JPH03258944A - Air-fuel ratio control device for engine - Google Patents

Abstract

PURPOSE:To eliminate a fine error of a air-fuel ratio different per the operated condition by processing the output of an upstream side air-fuel ratio sensor to get a mean value, and comparing it with a learning value of a slice level stored correspondently to a small area to judge a reverse of the air-fuel ratio. CONSTITUTION:The output of an upstream side air-fuel ratio sensor 34 is processed to get a mean value, and is read out from a means 38 in response to a result of a judgement by a means 36 for judging a small area of the real operated condition. The read out value is compaired with a learning value MSL of a slice level corresponding to the target air-fuel ratio of a correspondent small area of a storage means 36, and a means 39 judges that the air-fuel ratio is reversed or not, and a computing means 49 computes an air-fuel ratio feed back correction quantity in response to a result of that judgement. This learning value MSL is renewed by a renewing means 46 in response to a result of rich and lean judgement of the output of a downstream side air-fuel ratio sensor 44 to be judged by a means 45. Fine error of an air-fuel ratio different per the operated condition is thereby eliminated to improve the control accuracy.

Description

DETAILED DESCRIPTION OF THE INVENTION (Industrial Field of Application) The present invention is directed to devices that perform feedback control on the air-fuel ratio of an engine, particularly those that incorporate a learning function.

(Prior art) # elemental sensors (0) are installed upstream and downstream of the catalytic converter.
There is a device with a so-called double O2 sensor system (see US Pat. No. 3,939,654).

This will be explained with reference to FIG. 16. The figure shows the air-fuel ratio feedback correction coefficient α based on the upstream side 02 sensor output VFO.
This is a routine for calculating , and is performed at predetermined intervals.

S1 checks whether the air-fuel ratio feedback control conditions (abbreviated as 02F/BJ in the figure, the same in Figures 5 and 11) by the upstream 02 sensor are satisfied, and if so, the S2 Proceed to.

For example, when the cooling water temperature Tw is below a predetermined value, at startup,
Immediately after starting, during fuel increase for warm-up, when the output signal of the upstream 02 sensor has never inverted, and when fuel is cut, the feedback control conditions are not satisfied. Fuel ratio feedback control conditions are met.

In S2, the slice level SL corresponding to the buried air-fuel ratio is looked up.

In S3 to S5, sensor output VFO and slice level s4
Based on the comparison, it is determined whether the air-fuel ratio has reversed from the stoichiometric air-fuel ratio. In terms of layout, we use the current comparison results and the previous comparison results to divide the cases into four cases: if VFO≧SL, it is on the rich side, and vice versa, when VFO<SL, it is on the scene side. If it is, it will be cut off.

In steps 86 to S9, an air-fuel ratio feedback correction coefficient α is calculated according to the above determination result. In summary, it is as follows.

(i) It is determined that it is immediately after the change from lean to rich by 1 in S3→S4→S6, and the proportional fiPR is subtracted from α<(a=a−pR). With this, the air-fuel ratio is returned to the lean side in steps.

(ii) In S3 → S4 → s7, it is determined that it is rich this time as well, and the integral IR is subtracted from a<Ca=a-I R)
As a result, the air-fuel ratio is gradually returned to the lower side.

(iii) In S3→S5→s8, it is determined that the state has just changed from rich to lean, and the proportional amount PL is added to a (a=a0P, -). With this, the air-fuel ratio is returned to the rich side in steps.

(iv) In S 3-S 5-39, it is determined that it is lean this time as well, and the integral IL is added to a (a = Q + I
L).

With this, the air-fuel ratio is gradually returned to the rich side.

FIG. 17 shows a routine for correcting the slice level SL based on the downstream fIIIO2 sensor output VRO, which is executed every predetermined time opening.

In S21, the air-fuel ratio feedback control conditions (abbreviated as ra02F/BJ in the figure) by the downstream 02 sensor are determined.

6 and 12) is satisfied, and if the feedback control condition is satisfied, the process proceeds to S22.

