0, the #4 cylinder is the first reference cylinder c1, the #3 cylinder is the second reference cylinder c2, and the #2 cylinder is the third reference cylinder c3. When the #3 cylinder is the subject cylinder c0, the #2 cylinder is the first reference cylinder c1, the #1 cylinder is the second reference cylinder c2, and the #4 cylinder is the third reference cylinder c3. The subject cylinder c0 is sequentially switched every time the crank angle changes by 180°.

An “O” mark in FIG. 4 indicates a change in the misfire parameter α in a case where no misfire occurs in all the cylinders c0 to c3, that is, in the case of a normal cycle in which all the cylinders c0 to c3 are normal cylinders. A “x” mark indicates a change in the misfire parameter α in a case where misfire has occurred in some cylinders, that is, in the case of a misfire cycle including misfire cylinders. The misfire parameter α varies to some extent every cycle, and the “0” mark and the “x” mark are the average value or the median value of the misfire parameter α in consideration of the variation.

The misfire pattern includes a single-cylinder misfire in which only the subject cylinder c0 is misfired and a multi-cylinder misfire in which the subject cylinder c0 and any of the first reference cylinder c1, the second reference cylinder c2, and the third reference cylinder c3 are misfired. The multi-cylinder misfire is divided into a facing misfire, a subsequent continuous misfire, and a preceding continuous misfire. In the facing misfire, the subject cylinder c0 and the second reference cylinder c2 in which the order of the combustion stroke is not continuous with the subject cylinder c0 are misfired. In the subsequent continuous misfire, the subject cylinder c0 and the third reference cylinder c3 in which the order of the combustion stroke is continuous to the subject cylinder c0 and which undergoes the combustion stroke after the subject cylinder c0 are misfired. In the preceding continuous misfire, the subject cylinder c0 and the first reference cylinder c1 in which the order of the combustion stroke is continuous to the subject cylinder c0 and which undergoes the combustion stroke before the subject cylinder c0 are misfired.

In FIG. 4, a cylinder with a small combustion torque, that is, a cylinder with weak combustion is indicated by hatching. The misfire pattern of FIG. 4 is regularly repeated. Therefore, for example, in the subsequent continuous misfire, misfire occurs in the subject cylinder c0 and the third reference cylinder c3 next to the subject cylinder c0. If the current subject cylinder is the #3 cylinder and the subsequent continuous misfire occurs, the misfire repeatedly occurs in the #3 cylinder and the #4 cylinder in the combustion cycle of the engine 1. In FIG. 4, the subject cylinder c0 is a cylinder of weak combustion, and the misfire has occurred in cylinders of weak combustion (hatched region) including the subject cylinder c0 in all misfire patterns.

When misfire occurs in the subject cylinder c0, the base rotational speed (the rotational speed calculated in the vicinity of the compression top dead center) RV0 at the time of calculating the misfire parameter of the third reference cylinder c3 to be the next subject cylinder c0 decreases. Therefore, as illustrated in FIG. 4, in all misfire patterns, the misfire parameter α of the third reference cylinder c3 is larger than the misfire parameter α of the subject cylinder c0. In addition, the misfire parameter α (“x” mark) when misfire occurs in the subject cylinder c0 is smaller than the misfire parameter α (“O” mark) at the time of normal combustion for the subject cylinder c0. However, in the multi-cylinder misfire, the misfire parameter α of the subject cylinder c0 is affected by the misfire of the other cylinders c1 to c3, and thus the amount of decrease in the misfire parameter α is smaller than that at the time of the single-cylinder misfire. In particular, in the case of the preceding continuous misfire, the base rotational speed RV0 at the start of the combustion stroke in the subject cylinder c0 decreases due to the strong influence of the misfire in the immediately preceding first reference cylinder c1, and thus the relative rotational speed RV1 increases, and the misfire parameter α of the subject cylinder c0 tends to increase.

