JPS6358255B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to an air-fuel ratio control method for correcting cylinder-to-cylinder air-fuel ratio differences in a multi-cylinder engine.

In the conventional air-fuel ratio control method using a known air-fuel ratio sensor that detects the air-fuel ratio from engine exhaust gas components, the stoichiometric air-fuel ratio (i.e. excess air ratio λ =
1.0), the output of the air-fuel ratio sensor changes quite rapidly, so the output of the air-fuel ratio sensor is integrated and feedback control is performed so that the air-fuel ratio becomes the stoichiometric air-fuel ratio (λ = 1.0).

However, in the case of an electronically controlled fuel injection engine, fuel is supplied by an electromagnetic fuel injection valve installed in each cylinder, but this electromagnetic fuel injection valve has manufacturing tolerances, and the injection hole area, There were variations in the number of turns of the electromagnetic coil, etc., which caused variations in the amount of fuel supplied. Furthermore, there are differences in the amount of intake air between cylinders in the opening and closing timings of the intake valve and exhaust valve of the engine, resulting in variations in the amount of air taken into each cylinder.

As a result, the air-fuel ratio varies between the cylinders, making it difficult to control the air-fuel ratio to the stoichiometric air-fuel ratio using the air-fuel ratio sensor, which deteriorates the purification rate of the three-way catalyst. Furthermore, the difference in air-fuel ratio between cylinders has an effect on engine misfires and the like, and has a particularly large effect on idling stability, which has been a hindrance to lowering the idling speed to improve fuel consumption.

The present invention has been made in view of the above problems, and
It is an object of the present invention to provide an air-fuel ratio control method that corrects variations in air-fuel ratio between cylinders of a multi-cylinder engine with a simple configuration that uses a single air-fuel ratio sensor.

In the preferred embodiment of the invention described below, a multi-cylinder engine with a fuel injector for each cylinder is equipped with a single air-fuel ratio sensor, and the air-fuel ratio sensor has a "rich" output and a "lean" output. After adjusting the injection amount of all cylinders so that the temporal ratio with the output expressed is an air-fuel ratio shifted by a predetermined value from the stoichiometric air-fuel ratio, the amount of fuel supplied to each cylinder is individually varied for a predetermined period of time. , the “richness” and “leanness” of the air-fuel ratio sensor during that predetermined period.
The temporal ratio of output is calculated and memorized sequentially, these calculated ratios are compared, and the fuel injection pulse width to the cylinder that changes the most with respect to the initially adjusted ratio of "rich" and "lean" output is corrected. By repeating similar operations, variations in air-fuel ratio between cylinders are corrected.

This example will be explained below. In FIG. 1, an engine 1 is a known four-cycle spark ignition engine mounted on an automobile, and intakes combustion air through an air cleaner 2, an intake pipe 3, and a throttle valve 4. Further, fuel is supplied from a fuel system (not shown) through electromagnetic fuel injection valves 5 provided corresponding to each cylinder. The exhaust gas after combustion passes through the exhaust manifold 6, the exhaust pipe 7, the oxidation catalytic converter 8, etc., and is released into the atmosphere. The intake pipe 3 includes a potentiometer-type intake air amount sensor 11 that detects the amount of intake air taken into the engine 1 and outputs an analog voltage according to the amount of intake air, and a potentiometer-type intake air amount sensor 11 that detects the temperature of the air taken into the engine 1 and outputs an analog voltage according to the amount of intake air. A thermistor-type intake air temperature sensor 12 is installed that outputs an analog voltage (analog detection signal) depending on the air temperature. The engine 11 is also equipped with a thermistor-type water temperature sensor 13 that detects the coolant temperature and outputs an analog voltage (analog detection signal) according to the coolant temperature, and the exhaust manifold 6 is equipped with an oxygen concentration sensor in the exhaust gas. Detects the air-fuel ratio from the air-fuel ratio, and outputs a voltage of about 1 volt (high level) when the air-fuel ratio is smaller than the stoichiometric air-fuel ratio (rich), and about 0.1 volt (low level) when it is larger than the stoichiometric air-fuel ratio (lean). A sensor 14 is installed. The rotational speed (number) sensor 15 detects the rotational speed of the crankshaft of the engine 1 and outputs a pulse signal with a frequency corresponding to the rotational speed.

The cylinder discrimination sensor 16 outputs a pulse signal at the time of injection of the first cylinder, which serves as a reference, in order to specify each cylinder to which fuel is supplied according to the combustion order of each cylinder.