In S22, the sensor output VRO and the slice level SL2 corresponding to the stoichiometric air-fuel ratio are set to Jl[L, and if VRO≧SL2, it is determined that the rich side is present, and the process proceeds to S23.
<The reason why the slice level SL2 is provided separately from the above slice level SL, in which it is determined that the 02 This is to take into consideration the fact that the output characteristics of the sensors and the rate of deterioration of the sensors are different.

In S23, upstream lI! 02 Subtract a constant value Δ from the slice level SL to be compared with the sensor output VFO (SL
=SL-Δ), and in S25, a constant value Δ is added to the slice level SL (SL=SL+Δ).

For example, if you perform the operation in S25, the slice level S
L rises from the left to the right in the upper row of FIG. Due to this increase, the time on the lean side changes from TL to T.
L' becomes longer (on the contrary, the time on the rich side is TR
(from TR' to TR'), the controlled air-fuel ratio is returned to the rich side. In other words, control is performed to shift the air-fuel ratio to the rich side.

Figure 18 shows the fuel injection pulse width T output to the injector.
This is a routine for calculating i[msl, and is executed at every predetermined crank angle (for example, every 360° CA).

In S31, the basic injection pulse width Tp (=-Qa/Ne, where K is a constant) [mS] is determined by referring to a map from the intake air tQ& and the rotational speed Ne.

In S32, the sum CO of 1 and various correction coefficients (for example, water temperature increase correction coefficient KTV) is calculated.

In S33, the fuel injection pulse width Ti to be output to the injector is determined by Ti=Tp-Co·Q+Ts.

Note that Ts[s+s] is the invalid pulse width.

In S34, Ti is set.

(fiM to be solved by the invention) By the way, in the upper part of FIG. 4, if the ratio of the time on the rich side and the time on the lean side is H, then the slice level SL is changed by Δ based on the downstream side 02 sensor output. The reason for increasing the time ratio was to reduce this time ratio H.

However, with such 1IIIr&, if the upstream 02 sensor has good responsiveness, the downstream II
! This results in a situation where the effect of modifying the slice level SL based on the 02 sensor is not as desired.

The reason is as follows. The response waveform of the upstream @o2 sensor shown in the same figure is a raw one (however, the waveform itself is modeled), and the sharper the rise and fall (9), the better the response. This is because, with such a sensor with good responsiveness, even if the slice level SL is greatly changed, the time ratio H does not change much.

Therefore, the range in which the air-fuel ratio can be shifted is narrow, and there is a limit to the ability to absorb variations in air-fuel ratio errors.

In addition, although there is a large response delay in the downstream 02 sensor, the correction fIt of the slice level SL using the downstream 02 sensor output is uniform regardless of the operating conditions. With air-fuel ratio feedback control, it is difficult to accurately eliminate air-fuel ratio errors that vary depending on operating conditions.

This invention was made by focusing on such conventional problems, and the first invention averages the upstream O2 sensor output, compares the averaged sensor output with the slice level, and An important object of the present invention is to provide a device that eliminates air-fuel ratio errors that differ depending on operating conditions by introducing learning values for each small region into this slice level.

The object of the second invention is to provide a device that prevents deterioration of emissions due to sensor deterioration by intensifying the averaging process of the sensor output according to the degree of deterioration of the upstream '1IIO2 sensor.

