EP1068436A1

EP1068436A1 - Method for equalising the cylinders of a direct injection internal combustion engine

Info

Publication number: EP1068436A1
Application number: EP99913112A
Authority: EP
Inventors: Achim Przymusinski; Ralf Schernewski; Andreas KÄRCHER
Original assignee: Siemens AG
Current assignee: Siemens AG
Priority date: 1998-03-30
Filing date: 1999-03-10
Publication date: 2001-01-17
Anticipated expiration: 2019-03-10
Also published as: WO1999050547A1; DE19814155A1; DE59900863D1; US6302083B1; EP1068436B1

Abstract

The r.p.m. values of the crankshaft are corrected by means of an acausal mean value filter. Modification of the kinetic energy of the crankshaft in the expansion interval of a cylinder is calculated on the basis of the dynamically corrected r.p.m. values and applied to the maximum amount of fuel which can be supplied during this interval. The non-uniform residue thus obtained indicates whether too much or too little fuel has been injected for the cylinder in question. The calculated residues are used to derive the correction terms for the injection times of the individual cylinders. This enables an entire range of characteristics to be adapted, especially with respect to changing speeds.

Description

description

Method for cylinder equalization in an internal combustion engine working with direct injection

The invention relates to a method for cylinder equalization in an internal combustion engine working with direct injection according to the preamble of claim 1.

In the case of multi-cylinder, direct-injection internal combustion engines, due to various interfering influences, despite the same control, there are high variances in the mass flow between individual injection nozzles. The different amounts of fuel lead to different torque contributions from the individual cylinders, which in addition to increasing the uneven running due to fluctuations in the speed of the crankshaft can also lead to increased emissions.

From DE 41 22 139 AI a method for equalizing the cylinder with respect to the fuel injection quantities in an internal combustion engine is known, in which the rotational acceleration of each individual cylinder is detected. The individual measured values of the rotational acceleration are compared with one another and, in the event of deviations between the individual measured values, the fuel injection quantities of the individual cylinders are changed in such a way that deviations are ultimately avoided and thus rotational irregularities in the internal combustion engine are eliminated.

The invention is based on the object of specifying a method of the type mentioned at the outset, with which the systematic error of the individual injection nozzles of the injection system can be compensated for in a simple and rapid manner both when the internal combustion engine is operating in a stationary and in a non-stationary manner. 2

The stated object is achieved by the features of patent claim 1. Advantageous further developments are specified in the subclaims.

The energy released by the combustion in the individual cylinders is converted into kinetic energy of the crankshaft. Cylinder-specific combustion differences are expressed in speed fluctuations from which an error can be determined. This cylinder-specific error signal is characteristic of the systematic error during the injection process into the cylinder. In order to rule out incorrect adaptations even during unsteady operation of the internal combustion engine, for example when accelerating, the characteristic values, i.e. the speed values detected by the speed sensor, are corrected dynamically with an acausal filter. This dynamic correction makes it possible to determine an error even during speed transitions and to carry out an adaptation in the entire map area. Since the method uses the crankshaft speed supplied to the control device of the internal combustion engine anyway for other control and control purposes and therefore no additional hardware components are required, there is a very inexpensive implementation for increasing the smooth running of the internal combustion engine.

An exemplary embodiment of the invention is explained in more detail below with reference to the drawing. Show it

FIG. 1 shows the speed curve for faultless and faulty injectors in stationary operation,

FIG. 2 shows the speed curve for error-free injectors in stationary and transient operation,

FIG. 3 the speed curve and the averaging time periods for the dynamic correction, 3

FIG. 4 shows a diagram to clarify the running time of the mean filter,

5 shows a representation of the dynamic correction on the basis of an acceleration process and

Figure 6 speed and residual curve with slight acceleration of the internal combustion engine

In order to be able to correct a possible deviation of the actually injected fuel quantity from the target injection quantity, it is necessary to determine a measure for this deviation, that is to say an error. The signal from a speed sensor is used for this. The energy released by the combustion of the individual cylinders becomes m kinetic energy

