GB2249846A - Air-fuel ratio learning control system for automotive vehicle - Google Patents

Abstract

An air fuel ratio control system provides a basic fuel injection quantity and has a feedback loop providing a limited range of correction to the basic quantity in dependence on the output of an exhaust gas oxygen sensor. Under normal circumstances the feedback loop provides a compensation signal a depending on proportional and integral error and causes the air fuel ratio to oscillate around the optimum level. In the event of a large discrepancy between the optimum and the actual air fuel ratio, the cessation of oscillation is detected over a time interval t0 and a second correction factor, the learning value, is steadily increased to assist the return of the air fuel ratio to the optimum. While the learning value changing, the integration coefficient Ki in the feedback loop is reduced in value to limit the change in the compensation signal a. When the air fuel ratio returns to optimum, the recommencement of oscillation is detected over a predetermined number of cycles and any further correction DELTA alpha in the compensation signal a is transferred to the to, learning value. <IMAGE>

Description

2-2493,+0 1 AIR-FUEL RATIO LEARNING CONTROL SYSTEM FOR AUTOMOTIVE ENGINE

The present invention relates to an air-fuel ratio learning control system for an automotive engine, and particularly to a system for correcting an air-fuel ratio which deviates largely from a theoretical air-fuel ratio.

A conventional air-fuel ratio feedback control is performed in a manner of usually keeping a condition of the theoretical air-fuel ratio, even if a driving condition changes largely. The feedback control system adopts learning control in order to rapidly compensate for a discrepancy of the air-fuel ratio in dependency on the accuracy of a production and a change with time in such an intake air system as an intake air quantity sensor and in such a fuel system as an injector.

Namely, a stable state is judged by an engine speed and a load,, for example, basic fuel inject ' ion quantity.

in the constant condition, a center value of a compensation coefficient of a closed loop by an 02 sensor, namely, an air-fuel ratio feedback compensation coefficient a i s. repeatedly rich and lean in predetermined numbers by a proportion and integration control, the center value is stored as a learning value (a compensation coefficient of an open loop) so as to set the center value to a reference value ao (= 1) by a learning compensation.

The learning value therefore directly influences a fuel injection quantity to maintain an air-fuel ratio to the theoretical air-fuel ratio regardless to an integration speed (generally, a few %/SEC.) of the proportion and integration control when the driving condition changes.

However, the feedback com-Densation coefficient a converges to a compensation limit value so that the coefficient a cannot repeat the rich/lean air-fuel ratio 2 when the 02 sensor deteriorates to reduce its output characteristics and to maintain a discrepancy condition from a slice level, or when a conventional learning control cannot be applied by large deviation from the theoretical air-fuel ratio. Accordingly, a new learning control canno t be performed so that the conventional airfuel ratio control has a problem to deteriorate an operating performance.

in order to solve the above problem, there is proposed a so-called double 02 sensor system, in which the 02 sensors are respectively provided at both the upand down-streams of a catalytic converter, and the downstream sensor compensates for the deterioration of the output characteristics of the up-stream sensor so as to perform the air-fuel ratio control. For exam-Dle, Japanese Official Gazette of Patent Application Laid-open No. 63-45440 (1988) discloses a technology to improve a response speed of the down-stream 02 sensor and to prevent deteriorations of an exhaust emission, a fuel cost and a drivability in dependency on the deterioration of the downstream 02 sensor.

However, even if the double 02 sensor system is adopted, the conventional system has problems in that learning compensation cannot be performed when the system has a large deviation of the actual air-fuel ratio from the theoretical one, and the control system becomes complicated in the above condition, thereby increasing the production cost.

In view of the above situation, an object of the present invention is to provide an air-fuel ratio learning control system for an automotive engine for compensating a learning control when an actual air-fuel ratio has a large deviation from the theoretical air-fuel ratio, and preventing deterioration in the exhaust emission, the fuel cost and the drivability.

i i 3 In order to achieve the above object, the air-fuel ratio learning compensation system of the present invention, in which a fuel injection quantity is set by compensating a basic fuel injection quantity in dependency on an air-fuel ratio feedback compensation coefficient based on an output from such an exhaust gas sensor as an 02 sensor and a learning compensation coefficient based on the feedback compensation coefficient, comprises a reference time setting circuit for setting a reference time responsive to an output inversion frequency from the exhaust gas sensor against a predetermined slice level in dependency on an engine speed and the basic fuel injection quantity, a discriminating circuit for an inversion of an 0. sensor output for discriminating whether or not the output of the 02 sensor is inverted in the reference time set by the reference time setting circuit, a coefficient setting circuit for setting the air-fuel ratio feedback compensation coefficient in the manner that an integration constant of a proportion and integration control becomes smaller than the case that the sensor output is inverted in the reference time when the sensor is discriminated to not invert its output in the reference time, and a learning circuit for setting a learning compensation coefficient by increasing or decreasing a predetermined value of the learning value in the same direction as the feedback compensation coefficient responsive to a reduction of the integration constant when the sensor is discriminated to not invert its output in the reference time by the discriminating circuit.

In the air-fuel ratio learning control system according to the present invention, the reference time is first set in dependency on the engine speed and the basic fuel injection quantity responsive to the inversion frequency of the output from the exhaust gas sensor at the predetermined slice level, thereby discriminating 4 whether or not the output of the exhaust gas sensor is inverted in the reference time.