(Means for Solving the Problems) As shown in FIG. 1(A), the first invention includes sensors 31 and 32 that respectively detect the engine load (for example, intake air amount Qa) and the rotation speed Ne; means 33 for calculating the basic injection amount Tp according to these detected values; a first sensor (for example, 02 sensor) 34 that is interposed in the exhaust passage upstream of the catalytic converter and outputs an output according to the air-fuel ratio; It has a means 35 for averaging the sensor output VFO, and has the same number of addresses as a plurality of small areas divided according to operating conditions determined from at least the engine load and rotation speed, and has a target value corresponding to each small area. A means 36 for storing a slice level learning value MSL corresponding to the air-fuel ratio (stoichiometric air-fuel ratio), a means 37 for determining which of the above-mentioned small regions the current operating condition belongs to, and a small region to which the current operating condition belongs. The learning value MSL of the slice level stored at the address corresponding to
means 38 for reading out the value, a means 39 for determining whether or not the air-fuel ratio has been reversed by comparing the learned value MSL with the averaged sensor output MVFO; means 40 for calculating an air-fuel ratio feedback correction 1a so that the fuel ratio is controlled to be close to (1!ii!2 fuel ratio);
A means 41 for determining lTi, a means 42 for outputting this injection amount Ti to a fuel injection device 43, and a second sensor ( For example, by comparing the sensor output VRO with the second slice level SL2 corresponding to the target air-fuel ratio (theoretical air-fuel ratio), it is determined whether the air-fuel ratio is rich or rich. means 45 and the learned value MSL of the slice level stored at an address corresponding to the small area to which the operating condition at the time of judgment by the small area judging means 37 belongs, and the read learning value MSL is applied to the rich and lean values. The second invention is provided with a means 46 for updating according to the determination result of , as shown in FIG. Sensors 31, 32, means 33 for calculating basic injection 11 T p according to these detected values, and a first sensor (for example, 02 sensor) 34, means 35 for averaging the sensor output VFO,
It has the same number of addresses as multiple small areas divided according to operating conditions determined by at least the engine load and rotation speed, and learns a slice level corresponding to the target air-fuel ratio (theoretical air-fuel ratio) for each small area. means 36 for storing the value MSL;
a means 37 for determining which of the small areas the current operating condition belongs to; a means 38 for reading the learned value MSL of the slice level stored at an address corresponding to the small area to which the current operating condition belongs; means 39 for determining whether or not the air-fuel ratio has been reversed by comparing the learned value MSL with the averaged sensor output MVFO; The air-fuel ratio feedback correction amount a is controlled to be close to
means 40 for calculating the air-fuel ratio feedback correction amount a, and corrects the basic injection amount Tp using the air-fuel ratio feedback correction amount a to adjust the fuel injection ITi.
means 41 for determining the injection amount Ti, means 42 for outputting this injection amount Ti to the fuel injection device 43,
J2 sensor (for example, 02 sensor) 44, the air-fuel ratio is rich by comparing this sensor output VRO with a second slice level SL2 equivalent to the target air-fuel ratio (stoichiometric air-fuel ratio).
reading out the learning value MSL of the slice level stored at an address corresponding to the small area to which the operating condition belongs at the time of judgment by the means 45 for determining which side of lean it is on and the operating condition at the time of judgment by the small area determining means 37; A means 46 for updating the read learning value MSL according to the rich/rich determination result, a means 47 for measuring the degree of deterioration of the upstream air-fuel ratio sensor 34; Means 48 for increasing the degree of averaging processing is provided.

(Operation) In the first invention, since the upstream 111 air-fuel ratio sensor output VFO is averaged, the rising edge of the output waveform
The lowering blade r9 becomes gentle.

Therefore, when the slice level is changed, the ratio of rich time to lean time changes greatly.

Furthermore, when the slice-level learning value MSL corresponding to the small area to which the current operating condition belongs is read out and used, the learning value MSL changes stepwise at the boundary of the small area, so the correction coefficient a matches the required value well.

In the second invention, the degree of deterioration of the upstream air-fuel ratio sensor 34 is measured, and as long as no deterioration occurs, the air-fuel ratio is controlled by directly using the upstream air-fuel ratio sensor output VFO without performing weighted averaging. executed.

On the other hand, when the degree of deterioration of the upstream air-fuel ratio sensor 34 increases, the weighted average of the sensor output VFO is strengthened, and the rise and fall of the output waveform are made gentler.

(Embodiment) FIG. 2 is a system diagram of an embodiment common to each invention. In FIG. Based on the injection signal from the injector (fuel injection!tiI device) 4, the fuel is injected toward the suction port of the upstream engine 1.

The gas burned in the cylinder is introduced into the catalytic converter 6 located downstream of the wandering pipe 5, where the harmful components (Co, HC, No) in the combustion gas go up to the three-way catalyst and are purified and discharged. be done.

The intake air amount Qa is detected by the air 70-meter 7,
The flow rate is controlled by a throttle valve 8 that is linked to an accelerator pedal. The engine speed Ne is detected by a crank angle sensor 10, and the cooling water temperature Tw of the engine jacket is detected by a water temperature sensor 11.

02 sensors (air-fuel ratio sensors) 12A are provided in the #S pipes upstream and downstream of the catalytic converter 6, respectively.