Focusing on a difference Δα between the misfire parameter α (minimum value of the variation of the misfire parameter) of the subject cylinder c0 in the normal cycle and the misfire parameter α (maximum value of the variation of the misfire parameter) of the subject cylinder c0 in the misfire cycle, the magnitude of the difference Δα varies depending on the pattern of misfire, and the difference Δα is the smallest in the preceding continuous misfire. As a result, there is a possibility that the misfire cannot be accurately determined only by comparing the magnitude of the misfire parameter α with the threshold A1 (FIG. 3B). That is, when the threshold A1 is set excessively small, it may be erroneously determined that the combustion is normal although the misfire has occurred, and when the threshold A1 is set excessively large, it may be erroneously determined that the misfire has occurred although the combustion is normal. Therefore, in order to accurately determine the presence or absence of misfire even at the time of weak combustion or the like, the present embodiment configures a misfire determination apparatus as follows.

FIG. 5 is a block diagram illustrating a configuration of main portion of the misfire determination apparatus 10 according to the present embodiment. As illustrated in FIG. 5, the misfire determination apparatus 10 includes a crank angle sensor 21, a storage unit 30, a parameter calculation unit 40, and a determination unit 50. The storage unit 30, the parameter calculation unit 40, and the determination unit 50 are functional configurations of the controller 20. Therefore, the misfire determination apparatus 10 is configured by the crank angle sensor 21 and the controller 20 including a CPU (microprocessor) and memory connected to the CPU in terms of hardware.

The storage unit 30 is a buffer memory which temporarily stores the crank signal from the crank angle sensor 21. The storage unit 30 stores data of the generation time interval of the crank signal in a range from the current crank angle to the crank angle which is at least one cycle (720°) back therefrom, that is, the latest crank angle data for one cycle. In other words, the crank angle data that makes it possible to calculate the change in the relative rotational speed RV1 and the misfire parameter α for each of the subject cylinder c0, the first reference cylinder c1, the second reference cylinder c2, and the third reference cylinder c3 as illustrated in FIG. 4 is stored.

The parameter calculation unit 40 includes a base parameter calculation unit 41 and a sum parameter calculation unit 42. As described above, on the basis of the crank angle data stored in the storage unit 30, the base parameter calculation unit 41 calculates the relative rotational speed RV1 of each cylinder in the combustion stroke of the engine 1 and calculates the misfire parameter α of each cylinder by integrating the relative rotational speed RV1 over the combustion stroke (expansion stroke) of each cylinder. That is, the misfire parameter α of each of the subject cylinder c0, the first reference cylinder c1, the second reference cylinder c2, and the third reference cylinder c3 is calculated. In FIG. 5, the misfire parameters α of the subject cylinder c0, the first reference cylinder c1, the second reference cylinder c2, and the third reference cylinder c3 are represented by α0, α1, α2, and α3, respectively.

The misfire parameter α0 may be stored in the storage unit 30 every time the base parameter calculation unit 41 calculates the misfire parameter α0 of the subject cylinder c0, and the misfire parameters α0, α1, and α2 may be changed to α1, α2, and α3, respectively, every time the cylinder in the combustion stroke moves. As a result, since the misfire parameters α of the first reference cylinder c1, the second reference cylinder c2, and the third reference cylinder c3 are already stored in the storage unit 30 at the time of calculating the misfire parameter α of the subject cylinder c0, the base parameter calculation unit 41 only needs to calculate the misfire parameter α0 of the subject cylinder c0, and a processing load can be reduced.

The sum parameter calculation unit 42 includes addition circuits 43 to 45 which perform the addition processing of the misfire parameter α. The addition circuit 43 adds the misfire parameter α0 of the subject cylinder c0 and the misfire parameter α1 of the first reference cylinder c1 to calculate a sum parameter α11. The addition circuit 44 adds the misfire parameter α0 of the subject cylinder c0 and the misfire parameter α2 of the second reference cylinder c2 to calculate a sum parameter α12. The addition circuit 45 adds the misfire parameter α0 of the subject cylinder c0 and the misfire parameter α3 of the third reference cylinder c3 to calculate a sum parameter α13.