The integration constant of the proportion and integration control is set to smaller than the constant of the time when the output of the sensor is discriminated to invert in the reference time, when the output is discriminated not to invert in the reference time, thereby setting the air-fuel ratio feedback compensation coefficient. The learning compensation coefficient is set by increasing and decreasing the predetermined amount of the learning value with the same direction as the feedback compensation coefficient corresponding to the reduction of the integration constant.

The basic fuel injection quantity is compensated in dependency on both the feedback and learning compensation coefficients, thereby setting the fuel injection quantity to control the air-fuel ratio.

As described above,, as the system of the present invention judges whether or not the output of the exhaust sensor is inverted in the reference time to change the integration constant, it is possible to prevent the deterioration of' the air-fuel ratio when the feedback compensation coefficient reaches the uppermost and lowermost limits so that learning is not performed.

Namely, the present invention has the excellent effect that the learning compensation can be properly performed even when the actual air-fuel ratio is inconsistent with the theoretical air-fuel ratio, and that it is possible to prevent the deterioration of the exhaust emission, fuel cost, and drivability.

FIG. 1 is a block diagram showing a construction of an air-fuel ratio learning control system according to an embodiment of the present invention; FIG. 2 is a schematic view showing an engine control system in which there is provided the air-fuel ratio i i t 1 i learning control system of the embodiment shown in FIG.

1; FIG. 3 is an explanatory view showing a table for judging a constant condition and a learning value table; FIG. 4 is an explanatory view showing a reference time map for discriminating a changeover of rich/lean; FIG. 5 is a characteristic diagram showing air-fuel ratio feedback and learning compensations in dependency on an output voltage of an 02 sensor, respectively; FIG. 6 is a flow chart showing a control procedure of a fuel injection by the control system of the embodiment; FIG. 7 is a flow chart showing a procedure to set the air-fuel ratio feedback compensation coefficient; and FIG. 8 is a f low chart showing a procedure to renew the learning value.

[Configuration of Engine Control System] As shown in FIG.. 2, numeral 1 denotes an engine which is an example of a horizonta lly opposed piston arrangement having four cylinders. The engine 1 has a cylinder head 2 having an intake port 2a, an intake manifold 3 connected to the intake port 2a, an air chamber 4 connected to the uP-stream side of the manifold 3, a throttle chamber 5 connected to the air chamber 4, an intake pipe 6 connected to the upper stream side of the throttle -chamber 5, and an air cleaner attached to the intake pipe 6.

Such an intake air quantity sensor 8 as a hot wire type air flow meter is provided immediately down-stream of the air cleaner 7 of the intake pipe 6. A throttle opening sensor 9a, and an idling switch 9b are provided to a throttle valve 5a which is provided in the throttle chamber 5. The throttle opening sensor 9a detects an opening degree of the throttle valve Sar and the idling switch 9b detects a full-closed condition of the valve 5a.

6 An injector 10 is mounted on an immediately upper stream of the intake port 2a of each intake manifold 3. An ignition plug 11 is attached to each cylinder of the cylinder head 2 and a top end thereof exposes in a combustion chamber of the engine 1.

The injector 10 is connected through a fuel supply pipe 12 to a f uel tank 13. The pipe 12 has a fuel pump 14 and a fuel filter 15 in the order from the side of the tank 13.

The injector 10 is further connected through a return path 16 to a fuel chamber 17a of a pressure regulator 17. The chamber 17a is connected to the fuel tank 13 at the down stream side thereof. A pressure chamber l7b of the regulator 17 is connected to the intake manifold 3 as shown by a chain line in FIG. 2.

Accordingly, fuel is sent with pressure by the pump 14 from the tank 13 through filter 15, injector 10 and regulator 17. A fuel injection quantity of the injector is controlled in thei.manner that the pressure change in the intake manifold 3 does not influence the change of the quantity. Because the fuel is supplied tothe injector 10-under the constant condition where there is constant of a differential pressure between pressures in the intake manifold 3 and fuels.

The engine 1 has a crank shaf t lb to which a crank rotor 18 is attached, and a crank angle sensor 19 is mounted to face the outer surface of the rotor 18. The sensor 19 is made from an electromagnetic pickup or the like in order to detect a crank angle of the engine 1.

The engine 1 also has a cam shaft lc to which a cgm rotor is attached. The cam shaf t lc rotates a half while the crank shaf t lb rotates once. A cam angle sensor 21 is mounted to face the outer surface of the cam rotor 20 in order to discriminate a predetermined cylinder.

The intake manifold 3 has a coolant path (not shown) as a riser. A coolant temperature sensor 22 is exposed in the coolant path. Such an exhaust gas sensor as an 02 i 1 1 7 sensor 24 exposes in an exhaust pipe 23 which is connected to an exhaust port 2b of the cylinder head 2.

Numeral 25 denotes a catalytic converter.

(Circuit of Control System] On the other handr numeral 30 denotes a control system comprising a-central processing unit (CPU) 31r a read only memory (ROM) 32, a random access memory (RAM) 33r a backup random access memory (backup RAM) 34, an input/output (1/0) interface 35r a busline 36 for connecting components 31 to 35 with each other, and a constant voltage circuit 37 for supplying a constant voltage having a predetermined voltage to the components.