12B has characteristics that suddenly change after the stoichiometric air-fuel ratio,
It outputs so-called binary values indicating whether the mixture is richer or leaner than the stoichiometric air-fuel mixture. Note that the sensor is not limited to the O2 sensor, and may be a wide range air-fuel ratio sensor, a lean sensor, or the like.

9 is a sensor that detects the opening degree of throttle valve 8; 13 is /
7 is a vehicle speed sensor, and 14 is a vehicle speed sensor.

Above air 70-meter 7. Outputs from the crank angle sensor 10, water temperature sensor 11, two 02 sensors 12A, 12B, etc. are input to the control unit 21, and the control unit 21 outputs a fuel injection signal to the injector 4.

tlIJ3 Figure 3 shows a block diagram of the Huff control unit 21, in which the CPU 23 adjusts the air-fuel ratio with a learning function according to the points shown in FIGS. 5 and 6, and also according to the points shown in FIGS. The I10 boat 22 that performs feedback control is shown in FIG. 1 (A).
and functions as the output means 42 in FIG. 1(B).

The upper part of FIG. 4 shows the response waveform of the upstream 02 sensor output VFO, and the rapid rise and fall indicates good responsiveness. However, with such characteristics, even if the slice level SL is raised or lowered, the ratio H of the rich time to the lean time does not change much.

In this case, as shown in the bottom row, the rise of the output waveform,
If the fall is made gentler, the time ratio H changes greatly even if the slice level SL is increased by Δ in the same way as in the upper stage.

Therefore, in this example, the sensor output VFO is subjected to averaging processing to make the output waveform gentler, and this gentler sensor output is compared with the slice level.

FIG. 5 shows an averaging process of the sensor output VFO and an air-fuel ratio feedback control routine based on the averaged sensor output, which is executed in rotational synchronization.

S41 is a part that performs the function of the averaging processing means 35 in FIG. 1(A), and here, the upper tlfMO2 sensor output VFO
The weighted average value MV of the sensor output VFO is calculated by the following formula using
Ask for FO.

MVFO = (1-1/K) MVFO + (1/K) VFO...■ However, 1/ is a weighting coefficient, and a value less than or equal to 1 (-constant value)
It is.

In this case, calculating the weighted average is equivalent to applying a filter in terms of electrical circuits, and the smaller 1/K (the larger the value of K), the more gradual the response waveform of the sensor output becomes. Teku.

In S42, check whether it is within the feedback control range,
If it is within this control range, proceed to 843.

S43 to S45 are parts that function as the inverting control means 39 in FIG. to determine whether it has been reversed.

Note that S44. S45 "FRLJ is a flag that stores the previous comparison result of MVFO and SL, and FRL=
R means that it was rich last time, and FRL=L means that it was lean last time. S46. In S54, since it is immediately after the reversal, the training of this 7-lag FRL is replaced.

The "tree" in S47 and 355 indicates that an instruction to start the routine of FIG. 6 is given, and this notation is also used in FIG. 11.

Figure 6 is a routine for updating the learned value MSL of the slice level on the side to be compared with the upstream ○: sensor output VFO based on the downstream @02 sensor output VRO. Executed every time there is a flip.

S71 is a part that performs the function of the current small region determining means 37 in FIG. 1(A), and here it is determined to which small region the current operating condition belongs.

As shown in the tlIJ7 diagram, this small area includes at least the engine speed Ne and the engine load (for example, Tp).
The total number of small areas, which are small operating areas divided into a plurality of areas according to the operating conditions determined from the above, is set to be an excessive number in relation to the memory capacity. In FIG. 7, the engine load, engine speed, and cooling water temperature are also classified as parameters.

S? Reference numeral 2 denotes a portion that performs the function of the learned value reading means 38 in FIG. 1(A). Here, by referring to the learned value 77, the slice level learned value MSL is stored at the address corresponding to the small area to which the current operating condition belongs.
Look up.

The contents of this map of learned values are shown in FIG. 8. This map has the same number of addresses as the plurality of small areas shown in FIG.
Separate learning values are stored for each small area. The learning value map starts from RAM25! t, be done.