The determination unit 50 includes comparison circuits 51 to 55 which compare the misfire parameter α with the thresholds A1 to A4, an OR circuit 56, and an AND circuit 57. In FIG. 5, among the plurality of cylinders c0 to c3, a cylinder which is a subject for the misfire determination is indicated by a black circle, and the other cylinders are indicated by white circles between the parameter calculation unit 40 and the determination unit 50. The comparison circuit 51 determines the magnitude of the misfire parameter α0 of the subject cylinder c0 and the predetermined threshold A1. The threshold A1 (FIG. 3B) is set in consideration of the amount of decrease in the misfire parameter α at the time of the single-cylinder misfire. That is, the threshold A1 is set in association with the change in the misfire parameter α when the single-cylinder misfire occurs in each of the plurality of cylinders.

FIG. 6 is a diagram illustrating an example of the misfire parameter α (“O” mark) of a normal cycle and the misfire parameter α (“x” mark) of a misfire cycle side by side. As illustrated in FIG. 6, in the misfire cycle, the misfire parameter α at the time of the single-cylinder misfire is smaller than the misfire parameter α at the time of the multi-cylinder misfire. In consideration of the amount of decrease in the misfire parameter α assumed at the time of the single-cylinder misfire and the amount of decrease in the misfire parameter α assumed at the time of the multi-cylinder misfire due to weak combustion, the threshold A1 is set to a value larger than the misfire parameter α assumed at the time of the single-cylinder misfire and smaller than the misfire parameter α assumed at the time of the multi-cylinder misfire. When the comparison circuit 51 of FIG. 5 determines that the misfire parameter α0 of the subject cylinder c0 is less than the threshold A1, there is a possibility that the single-cylinder misfire has occurred in the subject cylinder c0. In this case, the comparison circuit 51 outputs a flag signal f0 indicating the possibility of the single-cylinder misfire.

The comparison circuit 55 determines the magnitude of the misfire parameter α0 of the subject cylinder c0 and a predetermined threshold A2. The threshold A2 is set in consideration of the amount of decrease in the misfire parameter α0 assumed at the time of the multi-cylinder misfire due to weak combustion. That is, the threshold is set in association with the change in the misfire parameter α0 of the subject cylinder c0 when each of all combinations (c0, c1), (c0, c2), and (c0, c3) of the subject cylinder c0 and the other cylinders c1 to c3 among the plurality of cylinders causes misfire (multi-cylinder misfire). As illustrated in FIG. 6, the threshold A2 is set to a value larger than the threshold A1. More specifically, in consideration of the amount of decrease in the misfire parameter α assumed at the time of the multi-cylinder misfire and the amount variation in the misfire parameter α assumed at the time of normal combustion, the threshold A2 is set to a value larger than the misfire parameter α0 of the subject cylinder c0 assumed at the time of the multi-cylinder misfire and smaller than the misfire parameter α assumed at the time of normal combustion.

When the comparison circuit 55 of FIG. 5 determines that the misfire parameter α0 of the subject cylinder c0 is less than the threshold A2, there is a possibility that the single-cylinder misfire has occurred in the subject cylinder c0 (α<A1<A2), or the multi-cylinder misfire has occurred in the subject cylinder c0 and any of the other cylinders c1 to c3 (A1<α<A2). In this case, the comparison circuit 55 outputs a flag signal f4 indicating that there is a possibility of the single-cylinder misfire or the multi-cylinder misfire. As described above, the condition for determining that the misfire occurs in the subject cylinder c0 is that at least the flag signal f4 is output, and when the flag signal f4 is not output, it is determined that misfire does not occur in the subject cylinder c0.

The comparison circuit 52 determines the magnitude of the sum parameter α11 and a predetermined threshold A3. The comparison circuit 53 determines the magnitude of the sum parameter α12 and a predetermined threshold A4. The comparison circuit 54 determines the magnitude of the sum parameter α13 and the predetermined threshold A3. The threshold A3 is set corresponding to the continuous misfire, and the threshold A4 is set corresponding to the facing misfire.