The constant voltage circuit 37 is connected through a control relay 38 to a battery 39 and a key switch 40.

is An electric power is supplied to each part to be controlled when a relay contact of the control relay 38 is closed by turning on the key switch 40, and a power line is directly connected to the battery 39. On the contrarYr when the contact of the relay 39 is open by turning off the key switch 40r the power source is supplied to the backup RAM 34 with backup power so as to maintain data.

The sensor 8, 9a, 19r 21, 22 and 24 and the idling switch 9b are connected to an input port of the 1/0 interface 35, and a plus terminal of the battery 39 is also connected to the input port of the interface 35 so as to monitor -a terminal voltage V. of the battery 39.

On the other hand, the ignition plug 11 is connected through an igniter 26 to an output port of the interface 35, and the injector 10 and fuel pump 14 are also connected through a driving circuit 41 to the output port of the interface 35.

The ROM 32 stores such fixed data as a control program, a reference time map MPtO and the like, and the RAM 33 stores values output from each sensor after a data processing and data calculated by the CPU 31. The backup 1 1 8 RAM 34 stores a learning value table TBLR and maintains the stored data even if the switch 40 is turned off.

The CPU 31 calculates an intake air quantity in dependency on a signal output from intake air quantity sensor 8 according to the control program stored in the ROM 32, a fuel injection quantity adaptive to the intake air quantity in dependency on the various data stored in the RAM 33 and the backup RAM 34, and an ignition timing in dependency on the injection quantity, respectively. A driving pulse width signal is supplied through the driving circuit 41 to the injector 10 of the proper cylinder at a predetermined timing corresponding to the fuel injection quantity, so that the injector 10 injects fuel. An ignition signal is supplied through the igniter 26 to the ignition plug 11 of the proper cylinder at a predetermined timing.

As a result, mixture having a predetermined air-fuel ratio, is ignited and combusts in the proper cylinder, and the 02 sensor 24 exposed in the exhaust pipe 23 detects an oxygen density in exhaust gases. The CPU 31 compares a^ detection signal of the sensor 24 after filtering the waveform with a reference voltage signal, and judges whether gn actual air-fuel ratio condition is to the rich side or the lean side with respect to the theoretical air-fuel ratio, namely a desired air-fuel ratio, thereby storing a value "0" if the ratio is in rich, and a value '1111 if in lean in the RAM 33. The CPU 31 monitors an air-fuel ratio signal of the mixture at each predetermined time or each predetermined cycle, and calculates the succeeding data processing.

[Construction of Control System] The control system 30 comprises as a functional construction with relation to a fuel injection control (an air-fuel ratio control), as shown in FIG. 1, an 35 intake air quantity calculating circuit 50, an engine speed calculating circuit 51, a basic fuel injection quantity setting circuit 52, a various increase i 1 i i i 1 1 J i 1 9 compensation coefficient setting circuit 53, a voltage compensation coefficient setting circuit 54, a discriminating circuit 55 for completion of a feedback condition, a learning condition discriminating circuit 56, a reference time setting circuit 57 for an inversion of an output of the-exhaust gas sensor, a discriminating circuit 58 for discriminating the inversion of the exhaust gas sensor output, an integration constant compensation circuit 59, an air-fuel ratio feedback compensation coefficient setting circuit 60, a learning circuit 61, a fuel injection quantity setting circuit 62, and an injector driving circuit 63. The reference time setting circuit 57 comprises a reference time setting circuit 57a for discriminating a rich/lean changeover, and a reference time map MPtO for discriminating a rich/lean changeover. The learning circuit 61 comprises a learning value renewal circuit 61a, a learning compensating coefficient setting circuit 61b, and a learning value table TBLR 20 The calculating circuits 50 and 51 calculate an intake air quantity 0 and an engine speed SEY respectively, each in dependency on signals output from the intake air sensor 8 and crank angle sensor 19. The setting circuit 52 obtains a basic fuel injection quantity T P by a map retrieving or calculation in dependency on the intake air quantity Q and the engine speed S. each obtained by the calculating circuits 50 and 51.

In this embodiment, the basic fuel injection quantity T P is obtained by the calculation according to the equation of "TP = K x Qls,"r where K is a constant.

The setting circuit 53 reads a throttle opening (0) signal of the throttle valve opening degree sensor 9a, ON/OFF signal of the idling switch 9b, and a coolant temperature signal Tw of the coolant temperature sensor 22, and sets various increase compensation coefficient COEF such as an acceleration/deceleration compensation, a 1 full-opening increase compensation, an increase compensation after idling, and a coolant temperature compensation.

The setting circuit 54 sets a voltage compensation coefficient TS for interpolating an invalid injection time (a pulse width) of the injector 10 corresponding to a terminal voltage VB of the battery 39 after reading out the invalid time from a table (not shown).

The circuit 55 discriminates whether or not the output of the 02 sensor 24 is in an inactive region and reads out the coolant temperature signal Tw of the sensor 22, the engine speed S. calculated by the circuit 51, and the basic fuel injection quantity T p calculated by the circuit 52, thereby discriminating whether or not the air-fuel ratio feedback control condition is completed.