If it is determined in S74 that it is in the same small area as the previous time, the process advances to S75 and the counter value j is incremented by 1.

This counter value j represents the number of air-fuel ratio reversals.

In S76, j is compared with a predetermined number of reversals (for example, 5 times) n, and if j>n, it is determined that the vehicle has continuously stayed in the small region with the same operating conditions for a predetermined number of reversals (a predetermined period). Then, the process proceeds to S77. If it is determined in S74 that the small area is not the same as the previous time, the counter value j is reset in S83.

This requires that the vehicle continuously stay in a small area with the same operating conditions for a predetermined period of time as a learning condition. The reason why the condition is that they be in the same small region is that if the small 1i regions are far apart, the state of the intake air and fuel that affect the air-fuel ratio will vary accordingly.

Further, the predetermined number of reversals n represents the response delay time of the downstream 02 sensor. This is because the mixture formed from the amount of fuel corrected by the feed balance 9 correction image failure α is burned and discharged into the exhaust pipe, and the downstream l1lIo2 sensor 12
This is because there is a response delay of a predetermined period τ before reaching B, so the current air-fuel ratio obtained from the downstream 02 sensor output is for a small region to which the operating conditions before τ belong.

Instead of the number of reversals of the upstream @02 sensor output, engine speed, cumulative intake air amount, fuel amount, elapsed time, etc. may be used. For example, if Fig. 6 is started in time synchronization,
is the elapsed time, if the engine is started with rotational synchronization, j is the engine speed, and if the engine is started for each unit intake air amount or unit fuel amount, j is the integrated value of the intake air amount or fuel amount. It turns out.

S77, S79. S80. S82 is a part that performs the function of the learning value updating means 46 in FIG. 1(A).

In S77, the downstream side 02 sensor output VRO and the slice level 5L2 equivalent to the stoichiometric air-fuel ratio (a constant value, for example 500
+amV), if it is determined that the air-fuel ratio is on the rich side, the process proceeds to S79.

In S79, the slice level learning value MSL is updated using the following equation.

MSL=MSL-DSLR...■ In this case, the reason why the constant value DSLR is subtracted is as follows. Proceeding to S79 is when it is determined that the air is on the rich side, so the air-fuel ratio must be shifted to the lean side by increasing the ratio H between rich time and lean time. This is because it is sufficient to lower the slice level SL.

On the other hand, if it is on the lean side in S77, the process advances to S82, where the slice level learning value MSL is updated using the following equation.

MSL=MSL-DSLL...■ However, in equation 0, DSLL is also a constant value.

Generally, DSLL>DSLR.

In S80, the updated learning value MSL is stored at an address corresponding to the same small area.

S78 I skipped the explanation. In S81, the following equation is used to provide hysteresis in the slice level SL2.

SL2'=MSL2-ΔSL...■ SL2=MSL2+ΔSL...■ However, MSL2 in formulas ■ and ■ is a slice level <constant value> equivalent to the stoichiometric air-fuel ratio (1 standard air-fuel ratio), and ΔSL2 is the width of hysteresis. It is a specified value (for example, 25-V).

Returning to FIG. 5, S48. In S56 as well, instead of using the learned value MSL as the slice level SL as it is, hysteresis is provided using the following equation.

SL=MSL-ΔSL...-■ SL=MSL+ΔSL...■ S49 to S53. S57 to S61 are portions that perform the function of the air-fuel ratio feedback correction amount calculation means 40 shown in FIG. 1(A).

Of these, S49. S51. S57. In S59, by referring to each map of the proportional part and the integral part according to the determination results in the above steps 543 to S45, the proportional parts PR, PL and the integral part IRt! Find l, respectively.

In addition, in S52 and S60, "rip load correction" and "fiL load correction" are calculated by multiplying the map values (R and iL) by the engine load (for example, fuel injection pulse width Ti) by the integral I R , IL. This notation will also be used in FIG. 11, which will be described later.

I R= i RX T i- ■ I L= iLX T knee ■ The engine load is not limited to Ti, and may be Tp 10 FST, etc. However, Tp is the basic injection pulse width and 0FST is the offset amount.