More specifically, the threshold A3 is set in consideration of the amount of decrease in the plurality of misfire parameters α assumed at the time of continuous misfire due to weak combustion. That is, the threshold A3 is set in association with a change in the sum parameters α11 and α13 obtained by summing the misfire parameters α0 and α1 or the misfire parameters α0 and α3 assumed when the multi-cylinder misfire occurs in the combination (c0, c1) of the subject cylinder c0 and the first reference cylinder c1 and the combination (c0, c3) of the subject cylinder c0 and the third reference cylinder c3 among the plurality of cylinders. The threshold A4 is set in consideration of the amount of decrease in the plurality of misfire parameters α assumed at the time of facing misfire due to weak combustion. That is, the threshold A4 is set in association with a change in the sum parameter α12 obtained by summing the misfire parameters α0 and α2 assumed when the multi-cylinder misfire occurs in the combination (c0, c2) of the subject cylinder c0 and the second reference cylinder c2 among the plurality of cylinders.

More specifically, when a value obtained by adding the misfire parameters α (“O” marks in FIG. 4) of the cylinders in the normal cycle corresponding to the sum parameters α11, α12, and α13 is set as the sum parameter in the normal cycle, the thresholds A3 and A4 are set to values larger than the sum parameters α11, α12, and α13 in the misfire cycle and smaller than the sum parameter in the normal cycle. A value obtained by doubling the misfire parameter α (the misfire parameter α in the subject cylinder c0) of the normal cycle assumed at the time of weak combustion may be used as the sum parameter in the normal cycle. The threshold A3 and the threshold A4 may be set to the same value. The threshold A3 may be set to different values for the preceding continuous misfire and the subsequent continuous misfire.

A difference between the sum parameters α11, α12, α13 in the misfire cycle and the sum parameter in the normal cycle is larger than the difference Δα (FIG. 4) between the misfire parameter α in the misfire cycle and the misfire parameter α in the normal cycle. Therefore, it is easy to set the thresholds A3 and A4 so as to satisfy the condition that the threshold is larger than the sum parameters α11, α12, α13 in the misfire cycle and smaller than the sum parameter in the normal cycle. When determining that the sum parameter α10 is less than the threshold A3, the comparison circuit 52 outputs a flag signal f1 indicating a possibility that the preceding continuous misfire has occurred. When determining that the sum parameter α12 is less than the threshold A4, the comparison circuit 53 outputs a flag signal f2 indicating a possibility that the facing misfire has occurred. When determining that the sum parameter α13 is less than the threshold A3, the comparison circuit 54 outputs a flag signal f3 indicating a possibility that the subsequent continuous misfire has occurred.

When any one of the flag signals f0 to f3 is input, the OR circuit 56 outputs the relevant flag signal f0 to f3. When none of the flag signals f0 to f3 is input, the OR circuit 56 outputs an off signal.

The AND circuit 57 has a function as a determination circuit that determines the presence or absence of misfire of the engine 1. When any of the flag signals f0 to f3 is output from the OR circuit 56 and the flag signal f4 is output from the comparison circuit 55, the AND circuit 57 determines that misfire has occurred in the engine 1, that is, the engine 1 is operating in the misfire cycle. On the other hand, when the off signal is output from the OR circuit 56 or the flag signal f4 is not output from the comparison circuit 55, the AND circuit 57 determines that no misfire has occurred in the engine 1, that is, the engine 1 is operating in the normal cycle.

When determining that the engine 1 is operating in the misfire cycle, the AND circuit 57 further specifies the misfire pattern (a manner of misfire) according to the flag signals f0 to f3. That is, when the flag signal f0 is output and none of the flag signals f1 to f3 is output, the AND circuit 57 determines that the single-cylinder misfire has occurred in the subject cylinder c0. When the flag signal f1 is output and the flag signal f0 is not output, the AND circuit 57 determines that the multi-cylinder misfire (preceding continuous misfire) has occurred in the subject cylinder c0 and the first reference cylinder c1. When the flag signal f2 is output and the flag signal f0 is not output, the AND circuit 57 determines that the multi-cylinder misfire (facing misfire) has occurred in the subject cylinder c0 and the second reference cylinder c2. When the flag signal f3 is output and the flag signal f0 is not output, the AND circuit 57 determines that the multi-cylinder misfire (subsequent continuous misfire) has occurred in the subject cylinder c0 and the third reference cylinder c3. When the flag signal f0 is output and any of the flag signals f1 to f3 is output, the AND circuit 57 determines that the single-cylinder misfire has occurred in the subject cylinder c0 or the multi-cylinder misfire has occurred in a plurality of cylinders including the subject cylinder c0.