The discrimination as to whether or not the output of the 02 sensor 24 is in the active region, is performed by discrimination as to whether or not an output voltage VAP of the 02 sensor:. 24 is on or over a predetermined value. Even though the output voltage VA1, is over the predetermined value so as to discriminate the active condition, the air-fuel ratio feedback control condition is determined to be incomplete, when the coolant temperature Tw of the sensor 22 is under a predetermined value such as 50 degrees centigrader- when the engine speed S. is over a set speed SS such as 5, 200 r.p.m., or when the basic fuel injection quantity T p is over a set value Tp., namely, when the throttle valve is substantially full-opened. Accordingly, the air-fuel ratio feedback control condition is determined to be complete without the above three condition and when the 02 sensor 24 is in an active condition.

The learning condition discriminating circuit 56 reads out the engine speed S. calculated by the circuit 51 and the basic fuel injection quantity T p calculated by the circuit 52 when the circuit 55 discriminates the completion of the airfuel ratio feedback condition, so i i 1 i 11 as to discriminate whether or not both the speed S E and quantity T P are in a matrix MT (shown in FIG. 3) which is constructed by each address of the learning value table TBLR The discriminating circuit 56 determines a predetermined block -in the matrix MT when both the speed SE and quantity T P are in the matrix MT, to determine the learning condition completed when the determined block is the same as the previously selected block.

In the reference time setting circuit 57r the reference time setting circuit 57a sets a reference time to for discriminating the rich/lean changeover of the output voltage of the 02 sensor 24. The reference time to can be represented as an inversion cycle of which the output voltage VAr of the sensor 24 inverts against the slice level VS from rich to lean sides or from lean to rich sides of the air-fuel mixture.

The reference time to is set to a little longer time than an inversion cycle of the 02 sensor 24 at the normal time by retrieving the reference time map MPtO for discriminating the rich/lean - changeover by using the engine speed S. and the basic fuel injection quantity T P as parameters.

The ref erence time map MPtO is constructed from a map having the speed S. and the quantity T P as parameters, as shown in FIG. 4, in which each address stores the reference time to previously obtained by such as an experiment.

The desired time for the rich/lean changeover is shortened with increasing the load and rotation speed of the engine, and lengthened with decreasing the load and speed. For example, there is usually desired 1 second of the rich/lean change over time at idling, but the reference time to is desired as "to = 2-3 ('seconds)" at idling.

The discriminating circuit 58 reads the output voltage V., of the sensor 24 and discriminates whether or 1 11 12 not the voltage V.. is inverted f rom rich to lean sides or from lean to rich sides of the air-fuel ratio in the reference time tO against the slice level Vf thereby outputting the discrimination signal to the learning value renewing circuit 61a. At the same time, the circuit 58 determines the air-fuel condition to have a discrepancy with the theoretical ratio when the voltage VAF is not inverted in the reference time tO so as to output an instruction signal f or an I-value compensated calculation to the integration constant compensation circuit 59 and output an 1 value compensation information to the learning value renewing circuit 61a.

The compensating circuit 59 generates an I value compensation signal by multiplying the integration constant I of the proportional pulse integral control against the air-fuel ratio feedback compensation coefficient a by a predetermined compensation constant Ki (0 < Ki < 1), thereby outputting the I value compensation signal to the feedback compensation setting circuit 60.

The setting circuit 60 compares the output voltage VAF OE the 02 sensor 24 with the slice level Vs when the circuit 55 discriminates the air-fuel ratio control condition completed, and sets the - air-fuel ratio feedback compensation coefficient a by the proport ion and integration control.

Namely, as a result of the comparison, when the air- fuel ratio is rich (VAp > Vs), the feedback compensation coefficient a is lowered first in the proportion constant P (a ±--a - P), then in the integration constant 1 step 30 by step (a (- a - I), thereby setting the air-fuel ratio to be lean. As a result, when the air-fuel ratio is lean by reducing the fuel injection quantity Ti (Vxl, < Vs) r the coefficient a is raised in the proportion constant P step by step (a <-- a + P), and in the integration 35 constant I step by step (a x- a + I). thereby setting the air-fuel ratio to be rich. Both operations are repeated.

13 In the above condition, the coefficient a is set to "a ± a - Ki x V' ar "a (- a + Ki x Ill responsive to the air-fuel ratio condition when the compensating circuit 59 supplies with the I value compensation signal.

When the circuit 55 discriminates the sensor 24 to be inactive or the f.eedback control condition incompleted by judging the throttle valve being in the substantially fully-opened condition, the feedback compensation coefficient a is-fixed to "a = 1".

The learning circuit 61 renews the learning value KLR in the learning table TBLR in dependency on the I value compensation information from the inversion discriminating circuit 58 when the circuit 56 discriminates the learning condition completed, or in dependency on the feedback compensation coefficient a set by the circuit 60 when the circuit 56 discriminates that the learning condition is completed and the engine is in the constant condition. Furthermore, the learning compensation coefficient KBLRC 'S set by interpolation operation in dependency on the learning value stored in the learning value table TBLR Namely, in the circuit 61, when the discriminating circuit 58 supplies the I compensation value, the renewing circuit 61a calculates a differential amount Aa (Aa = ao - a) between the feedback compensation coefficient a and a reference value ao, and renews the learning value KLR by adding or subtracting a fine set value KLRSET according to plus or minus symbol of the differential amount Aa in the address corresponding to the basic fuel injection amount TP as a parameter (namely r KLR ---KLR + KUSET) Next, when the air-fuel ratio is discriminated in a normal condition and the engine is determined to be normal by stopping the I value compensation and by inverting more than n times such as three times of the output voltage VAp of the sensor 24, the renewing circuit 61a calculates the differential amount Aa between an 14 average value a of the compensation coef f icient a and reference value a such as 1 (Aa = ao - a), thereby continually renewing one time the learning value KLR in the differential amount Aa.