The control circuit 20 is a circuit that calculates the fuel injection amount based on the detection signals of the sensors 11 to 16, and adjusts the fuel injection amount by controlling the opening time of each cylinder of the electromagnetic fuel injection valve 5.

The control circuit 20 will be explained with reference to FIG.
100 is a microprocessor (CPU) that calculates the fuel injection amount. Reference numeral 101 is a rotation number counter that counts the engine rotation number based on a signal from the rotation speed (number) sensor 15.
Further, this rotation number counter 101 sends an interrupt command signal to the interrupt control section 102 in synchronization with the engine rotation. When interrupt control section 102 receives this signal, it outputs an interrupt signal to microprocessor 100 via common bus 150. 103
is a digital input port (not shown), but it has a built-in discrimination circuit that determines whether the air-fuel ratio is above the stoichiometric air-fuel ratio (lean) or below (rich) based on the signal from the air-fuel ratio sensor 14. A digital binary signal, such as a signal from a cylinder discrimination sensor 16 that generates a pulse signal indicating the cylinder to be supplied, is transmitted to the microprocessor 100. 104
is an analog input port consisting of an analog multiplexer and an A-D converter.
It has a function of converting the data into D and sequentially reading it into the microprocessor 100. Each of these units 101,
Output information from 102, 103, and 104 is transmitted to microprocessor 100 through common bus 150. 105 is a power supply circuit and RAM 10 will be described later.
7. 17 is a battery, and 18 is a key switch, but the power supply circuit 105 is directly connected to the battery 17 without passing through the key switch 18. Therefore, power is always applied to the RAM 107, which will be described later, regardless of the key switch 18. 1
06 is also a power supply circuit, but it is connected to a battery 17 through a key switch 18. Power supply circuit 10
6 supplies power to parts other than the RAM 107, which will be described later. Reference numeral 107 is a temporary memory unit (RAM) that is used temporarily during program operation, but as mentioned above, power is always applied regardless of the key switch 18, so even if the key switch 18 is turned off and engine operation is stopped, the memory contents are lost. It is configured as non-volatile memory. A read-only memory (ROM) 108 stores programs, various constants, and the like. Reference numeral 109 is a fuel injection time control counter including a register, which is composed of a down counter, and converts the digital signal representing the opening time of the electromagnetic fuel injection valve 5 calculated by the microprocessor (CPU) 100, that is, the fuel injection amount, to the actual electromagnetic The pulse signal is converted into a pulse signal with a pulse time width corresponding to the opening time of the fuel injection valve 5. 110 is a power amplification section that drives the electromagnetic fuel injection valve 5. 111 measures the elapsed time with a timer and transmits it to the CPU 100.
The rotational speed counter 101 measures the engine rotational speed once every half rotation of the engine based on the output of the rotational speed sensor 15, and when the measurement is finished, the interrupt control unit 102
provides an interrupt command signal to the The interrupt control unit 102 generates an interrupt signal from the signal, and causes the microprocessor 100 to execute an interrupt processing routine for calculating the fuel injection amount.

FIG. 3 shows a schematic flowchart of the microprocessor 100, and the functions of the microprocessor 100 will be explained based on this flowchart, as well as the operation of the entire configuration. The key switch 18 and a starter switch (not shown)
When the engine is turned on and the engine starts, the first step
After the start of step 1000, at step 1000A, to determine whether the battery 17, which will be described later, has been removed.
Check the contents of a specific address in RAM107,
If the contents of the RAM 107 are not normal, it is determined that it is necessary to re-adjust the variations in each cylinder, and the process proceeds to step 1000B, where the constants K ₄₁ to _{K 46} are initialized to 1.0. After that, the arithmetic processing of the main routine is started, and initialization processing is executed in step 1001, where the constants K ₂ = 1.0, K ₃₁ to _{K 36} = 1.0, and K ₅ = 1.0 are set, and at the same time they are used for determining various conditions. Set the flags to E=0, F=0, G=0, H ₁ = 1, H ₂ to _{H 6} =
0, J=0, M=0. Step 1002
At this step, analog signals corresponding to the cooling water temperature and intake air temperature from the analog input port 104 are A/D converted and digital values obtained are read. In step 1003, correction values are calculated according to the cooling water temperature and intake air temperature based on the results.
K ₁ is calculated and the result is stored in RAM 107. In step 1004, it is determined whether the cooling water temperature is above a predetermined temperature and the air-fuel ratio sensor is in an active state, which are the starting conditions for air-fuel ratio feedback control.If these air-fuel ratio feedback conditions are not met, the process branches to NO. Then go to step 1005 and
At step 1005, the air-fuel ratio feedback correction coefficient K ₂ is set to 1.0, and the process returns to step 1002. Step 1004
When the feedback condition is satisfied in
Branch to YES and proceed to step 1006. step
In step 1006, the opening of the throttle valve is detected in conjunction with the throttle valve 4 to detect an idling state (throttle valve fully closed), and the idling state is longer than a predetermined time (for example, 10 seconds) at a vehicle speed of 0 km/h.
If two or more of the correction coefficients K ₄₁ to _{K 46} are 1.0 and the air-fuel ratio feedback is being executed, it is assumed that the learning condition for correcting cylinder variations in the air-fuel ratio is satisfied, and the process branches to YES and the air-fuel ratio variations are corrected in step 1010. The calculation processing of correction coefficients K ₄₁ to _{K 46} for correcting is performed and the results are stored in the RAM 107.

