JPH02308949A - Output fluctuation detecting device for multi-cylinder engine - Google Patents

Abstract

PURPOSE:To precisely detect the output difference by detecting the output torque difference between cylinders at the engine rotating speed of each cylinder and at the average engine rotating speed, and prohibiting the detection of the output fluctuation when an engine is not controlled at the preset lean air-fuel ratio. CONSTITUTION:A fuel control means A controls the fuel feed quantity so that the air-fuel mixture fed to an engine is set to the lean side from the theoretical air-fuel ratio in the preset operation region where the output of the engine is not relatively required, e.g., in the light-load region. An output fluctuation detecting means E detects the output torque difference between cylinders based on detected results of a cylinder speed detecting means C and an average speed detecting means D and feeds it back to the fuel control means A. If a lean judging means B judges that the engine is not controlled at the preset lean air-fuel ratio, a prohibiting means F prohibits the detection of the output fluctuation. The output fluctuation is not detected under the operation condition when the effect of noise is large, thus the detection precision of the output fluctuation is improved, thereby the precision of the operation control can be improved.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a device for detecting an output torque difference between cylinders of a multi-cylinder engine.

[Conventional technology]

In multi-cylinder engines, output torque differences between cylinders due to manufacturing variations and aging are detected, and the output I of each cylinder is controlled by controlling the fuel injection amount and ignition timing for each cylinder.
- Devices designed to equalize the light are conventionally known (for example, Japanese Patent Application Laid-Open No. 59-201936).

According to the above-mentioned conventional device, each time the output torque of the engine is expressed as a function of the engine rotational speed, the engine rotational speed during the explosion stroke of each cylinder is compared with the average engine rotational speed of all cylinders. Detects the output torque difference between the two, corrects the fuel injection amount for each cylinder, and outputs]
- It is possible to reduce the fluctuation in torque.

[Problem to be solved by the invention]

However, in this method of detecting the output I-lux difference between cylinders from changes in engine speed, the third
As shown in the figure, the sensitivity of the rotational speed change to the torque change changes depending on the air-fuel ratio of the air-fuel mixture supplied to the engine, and the sensitivity tends to decrease as the air-fuel ratio becomes closer to Lynch.

Further, detection of the engine rotation speed in a normal electronic fuel injection control device is performed by a crank angle sensor built into the dispensing refuter for reasons such as cost and ease of mounting the sensor.

As shown in Fig. 4, the waveform of the rotational speed signal detected by this crank angle sensor is affected by hacklash between the distributor shirt I gear and the camshaft gear, engine load fluctuations due to uneven road surfaces, etc. The resulting noise component N is superimposed.

On the other hand, as a means of preventing exhaust pollution and reducing fuel consumption,
In light load ranges where engine output is relatively undemanding, the air-fuel ratio of the mixture supplied to the engine is controlled to be leaner than the stoichiometric air-fuel ratio, and during acceleration and in high load ranges, the air-fuel ratio is leaner than during normal driving. Lean burn systems that control this have been put into practical use in recent years. When the torque fluctuation detection device described above is applied to this lean burn system, the torque difference between the cylinders can be detected well when the engine is controlled at a lean air-fuel ratio, but as the air-fuel ratio approaches the lean There is a problem that the sensitivity of rotational speed changes to changes decreases and it becomes relatively susceptible to the effects of noise caused by huck rash, unevenness of the road surface, etc., resulting in the risk of incorrectly determining the torque difference between cylinders. .

The present invention detects the output torque difference between each cylinder with high precision in a multi-cylinder engine equipped with a fuel control means that controls the amount of fuel supplied so that the air-fuel ratio is leaner than the stoichiometric air-fuel ratio during normal running. An object of the present invention is to provide an output fluctuation detection device that can improve the accuracy of engine operation control.

[Means to solve the problem]

As shown in the configuration diagram of the invention in FIG. 1, the multi-cylinder engine output fluctuation detection device according to the present invention has an air-fuel ratio that is leaner than the stoichiometric air-fuel ratio in a predetermined operating region. An output fluctuation detection device for a multi-cylinder engine, comprising fuel control means A for controlling the fuel supply amount so that and cylinder-specific speed detection means C for detecting the engine rotation speed during the explosion stroke of each cylinder.
and average speed detection means for detecting the average engine rotation speed of all cylinders; and output fluctuation detection means E for detecting the output torque difference between the cylinders based on the engine rotation speed of each cylinder and the average engine rotation speed. The present invention is characterized by comprising a prohibition means F that prohibits detection of output fluctuations when the lean determination means determines that the engine is not controlled at a predetermined lean air-fuel ratio.