The circuit 61a renews the learning value KLR by adding or subtracting with a predetermined rate of the differential amount Aa (KLR <- K LR + Aa/M or KLR ±- K LR - Aa/M, where an addition is performed when "Aa < 0", the subtraction is performed when "Aa > 0", and M denotes a constant for determining the rate of the learning value renewed) in the same manner as described above, when the circuit 58 discriminates the inversion of voltage output from the 02 sensor 24 over n times such as three times.

The learning value table TBLR is formed in the 15 backup RAM 34, as shown in FIG. 3, wherein each address corresponding to a range TpOT'Pl, TplTp2, Tp2Tp3,.,, or Tpn-lTpn of the basic fuel injection quantity Tp has each learning value KLR, and "K is stored as an LR 1011 initial value. 11 The setting circuit 61b retrieves the learning value table TBLR by using the basic fuel injection quantity as a parameter calculated by the circuit 52, thereby setting the learning compensation coefficient KBLRC by an interpolation operation.

The fuel injection quantity setting circuit 62 compensates the basic fuel injection quantity Tp calculated by setting circuit 52 in dependency on the various increase compensation coefficient COEF set by the circuit 53, the feedback compensation coefficient a by the circuit 60, and the voltage compensation coefficient Ts by the circuit 54, and at the same timer the circuit 62 sets the fuel injection quantity Ti by learning compensation by the compensation coefficient KBLRC set by the setting circuit 61b (Ti = Tp x COEF X KBLRC x a + Ts), thereby outputting a driving pulse signal at a predetermined timing corresponding to the fuel injection 1 i i 1 quantity through the injector driving circuit 63 to the injector 10.

As shown in FIG. 5r when an actual air-fuel ratio is largely discrepant from the theoretical air-fuel ratio and when the voltage VAp output f rom the 0 2 sensor 24 has a large discrepancy-with the slice level Vs, the feedback compensation coef f icient a is under a usual amount for lower compensation value relative to the integration constant I (a (-- a Ki x I; where 0 < Ki < 1), and the learning value KLR is renewed in a f ine amount (KLR <- KLR KLMET) The basic fuel injection quantity is then compensated in depende.ncy on the air-fuel ratio feedback compensation coefficient a and the learning compensation coefficient KBLW by the learning value KLM thereby setting the fuel injection quantity Ti.

Accordingly, the present invention supported by the embodiment has an effect that the actual air-fuel ratio can be maintained the theoretical ratio by the feedback compensation coefficient a and the learning compensation coefficient KURC by preventing the sticking of the feedback compensation coefficient a to the control limiter. However, the conventional air-fuel control system does not learn to deteriorate the air-fuel ratio when the actual air-fuel ratio has great discrepancy with the theoretical ratio, as shown by the dashed line of FIG. 5, because the feedback compensation coefficient a is compensated by the integration constant I to stick the control limit such as "0.7 < a < 1.21'.

Next, there will be described a control procedure of the control unit 30 with reference to flow charts as shown in FIGS. 6 to 8.

[Control Procedure of Control Unit] FIG. 6 shows a -procedure of the fuel injection control which is repeated at every predetermined cycle in synchronization with the engine rotation.

At first, in a step S101, signals output from the crank angle sensor 19 and the intake air quantity sensor 16 8, are read out, thereby calculating the engine speed S. and the intake air quantity Q.

In a step S102, the basic fuel injection quantity T. is calculated independency on the speed S. and quantity Q calculated in the step S101 (TP = K x Q/S,, where K is a constant), thereby- advancing to a step S103..

In the step S103, the coolant temperature Twf throttle opening degree 6, and output of the idling switch, are read from the coolant temperature sensor 22.

throttle opening degree sensor 9a, and idling switch 9b.

thereby setting various compensation coefficient COEF relevant to the coolant temperature, acceleration and deceleration, throttle full-opening increase. and increase after idling compensationr in a step S104.

Next, advancing to a step S105f the voltage compensation coefficient T. is set to interpolate an invalid injection time of the injector 10 in dependency on the terminal voltage VB of the battery 39, thereby advancing to a step S106.

In the step S106f the learning value table TBLR 'S retrieved to set the learning compensation coefficient KURC by the interpolational operation by means of using the basic fuel injection quantity T p calculated in the step S102 as a parameter.

Advancing to a step S107, the air-fuel ratio feedback compensation coefficient a stored in the RAM 33 is read out, thereby advancing to a step S108. The coefficient a is set by the program for setting the air fuel ratio feedback compensation coefficient as mentioned later.