Figures 4A to 4D show this correction coefficient K ₄₁ to _{K 46}
This is a detailed flowchart of step 1010 in FIG. 3, which performs arithmetic processing and stores it in the RAM 107, that is, performs storage processing. In step 401, it is determined how many of the correction coefficients K ₄₁ to _{K 46} stored in the RAM have a coefficient of 1.0, and the number M is determined. In step 402, it is determined whether the number M determined in step 401 is 2 or more, that is, whether there are two or more cylinders that have not been corrected using the correction coefficients _K41 to _K46 , and if there is only one cylinder that has not been corrected. Then, the process branches to NO and the correction calculation process for correction coefficients K ₄₁ to K ₄₆ ends.
This is to leave only one cylinder with the most average air-fuel ratio as the reference cylinder to prevent the air-fuel ratio from deviating from the target air-fuel ratio. If there are two or more cylinders that have not been corrected, branch to YES to perform correction calculations.
Proceed to 403. In step 403, the air-fuel ratio is shifted slightly to the richer side than the stoichiometric air-fuel ratio to find the cylinder with the leanest air-fuel ratio (E=
0) or conversely, to shift slightly to the lean side from the stoichiometric air-fuel ratio to find the cylinder with the richest air-fuel ratio (E=1). Note that E=0 due to the processing of initialization step 1001.
Since the process is started from the beginning, the process branches to YES at the beginning and proceeds to step 404. Step 404 is a flag for checking whether the time ratio between the period representing "rich" and the period representing "lean" in the air-fuel ratio sensor output has reached a predetermined ratio. If so, the process branches to YES and starts correction calculation for each cylinder from step 419. If G=0, that is, the predetermined ratio has not been reached, the process branches to NO and proceeds to step 405.

Does F in step 405 indicate the start of setting the temporal ratio of the "rich" and "lean" outputs of the air-fuel ratio sensor 14 to a predetermined ratio (F=0) or is it in the process of setting it (F=1)? ), if F=0, that is, the setting has started at a predetermined ratio, the process branches to step 406, and steps 406 and 407 stop the air-fuel ratio sensor feedback, and store the air-fuel ratio feedback correction coefficient at the time of stop in the RAM 107. K ₅
and store it in a location assigned to K ₂ =
Set to 1.0. Step 408 begins measuring the number of injections (or number of ignitions) to begin measuring the time ratio of the "rich" and "lean" outputs of the air-fuel ratio sensor to a predetermined value. Step 409 is F because initialization is completed from step 406 to step 408.
= 1 and thereafter jumps from step 405 to step 410 until the predetermined number of injections is counted.