Converted crankshaft. Cylinder-specific combustion differences are expressed in speed fluctuations from which an error can be determined. The kinetic energy that is released during combustion in a cylinder 1 is calculated

^ΔEk , "(= 2 ^-ö" ( ^ö, u (,) - ^ö, OT ( ₁ )) ⁽¹⁾

with θ mean moment of inertia of the crankshaft ^ω _{0T (} angular velocity at top dead center (before the expansion phase) J JJJI angular velocity at bottom dead center (after the

Expansion phase)

The bottom dead center (index UT) of the cylinder l corresponds to the top dead center (index OT) of the next fired cylinder z + 1. Therefore this equation can also be given as follows: ΔEkin «* - ( ^ω θT (ι + l) ⁰ OT (ι), (2;

A positive change in the kinetic energy (AE _km (i)> 0) corresponds to an excessive injection quantity m _{B l} of fuel and a negative change in the kinetic energy

(AE _kιn (i) <0) corresponds to an injection _quantity m _{B l} that is too small.

If AE _kιn (l) = 0 _f , the correct _{amount of} fuel has been injected.

However, these statements only apply if one can assume a quasi-steady-state operating state (ie the mean speed remains constant) and the load torque shows no jumps.

The course of the speed n over time t for fault-free injectors is plotted in FIG. 1A, and the speed curve over time for faulty injectors is plotted for the steady-state operation of the internal combustion engine. In both cases, in the form of circles, the cylinder-individual signal values, namely the speeds before the ignition n (OT (ι)) and after the ignition n (UT (ι)) are identified for the cylinder I.

The angular velocity CO and thus also Δs _faΛ can be calculated from the speed n by simple conversion.

If the method is also used when the internal combustion engine is not in stationary operation, no statements can be made about the errors of the injectors. E.g. the internal combustion engine is accelerating, an error is recognized where, under certain circumstances, none has to be.

This problem is shown in FIG. 2. The temporal speed curve in FIG. 2A was simulated with faultless injectors in stationary operation. The above procedure provides a value for the change in kinetic energy AE _km i = 0, i.e. no error. In FIG. 2B, the internal combustion engine was accelerated using the same fault-free injectors. The method now calculates a value for the change in the kinetic energy Δ £ ^ (z)> 0, since the rotational speed after the combustion is greater than before and concludes that the current injector has a positive error, that is to say injects too much.

In order to eliminate the limitation of a quasi-stationary operating state and to be able to determine an error even in dynamic speed transitions, a dynamic correction of the speed is carried out. This dynamic correction is explained below using the example of a 4-cylinder internal combustion engine.

The basic idea behind dynamic correction is to take the tendency of the average speed into account. For this purpose, the actual _speeds n _oτ ^ _{l +]} ^ and n _oτ ^ are not used to determine the error, but corrected speeds _{üHι +])} and n _oHl) .

These are freed from the trend of the average speed and thus allow a statement about the injection behavior of the injector under consideration.

To determine this trend, average speeds are calculated and related to the current values.

However, in order to enable a comparison of the current speed with an average, the average filter used must have a group delay of r = 0. This can only be achieved with an acausal filter where the current time is in the middle of the averaging interval.

The averaging time should be as short as possible to avoid any changes in the speed trend quickly recognizable. On the other hand, however, at least one work cycle must be worked out in order to calculate the systematic errors of the injectors.

Since the internal combustion engine examined here is a 4-cylinder internal combustion engine, four speed values (each at top dead center) must be taken into account when communicating a work cycle. However, in order to be able to maintain the required runtime of the mean value filter of τ = 0, the current point in time must be in the middle of the averaging interval, as described above. Since there is no mean value for four speed values, the mean is averaged over five values. The dynamic correction of the speed results from the acausal averaging:

_ 1 1 1 1 1 ⁿ 0T (ι) ^~ g ⁿ θT ι-2) ^{+ ~} ^ ^H OT {ι- \) ⁺ ₄ ⁿ 0T (ι) ^{+ ~} ^ ⁿ θT (ι + \) ^{+ ~} ^ ⁿ θT (ι + 2) < ³ )

ⁿ _0T _(ι - ₂₎ ^{n unc} _{θT (ι + 2)} 9 ^e hear while at the same cylinder and are each only half as heavily weighted as the other three values. This guarantees notification of exactly one work cycle.