[Effect]

The fuel control means A controls the fuel supply so that the air-fuel ratio of the air-fuel mixture supplied to the engine is leaner than the stoichiometric air-fuel ratio in a predetermined operating range where engine output is relatively not required, for example in a light load range. Control quantity.

The output fluctuation detection means E detects the output torque difference between the cylinders based on the detection results of the cylinder-specific speed detection means C and the average speed detection means.

The prohibition means F prohibits detection of output fluctuations when the lean determination means B determines that the engine is not controlled at a predetermined lean air-fuel ratio.

Therefore, since the engine rotation speed under operating conditions where the influence of noise is large is not taken in as data for detecting output fluctuations, the accuracy of detecting output fluctuations is improved.

〔Example〕

Embodiments of the present invention will be described below with reference to the drawings.

FIG. 2 shows a four-cylinder engine to which an embodiment of the present invention is applied. In the figure, reference numeral 10 indicates an intake pipe connected to an air cleaner 12, and reference numeral 14 indicates a throttle valve provided in the middle of the intake pipe 10. The throttle valve 14 controls the amount of intake air in conjunction with an accelerator pedal (not shown).

The throttle switch 16 is connected to the rotating shaft of the throttle valve 14, and closes to generate a fully closed signal when the throttle valve 14 is in a fully closed state (idle position). This throttle fully closed signal is supplied to the input/output (1/○) boat 181 of the control circuit (ECU) +8.

A pressure sensor 22 is attached to a surge tank 20 connected to the intake pipe 10 to detect the pressure (absolute pressure) inside the intake pipe. The pressure sensor 22 outputs a voltage corresponding to the detected intake pipe internal pressure, and this output voltage is
It is supplied to an analog-to-digital (A/D) converter 18a of Ul8.

The surge tank 20 is connected to an intake manifold 24, and the intake manifold 24 is connected to the combustion chamber 26 of each cylinder.
connected to. Fuel injection valve 28 at the intake port of each cylinder
are attached to each. I1 from ECTJ18
An injection signal is supplied to each fuel injection valve 28 at an appropriate time for each cylinder in one cycle of the engine via the zero boat 18b and the drive circuit 18c, whereby each fuel injection valve 28 intermittently opens and closes. Pressurized fuel sent from a fuel supply system (not shown) is independently injected. A lean sensor 32 is attached to the exhaust pipe (or exhaust manifold) 30 and generates a current as shown in FIG. 5 depending on the concentration of oxygen components in the exhaust gas. The structure, characteristics, usage examples, etc. of such a lean sensor 32 are disclosed in Japanese Patent Application Laid-Open No. 58-143.
Since the iron is disclosed in Japanese Patent No. 108, etc., a detailed description of the iron will be omitted. The output of the lean sensor 32 is
After being subjected to current-voltage conversion by a conversion circuit 18d in l8, it is supplied to an A/D converter 18a.

The distributor 34 is rotationally driven by a camshaft (not shown), and distributes high voltage generated from an igniter 36 equipped with an ignition coil to the spark plugs 38 of each cylinder.
distributed and supplied to Igniter 36 is ECU1, I from 8
10 pole l-1, 81) and an ignition signal sent through the drive circuit 18h.

Inside the distributor 34 is a crank angle sensor 40.
42 is built-in. The first crank angle sensor 40 is arranged to face a timing rotor 46 that rotates together with the distributor shirt 1-44, and outputs a reference position signal every 720'CA (crank angle) of a crankshaft (not shown).

The second crank angle sensor 42 is disposed facing a timing rotor 48 that rotates together with the distributor shaft 44, and generates a pulse signal every 30'CA of the crankshaft. These signals are fed to the I10 port 18b of ECU1,8.

Note that 50 is a water temperature sensor attached to the cooling water system of the cylinder block to detect the temperature of the cooling water, and its output voltage is supplied to the A/D converters 1 and 8a of the ECU 18.

