and downstream of a three-way catalyst in the exhaust passage of the internal combustion engine, and which controls the air-fuel ratio so that the oxygen storage amount of the three-way catalyst becomes a target oxygen storage amount based on the upstream exhaust air-fuel ratio detected by the upstream air-fuel ratio sensor, the method comprising: setting, with respect to the downstream exhaust air-fuel ratio detected by the downstream air-fuel ratio sensor, a lean slice level indicating that the downstream exhaust air-fuel ratio is lean, and a rich slice level indicating that the downstream exhaust air-fuel ratio is lean, and when the downstream exhaust air-fuel ratio crosses the lean slice level to become lean, or when the downstream exhaust air-fuel ratio crosses the rich slice level to become rich, each of which occurs a predetermined number of times in succession, the method performs learning correction of the air-fuel ratio detection characteristics of the upstream air-fuel ratio sensor.

4. The air-fuel ratio control method for an internal combustion engine according to claim 3, wherein the learning correction is performed by adding a learning value of a fixed magnitude.

5. The air-fuel ratio control method for an internal combustion engine according to claim 3, wherein when the rich change occurs after the lean change, the number of lean changes counted up to that point is cancelled, and when the lean change occurs after the rich change, the number of rich changes counted up to that point is cancelled.

6. The air-fuel ratio control method for an internal combustion engine according to claim 1, further comprising the step of: performing a learning correction of the air-fuel ratio detection characteristics of the upstream air-fuel ratio sensor when the downstream exhaust air-fuel ratio crosses the lean slice level to the lean side or when the downstream exhaust air-fuel ratio crosses the rich slice level to the rich side a predetermined number of times in succession.

7. An air-fuel ratio control device for an internal combustion engine comprising: a three-way catalyst provided in an exhaust passage of the internal combustion engine; an upstream air-fuel ratio sensor and a downstream air-fuel ratio sensor provided respectively upstream and downstream of the three-way catalyst, which provide an output according to the exhaust air-fuel ratio; and a controller that controls the air-fuel ratio based on the upstream exhaust air-fuel ratio detected by the upstream air-fuel ratio sensor so that the oxygen storage amount of the three-way catalyst becomes a target oxygen storage amount, wherein the controller compares the downstream exhaust air-fuel ratio with a lean slice level that indicates that the downstream exhaust air-fuel ratio detected by the downstream air-fuel ratio sensor is leaner and a rich slice level that indicates that the downstream exhaust air-fuel ratio is richer, which are set for the downstream exhaust air-fuel ratio detected by the downstream air-fuel ratio sensor; and when the downstream exhaust air-fuel ratio remains leaner than the lean slice level or richer than the rich slice level for a predetermined period of time, the controller performs learning correction of the air-fuel ratio detection characteristics of the upstream air-fuel ratio sensor.

8. An air-fuel ratio control device for an internal combustion engine comprising: a three-way catalyst provided in an exhaust passage of the internal combustion engine; an upstream air-fuel ratio sensor and a downstream air-fuel ratio sensor provided respectively upstream and downstream of the three-way catalyst, which provide an output according to the exhaust air-fuel ratio; and a controller that controls the air-fuel ratio based on the upstream exhaust air-fuel ratio detected by the upstream air-fuel ratio sensor so that the oxygen storage amount of the three-way catalyst becomes a target oxygen storage amount, wherein the controller compares the downstream exhaust air-fuel ratio with a lean slice level that indicates that the downstream exhaust air-fuel ratio detected by the downstream air-fuel ratio sensor and a rich slice level that indicates that the downstream exhaust air-fuel ratio is biased toward the lean side and a rich slice level that indicates that the downstream exhaust air-fuel ratio is biased toward the rich side, which are set for the downstream exhaust air-fuel ratio detected by the downstream air-fuel ratio sensor, and performs learning correction of the air-fuel ratio detection characteristics of the upstream air-fuel ratio sensor when the downstream exhaust air-fuel ratio crosses the lean slice level to change to the lean side or crosses the rich slice level to change to the rich side a predetermined number of times in succession.

Images