Step 410 branches to NO and returns to step 1002 until the predetermined number of injections, which was started in step 408, reaches the predetermined number of injections. When the predetermined number of injections reaches a predetermined number (preferably an integral multiple of the number of cylinders in the engine in order to accurately determine the ratio of "rich" and "lean"), the process branches to YES and proceeds to step 411, where the microcomputer 100 The ratio is calculated from the "rich" and "lean" output count values of the air-fuel ratio sensor 14 read in synchronization with the clock, and is stored in location C of the RAM 107. In step 412, the calculated value stored in C of the RAM 107 is within a predetermined ratio of "rich" and "lean". For example, ("rich" count value) / ("rich" count value + "lean" count value) = 80
Determine whether the ratio is within ~90%, and if it is not within that ratio, branch to NO, and set the target ratio in step 413.
Compared with B _R = 85%, if it is larger than 85%, that is, richer than the target, the process branches to step 414 and the correction coefficient K ₅
By subtracting △ _K5 from △K5 and reducing the correction coefficient _K5 , it becomes slightly lean and flag F is set in step 416.
By setting = 0, steps 405 to 409 are performed and injection is performed again a predetermined number of times, and thereafter, the process is repeated until the ratio of "rich" and "lean" outputs reaches a predetermined value. If the target ratio is smaller than 85% in step 413, that is, leaner than the target ratio, branch to NO and proceed to step
415 and slightly richer correction as K ₅ = K ₅ + △K ₅ . After setting F=0 in step 416, the process returns to step 1002, passes through steps 405 to 409, and repeats in the same manner until the ratio of "rich" and "lean" outputs reaches a predetermined value.

If C is within the specified range in step 412, step 417
Branch to , set flag F = 0, and proceed to step 418.
By setting G to 1, the setting of the air-fuel ratio to the target value is completed, so from next time onwards, the step to determine the leanest cylinder will be performed.
Jump from 404 to step 419. step
Step 419 determines whether or not to slightly enrich only the first cylinder, and in the initialization process of step 1001, initially H ₁ = 1, branching to YES, and setting the correction coefficient K ₃₁ to the first cylinder, for example, K. ₃₁ = 1.02
Perform calculations as . A detailed flowchart of this step 420 is shown in FIG. When the first cylinder correction _K31 calculation process is started in step 420, it is determined in step 501 whether it is the beginning or middle of the first cylinder correction process, and if it is the first, it is determined that J=0 was set in the initialization in step 1001. Branch to NO and step
502 and the correction coefficient of the first cylinder K ₃₁ = K ₃₁ + △K ₃
(For example, K ₃₁ = 1.00 + 0.04 = 1.04), and the correction coefficients K ₃₂ to _{K 36} of the other cylinders are kept at 1.00, so that only the first cylinder has a slightly rich air-fuel ratio. When the first cylinder correction _K31 is started in step 503, the measurement of the number of injections is started. By setting the flag J=1 in step 504, the correction initialization of the first cylinder is completed, and when step 504 is completed, the process returns to step 1002. After that, in step 501, since J=1, branch to YES and step
Proceed to 505. Step 505 continues until the measurement of the number of injections started in step 503 reaches a predetermined number.
Branches to NO and returns to step 1002. step
When the predetermined number of injections is reached in step 505, the process branches to YES in step 506, where the temporal ratio of the "rich" and "lean" outputs of the air-fuel ratio sensor 14 during the number of injections in step 505 is calculated and stored in the RAM 107. In step 507, since the correction process for the first cylinder has been completed, the flag H ₁ is set to 0 and the engine ignition order (for example, in a 6-cylinder engine, 1, 5, 3, 6, 2,
4 cylinders), a flag for performing correction processing for the fifth cylinder is set (H ₂ =1).
When step 507 is completed, the process returns to step 1002.
After completing the processing shown in FIG. 5 above, in step 419 H ₁ =
Since it is 0, the process branches to NO and proceeds to step 421.
At step 421, since H ₂ =1 was set at step 507, the process branches to YES and at step 422, the fifth
Cylinder correction calculation processing is performed in the same manner as in FIG.
From step 423 to step 430, correction calculation processing is performed for each cylinder in the same way.

In step 431, the cylinder having the highest "rich" to "lean" ratio, that is, the richest cylinder, is found among the ratios of the first to sixth cylinders stored in the RAM 107, and is stored in a location in the RAM 107. Step 432 is a step
For the richest cylinder found in 431, as shown in Figure 6 (the figure shows the case where the fifth cylinder is corrected), correct by △K ₄ at pulse width T ₁ with pulse width T ₂ as the axis. characteristics.
Step 433 then sets the cylinder slightly leaner to find a richer cylinder and sets a flag to perform correction processing.
Since E=1, proceed to step 434, and proceed to step 404.
From this, correction processing is performed using the same concept as 433.