In the step S108, the basic fuel injection quantity TP calculated in the step S102,. is compensated by the various increase compensation coefficient COEF in the step S104r the voltage compensation coefficient T. in t he 35 step S105, and the feedback compensation coefficient a in the step S107, and the learning compensation of the basic quantity T p is performed by the learning compensation 1 i i i i i i i i 17 coefficient KBLRC to set the actual fuel injection quantity Ti (Ti = T p x COEF x K BLRC X a + TS).

In a step S109, the driving pulse width signal is output to the injector 10 of the concerned cylinder in a predetermined timing corresponding to the actual fuel injection quantity Tli set in the step S108.

[Procedure of Setting Feedback Coefficient] There is described a procedure for setting the airfuel ratio feedback compensation coefficient a with reference to a flow chart as shown in FIG. 7.

A program of the setting procedure is one in which an operation is repeated in every predetermined time or every predetermined cycle.

At first, in step S201, an operation is as to whether or not the air-fuel ratio control condition is completed in dependency on the coolant temperature Tw, engine speed SE, and basic fuel injection quantity Tp.

If the feedback control condition is determined to and F thereby finishing the program. If the condition is determined to be completed, thereby advancing the operation from steps S201 to S202.

The voltage V.. output from the 02 sensor 24 is read 25 out in the step S202, and in a step S203r the voltage VAp is compared with the predetermined slice level Vs., thereby discriminating whether the present air-fuel ratio exists in a rich side or a lean side.

be incompleter an operation advances to a step S216 the feedback coefficient a is fixed to be "a = In the step S203. if the air-fuel ratio is 30 determined to be in the rich side, namely, "V,,,;.. V.511.. operation advances from S203 to a step S204. In the step S204, the discrimination flag PLAG1 for changing between rich/lean is determined as to whether it is set or not.

The value of the FLAG1 changes by an inversion of 35 the voltage VAF output from the sensor 24 from the lean side to rich side of the air-fuel ratio, or from rich to 18 the lean sides, thereby setting 111 --. 011 in the former case, and "0 --> 111 in the latter case.

Accordingly, as the feedback coefficient a is compensated previously by the integration constant I after skipping to the plus direction by the proportional constant P so that -the air-fuel ratio becomes rich when "FLAG1 = 111 in the step S204, an operation advances from the step S204 to a step S205. In the step S205, the feedback coefficient a is skipped to the minus direction with the proportional constant P (a ± a - P), and in a step S209, -the flag PLAG1 is clarified (FLAG1 0), thereby finishing the program.

on the other hand, the I value compensation flag FLAG2 is determined as to whether it is set or not in a step S206, when "FLAG1 = 011, namely, the skipping in the minus direction by the proportional constant P has been already performed in the step S204.

The I compensation f lag FLAG2 is as to whether or not the voltage VAI,-,_output from the 02 sensor 24 is inverted by reaching to the slice level V. in a reference time tO for discriminating the rich/lean changeover. In the step S206, when the voltage VAp is determined to be inverted-in the reference time tO, namely, '1FLAG2 = 011, an operation advances from the step S206 to a step S207, thereby reducing the feedback coefficient a in the integration constant I (a (- a - I). so that the operation advances to a step S209 to finish.

On the other hand, when the voltage VAF is determined not to be inverted in the reference time tO in the step S206, namely,, 11FLAG2 = 111,, in a step S208, the integration constant 1 decreases to less than the normal value by multiplying by a predetermined compensa-tion coef f icient Ki (0 < Ki < 1) so that it prevents the feedback coef f icient a from the reduction (a (- a - Ki X 1), thereby finishing the program through the step S209.

Returning to the step S203, when the air-fuel ratio is determined to be in the lean side, namely, "VAF < VS H r 1 1 19 operation advances from the step S203 to a step S210, the flag FLAG1 is determined as to whether it is set or not.

When '1FLAG1 = 011, namely, the air-fuel ratio becomes lean condition because the feedback coefficient a decreases step by step by means of the integration constant 1 after skipping in the minus direction by the proportional constant Pr an operation advances to a step S211.

In the step S211, the feedback compensation coefficient is skipped in the plus direction with the proportional constant P (a - a + P), advancing to a step S215, and the flag is set to 11FLAG1 = l", thereby finishing the program. When '1FLAG1 = Pr namely, the skipping in the plus direction is performed by the proportional constant P with respect to the feedback coefficient a, operation advances from the step S210 to a step S212, thereby discriminating whether the I value compensation flag FLAG2 is set or not.

In the step S212, when "FLAG2 011, operation advances to a step;. S213 to increase the feedback coefficient a with the integration constant I (a - a + 1), thereby advancing to a step S215. In contrast, when 11FLAG2 = 1, namely, the voltage VAp of he sensor 24 is not inverted in the reference time tOr the operation advances from the step S212 to a step S214.

In the step S214, the feedback coefficient a is compensated by the integration constant 1 smaller than in the ordinal case by multiplying the integration constant 1 by the compensation coefficient Ki (a ±-- a + Ki X 1), thereby advancing to a step S215.

In the step S215p the flag FLAG1 for discriminating the rich/lean changeover is clarified (FLAG1 0), thereby completing the program. [Renewing Procedure of Learning Value] A flow chart of FIG. 8 shows a program of the renewing procedure of the learning value, and the procedure is repeated in every predetermined time or cycle.