This kind of load correction is necessary because a f) In the operating range where the IIJ control period is long, the amplitude of α becomes large and the exhaust purification performance of the three-way catalyst may deteriorate. This is to keep it almost constant regardless of the control cycle.

S50. S53. S58,561tllJt example,! −
The air-fuel ratio feedback correction coefficient a is calculated using M minutes.

From a thus obtained, the fuel injection pulse width T1 is determined according to FIG. 18. At S31 in Figure 18, the first
The function of the basic injection amount calculation means 33 shown in FIG.
At S33, the function of the fuel injection amount determining means 41 shown in FIG. 1(A) is performed.

Here, the operation of this example will be explained.

In this example, since the upstream @02 sensor output VFO is averaged, its output waveform changes from the upper stage to the lower stage in FIG. For this reason, in the lower row, when the slice level SL is increased, the ratio H between the time in the rich state and the Ftf open time in the lean state becomes significantly smaller. This means that even if the same bamboo slice level as in the upper stage is changed, the control air-fuel ratio is shifted significantly, that is, the air-fuel ratio shift sensitivity increases as the slice level changes.

As a result, even if the upstream 02 sensor has good responsiveness, the air-fuel ratio feedback control based on this sensor increases the ability to absorb variations in air-fuel ratio errors, and errors from the stoichiometric air-fuel ratio are resolved with good response. Ru.

Furthermore, in this example, since the learning area is divided into small sections, it is possible to accurately correct air-fuel ratio errors that vary depending on operating conditions.

For example, if Figure jII19 shows the changes when the vehicle speed is increased during driving, the driving conditions are A- before and after the heavy speed change.
A small region different from O13-+C is migrated.

In order to improve the ability of the correction coefficient a to catch up with such changes in vehicle speed using simple air-fuel ratio feedback control without a learning function, it is necessary to increase the rate of change of a.This just means that the slope of a This corresponds to increasing as shown by the broken line in the figure. However, if the slope is increased, the catching up during the transient becomes better, but in the steady state before and after the transient, hunting also increases due to the large rate of change.

On the other hand, in this example, a separate learning value MSL is provided for each different small M region, and the learning value MSL corresponding to the small region to which the current operating condition belongs is read out and used. This means that it changes stepwise at the boundary of the small area as shown in the figure, so it is well matched to the required value of a.
'Cysru.

Further, since the learning value is updated in consideration of a predetermined period, that is, the response delay of the downstream 02 sensor output, learning accuracy is guaranteed.

Furthermore, since the learning value is updated every time the upstream 02 sensor output VFO is reversed, the air-fuel ratio feedback control by the upstream 1102 sensor and the learning control by the downstream 02 sensor can be matched. For example, upstream@
02 sensor output is not present immediately after reversal, upstream 110
The 2 sensor output itself is also a colossus! ! Since it has not caught up with the fuel ratio, even if the learned value is updated in this state, it will not be consistent. In addition, due to the above learning effect, upstream@02
Since the inversion period of the sensor output becomes shorter, the frequency of updating becomes higher accordingly, and control accuracy improves.

FIG. 10 shows the emission characteristics of this example in comparison with a conventional example. In the same figure, B is the downstream @ 02 sensor output, and when the slice level on the side to be compared with the upstream 02 sensor output is corrected (conventional example), C is the further upstream 0
When averaging processing is applied to the two sensor outputs, D is the learning I for each small area on top of that. These are the characteristics of the present invention. Note that A is the characteristic of the type in which the 02 sensor is not provided downstream.

11 to 14 show an embodiment of the second invention, of which FIG. 11 and FIG. 12 correspond to FIG. 5 and FIG. 6, respectively. This example takes into consideration the deterioration that occurs in the upstream 02 sensor.

In FIG. 12, S 91 , S 92 and S 95 . S
In 96, the learning value updated in S79 and S82 is set to the lower limit -
in or the upper limit value - ax. This is to stabilize the air-fuel ratio control by limiting the range that can be controlled by the learned value.

However, S91. MSL on S95 <winS MSL>
It is considered that the upstream mO2 sensor has deteriorated due to the occurrence of *state called wax. As a countermeasure for this, S93. In S97, the routine shown in FIG. 13 is activated.