You can get a handle on the acausality by performing the calculation for the cylinder that is current at time l only at the end of the averaging period at time i + 2. Since the corresponding value is only needed again in the next work cycle, i.e. at time 1 + 4, this is possible without any problems.

An average corresponding to equation (3) is also calculated for OT ₍₁₊ i):

_ 1 1 1 _ J_ ^π θ7 (, + ι) - g ⁿ oι (, - \) +. ⁿ oτ (\) ⁺ 4 ⁿ oι (> + \) ⁺ ₄ "θ7 (ι + 2) ⁺ g ⁿ oτ (, + ι) ⁽⁴ ι 7

FIG. 3 shows the speed _curve over two work cycles ASP and the averaging time periods for “ _{OT (/ + 1)} and n _{oτ {l +])} for the dynamic correction.

To clarify the influence of the transit time T of the mean filter, reference is made to FIG. 4. A sinusoidal signal is recorded in FIG. 4A. Below (FIG. 4B) is the course of the mean value when averaging over a quarter of the period when using a classic method with the aid of a causal mean filter whose runtime is r> 0. Only values from the past are used to calculate the mean value for the current point in time (indicated here by a vertical straight line). A phase shift between the smussignal and the middle curve (mean value 1) can be clearly seen.

FIG. 4C shows the mean value curve when using an acausal filter (T = 0). The same number of values from the past and the future are used for the calculation (the current point in time is in the middle of the averaging interval). It can be clearly seen here that the sinusoidal signal and the mean value signal 2 are in phase.

The corrected speeds are now calculated using the mean values from equations (3) and (4):

_ ⁿ θT {ι + \) ^{~ n} θT (ι) ⁿ θT {ι + \) ^{~ H} OT {ι + \) ö ⁽⁵ >

01 (ι + ϊ - nn ^■ 07 (ι) or (ή = n ^{, l} o _r τ (ή

The values n ₀ τ < _ι ) ^and d ⁿ oτu ₊ u denote the values measured with the aid of the speed sensor.

An example of dynamic speed correction is shown in FIG. Flawless injectors are assumed. This can be recognized by the fact that the current speed values n ₀ τ ₍ Λ and fT-oτ ( _{ι + \)} ^unc the associated mean _values n _oτ iΛ and ^~ ₀ τ _{(ι + \) have} the same distances (here Δλ? = 10) The internal combustion engine is accelerated

_oτ iΛ = 1220 [1 / min] ^~ oτ { _{ι + \} ) ^{= 1220} [1 min]

The corrected speeds are therefore the same, from which it can be concluded that the injectors are working correctly. The rising trend of the speed could therefore be filtered out.

With the corrected speed values from equations (5) and (6), the change in the kinetic energy IΔE _kιnZ , J is now calculated for the cylinder Z (i) according to the following equation:

ΔE km ^{<) F 2F n ² ή ⁿ θ ² T (ι and from this a residual is calculated

R∑ (ι) - Knorm ^'n θT (k, ι + \) OT {k, ι) with

1 . (2π ²

K, _m „„ = ² _H ^ 300 and R _z ) residual of the cylinder Z (i)

Θ average moment of inertia of the crankshaft

(Applied) H _u lower calorific value for the fuel used πϊBmax maximum amount of fuel that can be injected n _{ω (l)} corrected speed at top dead center of the cylinder i " _{f7 (ι + D} corrected speed at top dead center of the cylinder l + l K _norm as a normalization factor

that, with appropriate standardization, contains a statement as to what percentage of fuel was injected too much or too little.