In a step S301, the feedback control condition is determined as to whether it is completed or not in the same manner as the step S201 of the feedback coefficient setting program in dependency on the coolant temperature T,,, engine speed SE and basic fuel injection quantity T p. The operation ends when the condition is not completed, and advances to a step S302 when the condition is completed.

In the step S302, the engine speed S. and the basic quantity T p are read out. In a step S303, the speed SE and the basic quantity T p are determinedas to whether it is in a region of the matrix MT shown in FIG. 3 or not (SO 5 SE; Sn, T13O;5 T1);g Tpn) The operation ends when both the value is determined to be out of the region in the step S303, but operation advances to a step S304 when both the values are determined to be in the matrix MT and in the region f or the learning value renewing. In the step S304, the speed S. and basic quantity:, T p specify the divisional position in the matrix MT such as the division D1, thereby discriminating whether or not the learning condition is completed by comparing the position previously specified from the matrix MT with one presently specified.

The operation advanbes f rom. the step S304 to a step S305 in dependency on the discrimination in which the learning condition is incomplete when the previously specified position is different from the presently specified position. In the'step S305r the RAM 33 stores the divisional position in the matrix MT which is presently specified by the present program as the previously divisional position data, thereby clarifying the counter value C (C - 0) to complete the program in a step S306.

In the first program execution, the operation advances from the step S303 to the step S305 by jumping, thereby completing the program through the step S306, 1 i i 1 21 because there is no positional data of the previous division.

On the other hand, the operation advances f rom the step S304 to a step S307 when both the positions previously specified division and presently specified division are the same. In the step S307, the reference time map MPtO is retrieved by using the engine speed SE and the basic quantity TP read out in the step S302 as the parameter to set the reference time tO for discriminating the rich/lean changeover.

Advancing to step S308, the voltage V,,, output from the 02 sensor 24 is read out and determined as to whether or not the voltage VAp is inverted f rom the rich side to the lean side or f rom the lean side to the rich side in the reference time tO which is set in the step S307.

If the voltage VAF is not inverted in the reference time tO in the step S308, the operation advances from the step S308 to a step S309, thereby setting the I value compensation flag FLAG2 "FLAG2 (-- 1".

Next, in a step S310, a differential amount Aa between the feedback coefficient a and a reference value ao of the feedback coefficient with respect to -the basic air-fuel ratio is calculated (Aa ± ao - a), thereby discriminating whether or not the differential amount Aa is "Aa < 011 in a step S311.

The operation advances from the step S311 to a step S312 when the differential amount Aa is determined as "Aa < 0" in the step S311, thereby retrieving the learning value table TBLR in dependency on the basic quantity T P read out in the step S302 as the parameter.

In a step S3131 the learning value is renewed to set a new value by adding the learning value KLR with a f ine set value KLRSET (KLR ± KLR + KLRSET) I thereby advancing to a step S317.; The learning value is renewed corresponding to the I value compensation in the step S214 of the air-fuel ratio 22 compensation coefficient setting program as abovedescribed.

On the other hand, the operation advances f rom. the step S311 to a step S314 when the differential amount Aa is on or over 0, namely, "Aa - 011. In the step S314, the differential amount Aa is determined as to whether it is over 11011 or on 110", namely, "Aa = 011 or "Aa > 011. When "Aa = 0". the sensor 24 judges an abnormality, thereby completing the program, and when "Aa > 0", the 10 operation advances to a step S315.

In the step S315, the learning table TBLR is retrieved by using the basic fuel quantity TP as the parameter which is read out in the step S302. Then, in a step S316, the fine set value KLRSET is subtracted from the learning value KLR to set the new learning value (KLR KLR - KLRSET corresponding to the I value compensation in the step S208 of the feedback coefficient setting program.)r thereby advancing to the step S317.

The I value compensation learning value renewing flag FLAG3 is set in "FLAG3 - 1" in the step S317. The FLAG3 represehts that the voltage VAP is not inverted in the reference time tO for discriminating the rich/lean changeover, and that the learning value KLR is renewed corresponding to the I value compensation of the air-fuel ratio feedback compensation coefficient.

on the other hand, the operation advances from the step S308 to a step S318 when the voltage V., output from the 02 sensor 24 is inverted in the reference time tO for discriminating the rich/lean changeover in the step S308.

In the step S318, the I value compensation flag FLAG2 is clarified (FLAG2 ±- 0), thereby counting up the counter value C of the counter in a step S319 (C - C + 1).

In a step S320, the program i5 completed by discrimination that the condition is not constant when the counter value C is under a predetermined number n such as three. In contrast, when the value C is over the I 1 1 1 number nr the constant condition is determined and operation advances to a step S321.

Namely, the constant condition is determined when the driving condition by the engine speed S. and the basic quantity TP are the substantially same as previously and when-the voltage VAp of the sensor 24 is inverted n times under the condition, thereby clarifying the counter (C ± 0) in the step S321.