FIG. 7 shows a part of steps from step 404 to step 433, with the number of ignitions plotted on the horizontal axis and the air-fuel ratio sensor and the air-fuel ratio of each cylinder. The number of engines in this case is 6, and the number of firings for calculating the ratio of "rich" to "lean" power is 12 firings. When the ratio of "rich" and "lean" outputs between 1st to 12th ignition reaches a predetermined ratio, the correction coefficient △K ₃ for only the 1st cylinder is changed between 13th to 24th ignitions to K ₃₁ = K ₃₁ + △ Make it rich by adding K ₃ . As is clear from the figure, the first
Since the cylinder is rich, even if the ratio between the "rich" and "lean" outputs during this period is calculated, the output of the air-fuel ratio sensor 14 will not change because the "rich" output is changed to rich. However, if the amount of the 5th cylinder is increased by △K ₃ between the 25th and 36th ignitions, as you can see from the air-fuel ratio in Figure 5, the 5th cylinder will also become rich, and all cylinders will be rich, and the air-fuel ratio sensor will always read " Rich” output signal.
Similarly, even if the third cylinder, the sixth cylinder, the second cylinder, and the fourth cylinder are rich, the same result as the first cylinder is obtained, and only the fifth cylinder has all the air-fuel ratio sensor outputs showing "rich" output signals, As a result, it can be seen that the fifth cylinder is lean. As a result, a correction calculation of K ₄₂ is performed on the correction coefficient K ₄₂ and stored in the RM 107 .

FIG. 8 shows part of the process in step 434 for obtaining a rich cylinder by a slightly lean set. In FIG. 3, if the learning condition is not satisfied in step 1006, the process advances to step 1007 K ₃₁ to _{K 36}
are all coefficients of 1.0. In step 1008, the correction coefficient K ₅ is set to 1.0 during air-fuel ratio feedback. Step 1009 is the correction coefficient during air-fuel ratio feedback.
This is a calculation process for K ₂ , and K ₂ is determined by integral calculation based on the output of the air-fuel ratio sensor. step
When 1009 or 1010 is completed, the process returns to step 1002.

Note that the initialization process at step 1000A executes the following. In other words, the battery may be removed during vehicle inspection or repair. For this reason, RAM10
The correction amount _K4 stored in 7 may be corrupted and become a meaningless value. Therefore, in order to detect whether or not the battery has come off, a constant with a predetermined pattern is usually stored at a specific address in the RAM 107. When the program starts, it is determined whether the value of this constant is corrupted or incorrect, and if it is an incorrect value, it is assumed that the battery has been removed, and the program proceeds to step 1000B, where the correction amount K ₄₁ is determined.
Initialize all values of ~K ₄₆ to the initial value 1.0 and reset the constants of the determined pattern. If the pattern constant is not broken at the next startup, the process proceeds to step 1001 without initializing _K41 to _K46 .

Normally, the main routine processes 1002 to 1009 or 1010 are repeatedly executed according to the control program. When the interrupt signal for calculating the fuel injection amount is input from the interrupt control section 102, the microprocessor 100 immediately interrupts the main routine even if it is processing the main routine and moves to the interrupt processing routine at step 1020. In step 1021, a signal representing the engine speed N from the rotation speed counter 101 is taken in, and a signal representing the intake air amount (intake air amount) Q and a cylinder discrimination signal 16 specifying the cylinder to be injected are taken in from the analog input port. In step 1022, the rotational speed N and intake air amount Q are temporarily stored in the RAM 107 for use as parameters for correction storage processing of the correction amount _K2 in the main routine calculation processing. Next, in step 1023, the basic fuel injection amount (that is, the injection time t of the electromagnetic fuel injection valve 5) determined from the engine speed N and the intake air amount Q is calculated. The calculation formula is t=
F×Q/N (F: constant). Next, in step 1024, the fuel injection correction coefficients (K ₁ , K ₂ , K ₃₁ to _{K 36} , K ₄₁ to K ₄₆ , K ₅ ) obtained in the main routine are
It reads out from the RAM 107 and performs correction calculation of the injection amount (injection time width) that determines the air-fuel ratio.

The formula for calculating the injection time width T is It is. Here, K ₃₁ to K ₃₆ and K ₄₁ to K ₄₆ are calculated using correction coefficients corresponding to each cylinder.

For example, in the case of calculation for the fifth cylinder, T=t×K ₁ ×K ₂ ×K ₃₂ ×K ₄₂ ×K ₅ .

However, when the learning conditions of step 1006 are satisfied in the main routine, K ₂ =1.0. Also, during air-fuel ratio feedback, K ₃₁ to _{K 36} and K ₅ are 1.0.
It is. Next, in step 1025, the corrected and calculated fuel injection amount data is set in the counter 109. Next, the process advances to step 1026 to return to the main routine. When returning to the main routine, the process returns to the processing step at which it was interrupted due to interrupt processing. The functions of the microprocessor are as described above. In the manner described above, the fourth correction amounts K ₄₁ to K ₄₆ can be corrected to eliminate the air-fuel ratio difference between the cylinders.

In the case of a 6-cylinder engine, when the correction (K ₄ ) of five cylinders is completed, the correction coefficient is set in step 1006.
Since 1.0 among K ₄₁ to K ₄₆ is one, the flow branches to NO and thereafter no correction calculation processing is performed for K ₄₁ to _{K 46} , and thereafter only air-fuel ratio feedback is performed even at idle.

In the above embodiment, due to the manufacturing of the electromagnetic fuel injection valve 5, the error rate of the injection amount becomes larger when the injection amount is small (when the pulse width T is small). Since the variation was large, correction was performed only by idling, but correction in other driving ranges may be performed in combination with a constant running device such as auto drive.

Furthermore, although the period for determining the ratio of the "rich" and "lean" outputs of the air-fuel ratio sensor 14 is set to be the period of a predetermined number of injections (number of ignitions), the same effect can be obtained even if the ratio is determined over a fixed period of time. can. As described above, the present invention provides a method in which a multi-cylinder engine having a fuel injection valve for each cylinder is equipped with a single air-fuel ratio sensor, and the air-fuel ratio is controlled by the output signal of this air-fuel ratio sensor according to the exhaust gas components of the engine. In addition, by correcting the independent correction coefficient for each cylinder, the air-fuel ratio difference between each cylinder is controlled to be eliminated, thereby improving the controllability of the air-fuel ratio feedback and improving the purification rate of the three-way catalyst. In addition, idling stability is improved, which lowers idling speed and improves fuel consumption.Furthermore, as mentioned above, a single air-fuel ratio sensor can control the air-fuel ratio of multiple cylinders, which improves air-fuel ratios. The number of fuel ratio sensors can be reduced, cost can be reduced, and the installation configuration of the air-fuel ratio sensors in the exhaust system can be simplified. This has an excellent effect.

[Brief explanation of drawings]

FIG. 1 is a schematic diagram showing the overall configuration of the present invention. FIG. 2 is a block diagram showing a schematic configuration of the control circuit shown in FIG. 1. FIG. 3 is a schematic flow chart showing the operation of the microprocessor shown in FIG. 4A to 4D are detailed flowcharts of main parts of the flowchart shown in FIG. 3. FIG. 5 is a more detailed flowchart of the main part of the flowchart shown in FIG. 4B. FIG. 6 is a characteristic diagram for explaining air-fuel ratio correction for each cylinder. FIG. 7 is a waveform diagram of various parts showing how to obtain a lean cylinder. FIG. 8 is a waveform diagram of various parts showing how the rich cylinder is determined. 1...Engine, 11...Air amount sensor, 14
... Air-fuel ratio sensor, 15 ... Rotation speed sensor, 16
... Cylinder discrimination sensor, 20 ... Control circuit, 1
00...Microcomputer (CPU), 107
...A temporary storage unit (RAM) that forms non-volatile memory.

Claims (1)

[Claims] 1. The air-fuel ratio of the air-fuel mixture supplied to the engine is determined from the exhaust gas components sequentially discharged from each cylinder of the engine, which has a fuel injection valve for each cylinder and an air-fuel ratio sensor common to each cylinder. The air-fuel ratio is sequentially detected by the air-fuel ratio sensor, and the air-fuel ratio detected by the air-fuel ratio sensor is determined in correspondence with the cylinder of the engine to which the air-fuel mixture is supplied, and when the engine is in a predetermined steady operating state, Among the plurality of fuel correction amounts stored in memory for each cylinder, the fuel correction amount corresponding to the determined cylinder is corrected according to the detected value of the air-fuel ratio, and the intake air amount, rotation speed, etc. of the engine are adjusted. The fuel amount calculated according to the operating state is corrected using the fuel correction amount in the memory corresponding to the cylinder to which the air-fuel mixture is to be supplied, and the fuel injection valve for that cylinder is injected into that cylinder by the corrected fuel amount. An air-fuel ratio control method comprising: controlling the air-fuel ratio for each cylinder by correcting the fuel correction amount for each cylinder of the engine.