2π

In equation (8), the factor is used to convert conversion

60 rotations per minute (unit of n) m radians per second

(Unit of CO). By multiplying by - ^{1 •} ü n you get

2 an energy difference corresponding to that in equation (2). The division by H _tl - m _BmBX and the multiplication by 100 result in a percentage error, since the difference in chemical energy that occurs due to injector errors during ignition is related to the total energy of the injected fuel mass m _n .

FIG. 6 shows a speed curve with a slight acceleration of the internal combustion engine, in which a larger injection quantity of fuel was predefined on the cylinder Z (l). From the lower representation in FIG. 6 it can be seen that, despite an increase in the speed, the cylinder-individual residuals, which are synonymous with the errors, remain the same due to the dynamic correction indicated. Every fourth value belongs to the same cylinder l. One can clearly see that the error patterns remain the same.

Proportional injection corrections can now be made from the cylinder-individual residuals obtained with this method. Since the residuals are only relative measures for the change in the amount of fuel to be injected, the adaptation algorithm is also created from this aspect. It must be ensured that the internal combustion engine does not generate more or less fuel at any correction time. 10 holds as in the uncorrected case. The algorithm should therefore only take on the task of uniformly distributing the injection quantity. This results in the adaptation algorithm for a 4-cylinder internal combustion engine

^ z (ιμ Z (l), k-1 ' ^R Z (l), k Z (2), k ⁺ Z (3), k ^{+ R} Z (4), k Z (2), k Z (2 ), k-1 Z (2) k 1 ^R Z (3), k ^{+ R} Z (4), k ^{+ R} Z (l), k

+ a + - •

^ Z (3), k Z (3), k-1 Z (3), k 3 ^R Z (4), k ^{+ R} Z (l), k ^{+ R} Z (2), k ^ Z (4), k δ Z 4, kl V ^R Z (4), k7 VZ (l), k ⁺ Z (2), k ^{+ R} Z (3), k

^υ z (\)

3∑ (2), 0 with 3ψ) .0 ^ Z (4), 0 as initialization for a multiplicative adaptation case

Therein δ Z (i), k correction term for cylinder i after adaptation step k ^R Z (i), k residual of cylinder i at adaptation step ka is a positive, freely selectable adaptation parameter between 0 and 1, which determines the speed of the adaptation

If more fuel is injected into a cylinder (i.e. the residual of the cylinder was positive), this amount of fuel becomes proportional (i.e. one third each for a 4-cylinder internal combustion engine, generally l / (zl) for an internal combustion engine with z-cylinders ) deducted from the other cylinders.

If you inject less fuel into a cylinder (ie the residual of the cylinder was negative), this amount of fuel becomes proportional (ie one third each for a 4-cylinder internal combustion engine, generally l / (zl) for one 11

Internal combustion engine with z-cylinders) struck on the other cylinders.

This ensures that the torque remains constant during cylinder equalization, since the total amount of fuel to be supplied does not change.

The sum of the correction terms is equal to the number of cylinders at all times.

To correct the injection quantity m _B , ι into a cylinder, only a single control variable is available, namely the injection duration T _E. Because of the always positive slope of an inverted nozzle characteristic (injection quantity as a function of the injection duration), a longer activation pulls a larger one

Injection quantity with itself. The injection correction can thus take place directly over the injection duration by multiplying the correction terms δ _z tΛ _κ from the adaptation algorithm by the ideal injection times T _E , ιdeai specified by the engine control.

Claims

12 claims

1. Method for cylinder equalization with regard to the fuel injection quantities for an internal combustion engine working with direct injection, in which the fuel injection quantities can be controlled by changing the injection times and the injection times are applied with cylinder-specific correction terms in such a way that the smooth running of the internal combustion engine is increased

-the speed (n) of the crankshaft of the internal combustion engine is recorded both in the quasi-stationary and in the dynamic operating state of the internal combustion engine, ) the change in the kinetic energy iAE _kmι ) of the crankshaft in the expansion _{interval of} a cylinder (i) is calculated, from which a relative measure (residual Rz ₍ ± ₎ ) for each cylinder

(Z (i)) is derived, which contains the information about too much or too little injected fuel and

- From this measure (residual Rz ₍ i ₎ ) correction terms (δ _Z (Λ _k ) for the injection time (TE) can be calculated.

2. The method according to claim 1, characterized in that the correction of the speed values (n ₍₁₎ ) takes place according to the following relationship:

^l OT (ι n

"OT, (/ + 1) n, O.T (ι)

OT (ι + \) 13

_ ⁿ QT (ι + \) ~ ⁿ 0T (ή ⁿ oτ (ι) ^{~ n} oτ (ή ~ with h ₀ i _(<) '"or _{(/ +} ι ₎ ^as corrected speed of the cylinder i or i + 1 over a work cycle ή _{01 (l)} , n _{ül {l + λ)} mean value of the speed of the cylinder i, or i + 1 over a work cycle

3. The method according to claim 2, characterized in that in a 4-cylinder internal combustion engine the mean (n _oτι , _\ ) of the cylinder (i) is calculated to:

^l oτ (ι) - ^~ G ⁿ oτ (ι-2) ⁺ G ⁿ oτ (ι- \) ⁺ G ⁿ oτ (ι) ⁺ G ⁿ oτ {ι + \) ⁺ 1 G. ⁿ or (ι + 2 )

4. The method according to claim 2, characterized in that in a 4-cylinder internal combustion engine the mean value (« _{07 (l + 1)} ) of the cylinder (i + 1) is calculated to:

"« / (, +!) ^_ G "« / Η) ⁺ ₄ "OT (1) 'θ7' (ι + l) ⁿ () J (ι + 2) ⁺ _R ⁿ θl {, + 3)

5. The method according to claim 1, characterized in that the values for the change in the kinetic energy [ΔE _{km ι} \ is based on a value which indicates the maximum supplyable fuel energy m this interval and the measure

(Residual Rz ₍ ι ₎ ) is calculated.

6. The method according to claim 1, characterized in that the change in the kinetic energy lΔE _kιnZ , J is calculated according to the following equation

Δ £ faHv ^{= ~~} '^' ^ oτ ( _{ι + \} ) ^~ ** oτ ( _ι )) ^{un <} ^ from which the measure is determined 14

^ Z (ι) ^~ wπn ^' [ ⁿ oT (κ, ι + l) ⁿ 0T (k, ι)

With

Θ average moment of inertia of the crankshaft

H _u lower calorific value for the fuel used ^m Bmax maximum injectable fuel quantity ⁿ θl (ι) corrected speed at top dead center of the cylinder in TDC (ι + \) corrected speed at top dead center of the cylinder i + 1

K "a scaling factor that determines the value

Θ 1, 2

2π

2 H „m _BaΛ \ m

7. The method according to claim 1, characterized in that correction terms ( _{| t} ) are calculated from the calculated dimensions (residual Rz ₍ ι ₎ ) by which the values for the injection times ( ^T _E , i _d ea _l ) are multiplied.

8. The method according to claim 7, characterized in that the

Correction terms z (,), k ι are calculated

z (ι), k Z (l), k-1 Z (l), k 'Z (2), k ⁺ Z (3), k ^{+ R} Z (4), k

^ Z (2), k Z (2), k-1 RZ (2) k ^R Z (3), k ^{+ R} Z (4), k ⁺ Z (l), k

+ a + -

^ Z (3), k δ Z (3), k-1 RZ (3), k ^R Z (4), k ^{+ R} Z (l), k ^{+ R} Z (2), k Z (4), k _ * Z (4), kl ^R z (4), _k y VZ (l), k ^{+ R} Z (2), k ^{+ R} Z (3), k V 15

z (ι), o

Z (2), 0 with Z (3), 0 Z (4), 0 as initialization value and with

Zi.k correction term for cylinder i after adaptation step k

R Z (i), k residual of the cylinder i to the adaptation step k a positive freely selectable adaptation parameters between 0 and 1, which determines the speed of the adaptation