In a step S322, there is calculated the differential amount Aa between an average value 7 while the feedback coefficient a is skipped in n times and the reference value ao (Aa (- ao - -j). In a step S323, the learning value KLR is retrieved from the proper address of the learning value table TBLR by using the basic quantity T p as the parameter read in the step S302, thereby advancing to a step S324.

in the step S324. the flag FLAG3 for renewing the I value compensation learning value is determined as to whether it is set or not. When 11FLAG3 = 011r namely, the voltage VAp of the sensor 24 is inverted in the reference time tO for discriminating the rich/lean changeover in the normal condition, and the learning value KLR is not renewed corresponding to the 1 value compensation, operation advances from the step S324 to a step S325. In the step S325, the new learning value KLR 'S set in dependency on the retrieved learning value KLR in the step S323 and the differential amount Aa calculated in the step S322 (XLR - KLR - Aa/Mr where character M is a constant determining rate for renewing the le arning value), thereby advancing to a step S328.

On the other hand, operation advances from the step S324 to a step S326 when "FLAG3 = V' in the step S324,, namely, the voltage V.. is not inverted in the reference time tO at the previous time and the learning value KLR is renewed by the f ine set value KLRSET step by step. In the step S326, the proper address of the learning table TBLR is renewed as a new learning value by subtracting 24 the differential amount Aa calculated in the step S322 from the learning value KLR retrieved in the step S323 (KLR ±- KLR - Aa), thereby advancing a step S327. In the step S327, the feedback compensation coefficient a-is set to Ila = 111, and operation advances to a step S328.

in the step S328, the learning value renewing flag FLAG3 for compensating the I value is clarified (FLAG3 0), and the program completes.

According to the present inventionf as described above, as the system of the present invention judges whether or not the output of the exhaust sensor is inverted in the reference time to change the integration constant, it is possible to prevent the deterioration of the air-fuel ratio when the feedback compensation coefficient reaches the uppermost and lowermost limits so that learning is not performed.

Namely, the present invention has the excellent effect that the learning compensation can be properly performed even when,. the actual air-fuel ratio is inconsistent with the theoretical air-fuel ratio, and that it is p6ssible to prevent the deteriorations of the exhaust emiss-ion, fuel cost, and drivability.

While the presently preferred embodiments of the present invention have been shown and described, it is to be understood these disclosures are for the purpose of illustration and that various changes and modifications may be made without departing from the scope of the invention as set forth in the appended claims.

1 i

Claims (5)

CLAIMS:

1. An air-fuel ratio learning control system for an automotive engine having an exhaust gas sensor and detecting means for detecting an engine speed, wherein said system sets a fuel injection quantity to which a basic fuel injection quantity is compensated and generated by an air-fuel ratio feedback compensation coefficient.based on an output of the exhaust sensor and a learning compensation coefficient based on said feedback compensation coefficient, and said system comprising: reference time setting means for setting a reference time with respect to an outDut inversion cycle of said exhaust gas sensor which is inverted with reference to a predetermined slice level in dependency on said engine speed and said basic fuel injection quantity; inversion discriminating means for discriminating inversion of said output of the exhaust gas sensor in said reference time which is set by said reference time setting means; feedback coefficient setting means for setting said air-fuel ratio feedback compensation coefficient in the manner that an integration constant of a-proportional integration control is set to a small value when the output of said sensor indicates without inversion; and learning means for setting said learning compensation coefficient in the manner that a learning value increases and decreases in a predetermined amount and in the same direction of said feedback coefficient corresponding to a decrease of said integration constant when said inversion discriminating means discriminates that said output of said sensor is not inverted in said reference time.

2. The system of claim 1, further comprising: discriminating means for discriminating a feedback condition completed in dependency on said output j 26 of the exhaust gas sensor and outputting a feedback condition discriminating signal to said feedback coefficient setting means; learning condition discriminating means for discriminating and outputting a learning condition to said reference time setting means in dependency on a feedback condition discriminating signal, said basic fuel injection quantity, and said engine speed; and compensating means for compensating said integration constant in dependency on a signal that said output of the exhaust gas sensor is determined to be inverted and for outputting a compensated constant to said feedback coefficient setting means.

The system of claim 1, wherein said learning means comprises: learning value renewing means f or renewing said learning value in dependency on said feedback coefficient, said basic fuel injection quantityr and previous learning value, when said inversion discriminating means discriminates that said output of the exhaust' gas sensor is inverted in a predetermined times in said reference time; learning value table f or storing said learning value which is renewed by said renewing means at a predetermined condition; and learning compensation coefficient setting means for setting said learning compensation coefficient in dependency on said basic fuel injection quantity and said learning value previously stored in said learning table.

4. The system of claim 1, further comprising: various increase compensation coefficient setting means f or setting various increase compensation coefficient in dependency on a coolant temperature detected by a coolant temperature sensor, a throttle opening degree signal detected by a throttle opening 1 I 27 degree sensor, and an output from an idling switch with respect to a coolant temperature compensation coefficient, acceleration and deceleration compensation coefficient, full-opening increase compensation coefficient, and an increase compensation coefficient after idling; voltage compensation coefficient setting means for setting a voltage compensation coefficient in dependency on a battery output voltage detected by a battery terminal voltage meter mounted on a battery; and fuel injection quantity setting means for setting an actual fuel quantity in dependency on said learning compensation coefficient set by said learning means, said feedback coefficient set by said air-fuel ratio compensation coefficient setting means, said basic fuel injection quantity, said various increase compe nsation coefficient, and said voltage compensation coefficient.

5. An air-,. uel ratio learning control system for an automotive engine substantially as hereinbefore described with reference.to and as shown in the accompanying drawings.