When the AND circuit 57 determines that misfire has occurred in the engine 1, the controller 20 (FIG. 2) executes processing of suppressing the misfire. For example, when an ignition period is retarded, the controller 20 outputs a control signal to an ignition plug 11 to perform correction to advance the ignition period. Alternatively, the controller 20 outputs a control signal to an injector 12 to change an injection pattern and an injection timing.

FIG. 7 is a diagram illustrating the relationship between the flag signals f0 to f4 and the misfire pattern. In FIG. 7, an “O” mark indicates that the flag signal f0 to f4 is output, and an “x” mark indicates that the flag signal f0 to f4 is not output. For the flag signals f1 to f3, a case where any of the flag signals f1 to f3 is output is indicated by the “O” mark, and a case where none of the flag signals f1 to f3 is output is indicated by the “x” mark. When the flag signal f0 to f4 is output, at least the subject cylinder c0 is misfired, and cases where the normal combustion has occurred in the subject cylinder c0 and the misfire has occurred in other cylinders are not assumed here.

As illustrated in FIG. 7, when the flag signal f0 and the flag signal f4 are output and the flag signals f1 to f3 are not output, the misfire parameter α0 is less than the threshold A1 (FIG. 6), and it is determined that the single-cylinder misfire has occurred in the subject cylinder c0. When none of the flag signals f0 to f4 is output, it is determined that the engine 1 is combusted in the normal cycle.

Even when the flag signal f4 is output and the flag signals f0 to f3 are not output, it is determined that the engine 1 is combusted in the normal cycle. This case is considered such that although no misfire has occurred in the subject cylinder c0, the flag signal f4 is output since the misfire parameter α0 of the subject cylinder c0 becomes smaller than the threshold A2 due to weak combustion. That is, since the threshold A2 is larger than the threshold A1, the misfire parameter α0 may be smaller than the threshold A2 although no misfire occurs. However, in the present embodiment, not only that the flag signal f4 is output, but also that any of the flag signals f1 to f3 is output is set as the condition of the multi-cylinder misfire. Therefore, even in a case where α0<A2 is satisfied, the occurrence of the multi-cylinder misfire can be reliably determined to be distinguished from the normal combustion.

When the flag signal f0 and the flag signal f4 are output and any of the flag signals f1 to f3 is further output, it is determined that the single-cylinder misfire has occurred in the subject cylinder c0 or the multi-cylinder misfire has occurred in a plurality of cylinders including the subject cylinder c0. When any of the flag signals f1 to f3 is output and none of the flag signals f0 and f4 is output, it is considered that although no misfire occurs in the subject cylinder c0, any of the flag signals f1 to f3 is output since any of the sum parameters α11 to α13 becomes smaller than the thresholds A3 and A4 due to weak combustion. Therefore, it is determined that the engine 1 is combusted in the normal cycle.

When any of the flag signals f1 to f3 and the flag signal f4 are output and the flag signal f0 is not output, it is determined that multi-cylinder misfire has occurred in a plurality of cylinders including the subject cylinder c0. If the output of the flag signal f4 without the output of the flag signal f0 is set as a condition for the determination on the multi-cylinder misfire, it may be erroneously determined that the misfire has occurred although the engine 1 is combusted normally. On the other hand, by adding the output of any of the flag signals f1 to f3 to the condition for the determination on the multi-cylinder misfire, the presence or absence of occurrence of the multi-cylinder misfire can be determined reliably.