FIG. 13 shows Il! which can shift the air-fuel ratio. This is a routine for widening the range, and is executed every time the learned value MSL falls below the lower limit value 5hin or exceeds the upper limit value -ax.

5101 and 5104 are 864. in FIG. S94 in FIG. S98. Along with S99, lf! of the deterioration degree measuring means 47 in FIG. 1(B)! It is the part that performs the function.

At 5101, the counter value f is incremented by 1. This counter value 2 represents the number of times the learned value MSL has fallen below 2 or exceeded 2, that is, the degree of deterioration.

5102 and 8103 are average reinforcement steps 4 in Figure 1 (B)
This is the part that fulfills the functions of 8.

In 5102, 2 is compared with a predetermined number of times -, and if 2 exceeds 2, it is determined that the upstream 1102 sensor has deteriorated, and the process proceeds to 5103.

In step 5103, the value of K in equation (2) is increased by 1. In addition to this, the rise and fall of the averaged sensor output MVFO becomes even more gradual.

At 5104, the counter value 2 is reset.

FIG. 14 shows an initialization routine when the battery is dead or the like. If it is determined in step 5111 that the computer is in the initial state due to rising battery power, etc., the process proceeds to step 5112, where the value of K is returned to the initial value of 1.

According to this example, while there is no deterioration in the upstream @02 sensor (win < M S L < wax),
The responsiveness of air-fuel ratio control is ensured by using the upstream @02 sensor output as it is without performing weighted averaging, and the upstream [1102
If the sensor has deteriorated (M SL < sin or M SL > wax), increase the weighted average of the upstream 02 sensor output and widen the adjustment range of the air-fuel ratio shift (air-fuel ratio shift adjustment range due to change in slice level SL). This increases the sensitivity of the fuel ratio shift) and prevents deterioration of exhaust emissions.

Fig. 15 shows emitters with different values of K.

Indicates the characteristics of the Note that the case where K=1 is the case where weighted averaging is not performed. However, by increasing the degree of weighted averaging (
(by increasing the value of K), the air-fuel ratio can be shifted.
Although the range of adjustment becomes wider, on the other hand, it also increases the response delay of the upstream 02 sensor and weakens the purification performance, so it is possible to impose restrictions on increasing the degree of weighted averaging. .

(Effects of the Invention) In the first invention, the upstream air-fuel ratio sensor output is averaged, and the averaged sensor output is compared with a slice level learning value stored corresponding to a small area. In addition to determining whether the air-fuel ratio has been reversed,
We decided to update this learning value for each small region when the predetermined learning conditions are met, so even if the upstream air-fuel ratio sensor changes with good response, it is possible to eliminate small air-fuel ratio errors that vary depending on the operating conditions. , control accuracy can be improved.

In the second invention, the degree of deterioration of the upstream air-fuel ratio sensor is measured, and the degree of the averaging process is increased as the degree of deterioration increases. It can be purified.

[Brief explanation of drawings]

Figures 1 (A) and 1 (B) are claims correspondence diagrams of each invention, Figure 2 is a control system diagram of an embodiment common to each invention, and Figure 3 is a block diagram of the control unit of this embodiment. Figure 4 is a waveform diagram of the upstream side 02 sensor output to explain how to shift the air-fuel ratio by changing the slice level when the air-fuel ratio becomes lean. 1/117 is a flowchart for explaining the control operation of one embodiment of the invention, FIG. 8 is an operating region diagram for explaining the small region of this embodiment, and FIG. FIG. 9 is a waveform diagram for explaining the action of this embodiment, and FIG. 11110 is a characteristic diagram of exhaust emissions of this embodiment. 11 to 14 are flowcharts for explaining the control operation of an embodiment of the second invention, and FIG. 15 is a characteristic diagram of exhaust emissions of this embodiment. 4@16 to 18 are flowcharts for explaining the control operation of the conventional example, respectively. 4... Injector (fuel injection device), 5... Exhaust pipe, 6... Catalyst humbater, 7... Air 70-meter (engine load sensor), 10... Crank angle sensor (engine speed sensor), 11... water temperature sensor,
12A...Upstream 02 sensor (upstream air-fuel ratio sensor)
, 12B...Downstream side 02 sensor (downstream side air-fuel ratio sensor), 21...Control unit, 31...Engine load sensor, 32...Engine speed sensor, 3
3... Basic injection amount calculation means, 34... Upstream air-fuel ratio sensor (first sensor), 35... Averaging processing means,
36...Learned value storage steps, 37...Current small area determination means, 38...Learned value readout means, 39...Reversal determination means, 40...Air-fuel ratio feed ratio correction amount calculation means , 41...Fuel injection amount determining means, 42...Output means, 43...Fuel injection device, 44...Downstream air-fuel ratio sensor (second sensor), 45--Rich, lean determination Means, 46... Learning value updating means, 47... Deterioration degree measuring means, 48... Averaging reinforcement means. Fig. 4 Fig. 8 Fig. 10 NOx ・: A @: B ('L east example) Fig. 13 Fig. 14 Fig. 15 ・:ni-3 @:ni-2 0:ni=Ir front 02 Toguchi 11 Average Price 11 No. 17 Fig. 18