ECU18 includes the aforementioned A/D converter 18a, I10 boat 18b, drive circuits IElc and 18h, and conversion circuit]
, 8 (In addition to 1, there is a central processing unit (CP[I) 18
e, Random access memory (RAM) 18 f
, and read-only memory (IIOM) 18g. The A/D converter 18a also has a multiplexer function, and the pressure sensor 2
2, the voltage corresponding to the output current of the lean sensor 32, or the output voltage of the water temperature sensor 50 are selected and converted into a binary signal. The obtained binary signal, that is, the data representing the intake pipe internal pressure PM, the output LNSII of the lean sensor 32
The data corresponding to R and the data representing the cooling water temperature THW are R
It is stored in AM18f.

The ROM 18g stores in advance the control program and function table described later, and the ECU 8 detects the torque difference between each cylinder based on the signals from the above-mentioned sensors, and calculates the fuel injection amount for each cylinder. do.

Next, the operation of this embodiment will be explained using a flowchart.

FIG. 6 shows a control program for calculating the fuel injection amount TAUi, and the CPU 8e executes this processing routine at every predetermined crank angle, for example every 180° CA, during the main routine.

In step 100, the basic fuel injection amount TP is calculated from the engine rotational speed NE and the intake pipe internal pressure times.

In calculating the basic fuel injection amount TP, a two-dimensional map with the engine rotational speed NE and intake pipe internal pressure PM stored in advance in the ROM 18g as parameters is used. In the next step 102, the lean correction coefficient KLEAN
is calculated. This KLEAN is a correction coefficient for setting the target air-fuel ratio to a value on the leaner side than the stoichiometric air-fuel ratio, and RO
Engine speed NE stored in M18g in advance
It is determined from a two-dimensional map using the intake pipe internal pressure PM as a parameter.

If the target air-fuel ratio is the stoichiometric air-fuel ratio, KLEAN =
Set to 1.0.

In the next step 104, the fuel injection amount TA [Ii is the basic fuel injection amount 1'P, the air-fuel ratio feed bank correction coefficient FA
It is obtained from the following equation using F, lean correction coefficient KLEAN, I-lux fluctuation correction coefficient KTRQi, and other correction coefficients α and β.

TA[l1=TP・FAF −KLEAN −KT
Rlli·α+βFAF is a coefficient for performing closed-loop control of the air-fuel ratio, and is calculated in the processing routine shown in FIG. 7, which will be described later. In the next step 106, the determined fuel injection amount TAIIi is F,! It is stored in AM18f.

In the interrupt processing routine that is executed at each predetermined crank angle position of each cylinder, the injection start time and injection stop time are calculated from this fuel injection amount TAUi, and the injection signal is set to I10 during these times. It is output to the corresponding cylinder position and fuel injection is executed.

FIG. 7 shows a processing routine for calculating the air-fuel ratio feedback correction coefficient FAF based on the output LNSR of the lean sensor 32. CPU], 8e executes this processing routine at predetermined intervals during the main routine.

In step 200, it is determined whether closed-loop control execution conditions are satisfied. The closed-loop condition is not satisfied during engine startup, warm-up increase, high-load increase, etc., and the closed-loop condition is satisfied in other cases. If the closed loop condition is not satisfied (・), the process advances to step 202 and FAF=1.
0 and perform open loop control.

If it is determined that the closed loop condition is satisfied, the Ste knob 204
Then, the comparison reference value IR corresponding to the lean correction coefficient KLEAN obtained in the processing routine of FIG. 6 is calculated. R.O.
A KLEAN-IR function table as shown in FIG. 8 is stored in advance in M18g, and in step 204, the IR corresponding to KLEAN is determined using this function table. This lit is the output L of the lean sensor 32
This is a reference value for comparison of NSR, and by making this variable in accordance with the lean correction coefficient KLEAN, the target air-fuel ratio by closed loop control can be variably controlled in accordance with Kt and EAN.

In the next step 206, the output L of the lean sensor 32 is
N S It is compared with the comparison reference value IR to determine whether the current air-fuel ratio is richer or leaner than the target air-fuel ratio determined by the comparison reference value IR.

If L N S R≦IR, that is, if it is on the lynch side, steps 208 to 216 are performed. Step 20
In step 8, the skip flag CAFL used in steps 220 to 226 is reset to zero. Step 210
Then, it is determined whether the skit flag CAFR is 0 or not.

If you are transitioning from the Lean side to the Lynch side for the first time (:AP
Since R=0, the process proceeds to step 212, where the air-fuel ratio feedback correction coefficient FAF is decreased by SKP. Next, at the steering knob 214, the skip flag C/
Set IFR to 1. As a result, when the process of step 210 is executed next, the process proceeds to step 216 and is decreased by FAF. Here, SKP,
and are constants, and SkP is set to a much larger value than . SKP is FAF when it is determined that the air-fuel ratio has shifted from lean to lean relative to the target value.
This is to perform a process that greatly reduces the amount of data, that is, a skip process. Also, is for integral processing to gradually reduce FAR.

If LNSR>IR, that is, if it is on the lean side, steps 218 to 226 are performed. Step 218~
The processing at step 226 is similar to steps 208 to 216 described above except that the FAF is increased by 1 (2) if Sl<Pz.
The process is the same as that of . Step 202.212.216
．． 222 or 226] This FAF is stored in the RAM 18f in step 228.

FIG. 9 is a control program for detecting the output torque difference between each cylinder and calculating the torque fluctuation correction coefficient KTRQi, and the CPU 1i1e executes this processing routine at every predetermined crank angle during the main routine.

In step 300, the current predetermined cylinder is at compression top dead center (T
DC). This determination is made based on output signals from the crank angle sensors 4o and 42, as is conventionally known. That is, the reference position signal is set to be output, for example, at the compression TDC of the first cylinder, and from the detection of the reference position signal, 180.36Q,
540' After CA, compression 7 of cylinders 3, 4, and 2 will come. Step 30
If the answer is 0, the process proceeds to step 302, whereas if the answer is negative, this routine is temporarily terminated.

In step 302, the engine rotational speed NEi- in the explosion stroke is calculated from the time required for the cylinder that was at compression TDC immediately before the cylinder that is currently at compression TDC to rotate 180' CA. This engine speed NE
i-+ is proportional to the output torque of that cylinder.

In step 304, the cylinder counter j is incremented by 1, and in step 306, it is determined whether the calculation of the engine rotational speed NEi in step 302 has been completed for all cylinders. Step 30 for all cylinders
If the process of step 2 has been completed, the process advances to step 308, but if it has not yet been completed, this routine is temporarily ended. In this embodiment, since the engine has four cylinders, it may be determined in step 306 whether the cylinder counter i has exceeded four. Note that 1, 2, 3, and 4 of the cylinder counter i indicate the ignition order and correspond to cylinders No. 1, No. 3, No. 4, and No. 2, respectively.

In step 308, the output 1 of the lean sensor 32, NS
It is determined whether the engine is controlled at a predetermined lean air-fuel ratio based on whether I,' is greater than or equal to a predetermined value X.

If it is determined that the lean air-fuel ratio is being controlled, the process proceeds to step 310, where it is determined whether the engine rotational speed 11Ei of each cylinder calculated in step 302 is less than a predetermined value Y or not. Engine rotation speed NEi is a predetermined value Y
If it is smaller, steps 314 and subsequent steps are executed, but if a negative determination is made in step 308 or step 310, the process advances to step 328.

If the engine is not controlled at a predetermined lean air-fuel ratio, the sensitivity of changes in rotational speed to changes in output torque will be reduced, and the effects of hacking and noise from road surfaces will become relatively large. In the rotation range, the output l/lux difference between each cylinder decreases, and the influence of the torsion of the rotation system including the disc I/reducer 34 is large, so the output l/lux difference below step 314 decreases. detection processing cannot be performed with high precision. Therefore, in this case, in step 328, the integrated value SKi and the cycle counter,
8) is cleared to 0, and the cylinder counter i is cleared to 1 in step 330,
Exit this routine. On the other hand, when the engine is controlled at a predetermined lean air-fuel ratio and the engine is in the low-medium rotation range, the control in step 312 and subsequent steps 7 is executed.

In step 312, the average rotational speed NEAV during the explosion stroke of all cylinders is calculated. Then, in step 314, for each cylinder, the ratio Ki between the average rotation speed NEAV during the explosion stroke of all cylinders and the engine rotation speed NEi during the explosion stroke of that cylinder is determined. In other words, this ratio Ki is the average torque value of all cylinders and the 1-
It shows the ratio to

In step 316, the fuel supply is currently cut off (fuel
−) is determined. When the throttle valve is detected to be fully closed by the throttle valve and the engine rotation speed exceeds a predetermined value, it is determined that fuel is currently being cut, and steps 332 and subsequent steps are executed. However, at other times, it is determined that the fuel is not running, and steps 318 and subsequent steps are executed.

At step 332, the ratio Ki is integrated for each cylinder to obtain an integrated value 5KFCi, and the cycle counter Jf is incremented by one. That is, the integrated value is 5k
FCi is the summation of the ratio Ki between the average rotation speed during the explosion stroke of all cylinders during the fuel injection and the engine rotation speed during the explosion stroke of each cylinder. The integrated value 5KFCi of each cylinder during fuel cups 1 to 1 means the angle detection error of the crank angle sensor due to variations in the machining accuracy of the timing rotor 48, etc., since each cylinder is not in a combustion state.

In step 334, it is determined whether the cycle counter Jf has reached 50, that is, whether the ratio K) has been integrated for 50 cycles in step 332.

If the integration for 50 cycles has not been completed, the process proceeds to step 330, sets the cylinder counter i to 1, ends this routine, and then calculates the ratio Ki for all cylinders again.
(Step 314) is obtained. If the cycle counter Jf has reached 50 in step 334, the process advances to step 336, where the integrated value 5KFCi is changed to the integrated value SKF.
i, and the integrated value 5KFCi and the cycle counter Jf are cleared to 0.

In step 318, the ratio K of each cylinder during fuel supply is
i is integrated to obtain an integrated value SKi, and the cycle counter J is incremented by one.

The magnitude relationship of the integrated value SKi of each cylinder corresponds to the output torque difference of each cylinder.

In step 320, it is determined whether the cycle counter J has reached 50, that is, in step 318, the ratio Ki is 50.
It is determined whether or not the cycle has been integrated. If the integration for 50 cycles has not been completed, the program proceeds to step 330, sets the cylinder counter i to 1, and ends this routine, after which the ratio Ki (step 314) is determined for all cylinders again. If the cycle counter J has reached 50 in step 320, step 32
Proceed to step 2 and calculate the difference (SK +
SK;-1) and integrated value 5KFi during fuel cut,
5KFi-1 (SKFi-5KFi-1), and further calculates these differences, that is, the amount of change DSKi. This amount of change DSKi represents the amount of drop in torque that goes to that cylinder.

In step 324, it is determined whether the change 1l)SKi is equal to or greater than the set value 7. If the amount of change DSKi is equal to or greater than the set value 7, I-lux fluctuation correction is performed in step 326]
9 Coefficient KTRQi is increased by T and if the amount of change DSKj is smaller than the set value Z, step 326 is not executed and the current torque fluctuation correction coefficient MTRQi is maintained.

Step 324. Step 326 is executed for all cylinders, whereby the fuel injection amount is increased and corrected for the cylinder whose output torque is smaller than that of other cylinders.

In step 328, the integrated value SKi and the cycle counter J are cleared to 0, and in step 330, the cylinder counter i is set to 1.

As explained above, in this embodiment, when the engine is controlled at a predetermined lean air-fuel ratio and is in the low and medium rotation range, the average engine rotation speed of all cylinders and the explosion of each cylinder are calculated for each engine cycle (7206CA). The ratio Ki to the engine rotational speed during the stroke is determined, and the integrated value SKi for 50 cycles during fuel supply or fuel cut,
5KFi is required.

This integrated value is updated every 5 days. Next, SKi, 5Ki-1 for two cylinders with consecutive explosion strokes.
The difference (SKi 5Ki-+) and the integrated value 5KFi,
The difference between 5KFi-t (SKFi-5KFi-+) is calculated, and the change in these differences NDsKi is based on the set value.
If it is 2, it is determined that the output torque of the cylinder corresponding to the cylinder counter i is too small, and the fuel injection amount is increased.

As described above, the magnitude relationship of the integrated value 5KPi of each cylinder during fuel cut means an error in angle detection by the crank angle sensor since each cylinder is not in a combustion state. Therefore, if the 1 to 1 rook generated during the explosion stroke is uniform in each cylinder, the magnitude relationship of the integrated value SKi of each cylinder during fuel supply is the integrated value 5 of each cylinder during fuel power/throttle.
The size relationship is almost the same as that of KFi.

Figures 10(a) to (e) show cases in which the air-fuel ratio of the air-fuel mixture in each cylinder is made uniform and cases in which the air-fuel ratio of the air-fuel mixture in one cylinder is made leaner than that in other cylinders. , is the result of an experiment to determine the relationship between the integrated value SKi of each cylinder and the integrated value 5KFi at the time of fuel cut.

As shown in FIG. 10(a), the air-fuel ratio A/
F (solid line A), when the output power per herk is almost uniform, the integrated value 5ki during fuel supply (solid line B) and the integrated value 5KFi during fuel cut (broken line C) almost match in each cylinder, The amount of change DSKi (solid line D) takes a constant value for each cylinder. On the other hand, when the air-fuel ratio of one cylinder, for example, the No. 2 cylinder, is made lean to reduce the output l/lux, as shown in Fig. 10(b), the integrated value of 5 l during fuel supply is obtained. (i (solid line B) and the integrated value 5KFi (wave, vilC) at the time of fuel cut differ for each cylinder. Immediately, the integrated value SKi of the second cylinder becomes larger than the integrated value 5KFi, and the The integrated value SKi also changes, and as a result, only the amount of change DSKi of the second cylinder is clearly larger than the amount of change DSKi of the other cylinders. When the fuel ratio is made lean, Figures 10 (C), (d), and
As shown in (e), the amount of change DSKi for that cylinder increases. Therefore, if the output torque of a cylinder whose change amount DSKi is larger than a predetermined determination value is too small, it is possible to equalize the output torque of each cylinder by increasing the fuel injection amount of this cylinder.

In this example, the sensitivity of the rotational speed change to the I-lux change is reduced, resulting in a relatively hackish crash, an air-fuel ratio region where the influence of the road surface is large, and a high engine speed where the influence of the torsion of the rotational system including the distributor is large. It is configured so that the output 1-lux difference between each cylinder is not detected in the range. Therefore, erroneous detection of the output torque difference is prevented.

Furthermore, according to this embodiment, the angle detection error of the crank angle sensor is removed by determining the engine rotational speed at 1 to 10 hours of fuel consumption, so the output torque difference between each cylinder can be detected with high precision. .

As described above, according to this embodiment, the accuracy of detecting the output torque difference between the cylinders is improved, so that it is possible to uniformize the air-fuel ratio of each cylinder with high precision.

As a result, the difference in output and torque between each cylinder is reduced, so the torque fluctuation of the engine as a whole is reduced, which improves fuel efficiency by making the air-fuel ratio leaner.
The amount of NoX discharged can be reduced.

In order to adjust the output torque of each cylinder, it is not necessarily necessary to control the fuel injection amount as in the above embodiment, and the ignition timing may be controlled for each cylinder.

〔Effect of the invention〕

As described above, according to the present invention, the output difference between each cylinder can be detected with high accuracy, and it is therefore possible to improve the accuracy of engine operation control.

[Brief explanation of the drawing]

FIG. 1 is a block diagram of the invention; FIG. 2 is a diagram showing an engine to which an embodiment of the invention is applied; FIG. 3 is a diagram showing the relationship between the sensitivity of rotational speed change to torque change and the air-fuel ratio; Figure 4 is a diagram showing temporal changes in engine speed, Figure 5 is a characteristic diagram of the lean sensor, and Figures 6, 7, and 9 are flowcharts for explaining the operation of the control circuit in Figure 2. , Figure 8 shows KLEAN-
Characteristic diagram of IR function table, Figure 1O (a) to (e)
is a diagram showing the air-fuel ratio A/F of each cylinder, integrated value SKi, 5KFi, and variation DSKi. 1st - Control circuit, 32 - Lean sensor, 34 - Distributor, 40.42 - Crank angle sensor. Applicant: I-Yota Automobile Co., Ltd. t town month IF kidney (1)) Figure 10

Claims (1)

[Scope of Claims] Output of a multi-cylinder engine equipped with fuel control means for controlling the amount of fuel supplied so that the air-fuel ratio of the air-fuel mixture supplied to the engine is leaner than the stoichiometric air-fuel ratio in a predetermined operating range. A fluctuation detection device, comprising: lean determination means for determining whether or not the engine is controlled at a predetermined lean air-fuel ratio; cylinder-specific speed detection means for detecting the engine rotational speed during the explosion stroke of each cylinder; and all cylinders. an average speed detection means for detecting the average engine rotational speed of the engine; an output fluctuation detection means for detecting the output torque difference between the cylinders based on the engine rotational speed of each cylinder and the average engine rotational speed; and a lean determination means. An output fluctuation detection device for a multi-cylinder engine, comprising: prohibition means for prohibiting detection of output fluctuation when it is determined that control is not performed at a predetermined lean air-fuel ratio.