The present embodiment can achieve advantages and effects such as the following:

(1) A misfire determination apparatus 10 for an internal combustion engine includes: a crank angle sensor 21 that detects a rotational speed of a crankshaft 107 of an engine 1 having a plurality of cylinders; a parameter calculation unit 40 that calculates a misfire parameter α having a correlation with a change amount of a rotational speed in a combustion stroke of each of the plurality of cylinders, that is, a misfire parameter α which increases as an increase amount of the rotational speed increases, on the basis of the rotational speed detected by the crank angle sensor 21; and a determination unit 50 that has a comparison circuit 51 that determines whether or not a misfire parameter α0 calculated by the parameter calculation unit 40 (base parameter calculation unit 41) is less than a threshold A1 (a first predetermined value), and a comparison circuit 55 that determines whether or not the misfire parameter α0 is less than a threshold A2 (a second predetermined value) larger than the threshold A1, and determines the presence or absence of misfire of the engine 1 according to determination results of the comparison circuits 51 and 55 (FIG. 5). When the comparison circuit 51 determines that the misfire parameter α0 is less than the threshold A1, the determination unit 50 determines that single-cylinder misfire, which is misfire in a single cylinder of the plurality of cylinders, has occurred, and when the comparison circuit 55 determines that the misfire parameter α0 is less than the threshold A2, the determination unit 50 determines that the single-cylinder misfire or multi-cylinder misfire, which is misfire in two or more cylinders of the plurality of cylinders, has occurred (FIG. 5).

With this configuration, it is possible to satisfactorily determine the presence or absence of the multi-cylinder misfire of the engine 1. That is, since the amount of decrease in the misfire parameter α0 is different between the single-cylinder misfire and the multi-cylinder misfire, it is difficult to accurately determine the presence or absence of misfire only by comparing the misfire parameter α0 with the single threshold A1 or A2. However, it is possible to accurately determine the presence or absence of the single-cylinder misfire and the multi-cylinder misfire by comparing the misfire parameter α0 with two thresholds A1 and A2 different from each other.

(2) The determination unit 50 identifies a subject cylinder c0, which is a cylinder during the combustion stroke or immediately after completion of the combustion stroke, among the plurality of cylinders, determines that the single-cylinder misfire has occurred in the subject cylinder c0 when the misfire parameter α0 is determined to be less than the threshold A1, and determines that the multi-cylinder misfire has occurred in two cylinders including the subject cylinder c0 when the misfire parameter α0 is determined to be equal to or greater than the threshold A1 and less than the threshold A2 (FIG. 6). As a result, the cylinder in which the misfire has occurred can be identified, and the single-cylinder misfire and the multi-cylinder misfire can be determined to be distinguished.

(3) The parameter calculation unit 40 calculates, on the basis of the rotational speed detected by the crank angle sensor 21, the misfire parameter α0 having a correlation with the amount of change in the rotational speed in the combustion stroke of the subject cylinder c0 and the misfire parameters α1 to α3 having a correlation with the amount of change in the rotational speed in the combustion stroke of the reference cylinders c1 to c3 within a range of a crank angle which is one cycle of the engine 1 back from the combustion stroke of the subject cylinder c0 as a starting point, and calculates sum parameters (a sum misfire parameter) α11 to α13, each of which is a sum of the misfire parameter α0 and any one of the misfire parameters α1 to α3 (FIG. 5). The determination unit 50 further has comparison circuits 52 to 54 that determine whether or not the sum parameters α11 to α13 calculated by the parameter calculation unit 40 are less than the thresholds A3 and A4 (third threshold) (FIG. 5). Then, when the comparison circuit 51 determines that the misfire parameter α0 is less than the threshold A1, or the comparison circuits 52 to 54 determine that the sum parameters α11 to α13 are less than the thresholds A3 and A4, and the comparison circuit 55 determines that the misfire parameter α0 is less than the threshold A2, the determination unit 50 determines that misfire has occurred in the subject cylinder c0. When the comparison circuit 51 determines that the misfire parameter α0 is equal to or greater than the threshold A1, and the comparison circuits 52 to 54 determine that the sum parameters α11 to α13 are equal to or greater than the thresholds A3 and A4, even if the comparison circuit 55 determines that the misfire parameter α0 is less than the threshold A2, the determination unit 50 determines that no misfire has occurred in the subject cylinder c0 (FIGS. 5 and 7). The characteristics of the misfire parameter α at the time of multi-cylinder misfire can be satisfactorily reflected by the sum parameters α11 to α13. Therefore, by using the sum parameters α11 to α13, the presence or absence of multi-cylinder misfire can be accurately determined.

(4) When the comparison circuit 51 determines that the misfire parameter α0 is less than the threshold A1 and the comparison circuits 52 to 54 determine that the sum parameters α11 to α13 are equal to or greater than the thresholds A3 and A4, the determination unit 50 determines that single-cylinder misfire has occurred in the subject cylinder c0 (FIGS. 5 and 7). That is, when the misfire parameter α0 is less than the threshold A1, a condition that the misfire parameter α0 is less than the threshold A2 is satisfied, so that the determination unit 50 can determine that the single-cylinder misfire has occurred in the subject cylinder c0 without comparing the magnitude of the misfire parameter α0 and the threshold A2. As a result, the presence or absence of occurrence of the single-cylinder misfire can be accurately determined.

(5) When the comparison circuit 51 determines that the misfire parameter α0 is equal to or greater than the threshold A1, the comparison circuit 55 determines that the misfire parameter α0 is less than the threshold A2, and the comparison circuits 52 to 54 determine that the sum parameters α11 to α13 are less than the thresholds A3 and A4, the determination unit 50 determines that multi-cylinder misfire has occurred in two cylinders including the subject cylinder c0 (FIGS. 5 and 7). As a result, the presence or absence of occurrence of the multi-cylinder misfire can be accurately determined.

(6) The threshold A1 is set in advance in association with a change in the misfire parameter α0 when the single-cylinder misfire occurs in each of the plurality of cylinders (FIG. 6). By setting the threshold A1 in this manner, it is possible to clearly distinguish between the single-cylinder misfire and the multi-cylinder misfire, and to accurately detect the occurrence of the single-cylinder misfire.

(7) The threshold A2 is set in advance in association with a change in the misfire parameter α0 when the multi-cylinder misfire occurs in two cylinders including all combinations of the plurality of cylinders (FIG. 6). By setting the threshold A2 in this manner, it is possible to clearly distinguish between the single-cylinder misfire and the multi-cylinder misfire, and to accurately detect the occurrence of the multi-cylinder misfire.

(8) The thresholds A3 and A4 are set in advance in association with changes in the sum parameters α11 to α13 when the multi-cylinder misfire occurs in two cylinders including all combinations of the plurality of cylinders. By setting the thresholds A3 and A4 in this manner, it is possible to clearly distinguish between the multi-cylinder misfire and the normal combustion, and to accurately detect the occurrence of the multi-cylinder misfire.

In the above embodiment, the engine 1 is configured to have four cylinders. However, as long as the engine is configured to have a plurality of cylinders, the number of cylinders is not limited to four and may be five or more. In the case where the number of cylinders is, for example, six, every time the crank angle θ changes by 120°, a combustion stroke is sequentially started with different cylinders. In this case, the parameter calculation unit 40 may calculate the misfire parameter α for the subject cylinder c0 based on crank angle data detected by the crank angle sensor 21 from the start of the combustion stroke in the subject cylinder c0 until the start of the combustion stroke in the subsequent cylinder (another cylinder) (until the crank angle θ changes by 120°). Thus, even if the number of cylinders is 5 or more, it is possible to accurately determine the presence or absence of misfire in the engine 1.

In the above embodiment, the misfire determination apparatus 10 is applied to the engine vehicle traveling by the power of the engine 1, but it is also possible to apply the misfire determination apparatus to a hybrid vehicle including the engine 1 and a traveling motor. In this case, as illustrated in FIG. 8, the misfire determination apparatus 10 further includes a rotational speed detection part 23 such as a resolver that detects the rotational speed of the traveling motor, and the parameter calculation unit 40 may calculate the misfire parameter α based on crank angle data detected by the crank angle sensor 21 (a first rotational speed detector) and the rotational speed of the traveling motor detected by the rotational speed detection part 23 (a second rotational speed detector). Accordingly, it is possible to accurately determine the presence or absence of misfire of the engine 1 by excluding an influence on the rotational speed of the engine 1 caused by the rotation of the traveling motor. An electric equipment rotationally driven by the power of the engine 1 may be another electric equipment such as a generator instead of a traveling motor, and the rotational speed detection part 23 may detect the rotational speed of the generator or the like. A sensor for detecting a load acting on the electric equipment may be provided, and the parameter calculation unit