Claims (1)

[Scope of Claims] 1. A sensor that detects the load and engine speed of the engine, a means for calculating the basic injection amount according to these detected values, and a sensor installed in the exhaust passage upstream of the catalytic converter to calculate the exhaust air-fuel ratio. A first sensor that outputs an output according to the engine speed, a means for averaging the sensor output, and at least the same number of addresses as the plurality of small areas divided according to the operating conditions determined from the load and rotation speed of the engine. and means for storing a learned value of a slice level corresponding to a target air-fuel ratio corresponding to each small region;
means for determining which of the small areas the current operating conditions belong to; means for reading out the learned value of the slice level stored at an address corresponding to the small area to which the current operating conditions belong; and the learned value. means for determining whether the air-fuel ratio has been reversed by comparing the averaged sensor output with the averaged sensor output; and air-fuel ratio feedback so that the air-fuel ratio is controlled to be near the target air-fuel ratio according to the determination result. means for calculating a correction amount; means for correcting the basic injection amount using the air-fuel ratio feedback correction amount to determine the fuel injection amount; means for outputting the injection amount to the fuel injection device; A second sensor is installed in the exhaust passage of the engine and outputs an output according to the exhaust air-fuel ratio, and a comparison between this sensor output and a second slice level corresponding to the target air-fuel ratio determines whether the air-fuel ratio is rich or lean. read out the learned value of the slice level stored at an address corresponding to the small area to which the operating condition belongs at the time of judgment by the small area judgment means; 1. An air-fuel ratio control device for an engine, comprising: means for updating according to a lean determination result. 2. A sensor that detects the engine load and engine speed, a means to calculate the basic injection amount according to these detected values, and a sensor installed in the exhaust passage upstream of the catalytic converter to output an output according to the exhaust air-fuel ratio. It has a first sensor, a means for averaging the sensor output, and the same number of addresses as a plurality of small areas divided according to operating conditions determined from at least the engine load and rotation speed, and each small area has the same number of addresses. means for correspondingly storing a learned value of a slice level corresponding to a target air-fuel ratio;
means for determining which of the small areas the current operating conditions belong to; means for reading out the learned value of the slice level stored at an address corresponding to the small area to which the current operating conditions belong; and the learned value. means for determining whether the air-fuel ratio has been reversed by comparing the averaged sensor output with the averaged sensor output; and air-fuel ratio feedback so that the air-fuel ratio is controlled to be near the target air-fuel ratio according to the determination result. means for calculating a correction amount; means for correcting the basic injection amount using the air-fuel ratio feedback correction amount to determine the fuel injection amount; means for outputting the injection amount to the fuel injection device; A second sensor is installed in the exhaust passage of the engine and outputs an output according to the exhaust air-fuel ratio, and a comparison between this sensor output and a second slice level corresponding to the target air-fuel ratio determines whether the air-fuel ratio is rich or lean. read out the learned value of the slice level stored at an address corresponding to the small area to which the operating condition belongs at the time of judgment by the small area judgment means; , a means for updating according to a lean determination result, a means for measuring the degree of deterioration of the upstream air-fuel ratio sensor, and a means for increasing the degree of the averaging process as the degree of deterioration increases. An engine air-fuel ratio control device characterized by: