GB2257542A

GB2257542A - Method of equalising fuel injection between engine cylinders

Info

Publication number: GB2257542A
Application number: GB9214241A
Authority: GB
Inventors: Wilhelm Eyberg
Original assignee: Robert Bosch GmbH
Current assignee: Robert Bosch GmbH
Priority date: 1991-07-04
Filing date: 1992-07-03
Publication date: 1993-01-13
Also published as: JPH05187302A; GB9214241D0; DE4122139A1; JP3348107B2; DE4122139C2

Description

2 23 7342 - 1 METHOD OF EQUALISING FUEL_INJECTION BETWEEN ENGINE CYLINDERS

The present invention relates to a method of equalising fuel injection between the cylinders of a fuel-injected multicylinder internal combustion engine.

When an internal combustion engine is running, non-uniformities of the engine rotational speed occur due to differences in the quantities of fuel injected into the individual cylinders of the engine. In this case, inter alia, tolerances of the individual injection components play a part and such tolerances are reducibleonly through considerable cost and effort. The non-uniformities of rotational speed can cause, for example, vibrations in a motor vehicle equipped with the engine concerned.

There are known running smoothness regulation procedures, which serve to damp the oscillations due to differences in the quantities of injected fuel. For example, it is known to detect deviations in rotational speed of individual cylinders from the mean rotational speed of the engine. It has proved, however, that the function of such a running smoothness regulation can be optimised only for a limited rotational speed range, so that balancing-out of the oscillations is possible only to a limited extent.

According to the present invention there is provided a method of equalising fuel injection between the cylinders of a fuel-injected multicylinder internal combustion engine, comprising the steps of measuring the rotational speed acceleration for each combustion process, comparing the obtained individual measurement values with each other to recognise deviations therebetween, and in the case of recognised deviations correspondingly varying the quantities of injected fuel until the deviations are eliminated.

2 - A method exemplifying the invention may have the advantage, due to the structure of a PT1-loop, that it enables the avoidance of nonuniformities in engine rotational speed, due to differences in the quantities of injected fuel, over all or the substantial part of the entire operating range of the engine. The detection of the rotational speed acceleration of each combustion process forms the basis of the method. The obtained measurement values are compared with each other and deviations are ascertained. On the basis of such deviations, the fuel injection quantities for the individual cylinders are so varied that deviations are avoided and non-uniformities in engine rotatJonal speed ' which are caused by this phenomenon, are eliminated.

Preferably, a mean value of the measurement values is ascertained, for preference as a sliding mean over all cylinders. In this manner, balancing of the injection quantities can be achieved even in the case of transient engine operating states.

Preferably, also, in the case of a deviation of a measured rotational speed acceleration value from the mean value, an additional (positive or negative) injected quantity is fed to the associated cylinder during one of the next injection phases, for preference the next following injection phase.

c In a further preferred example of the method, the mean value is formed from the sum of the individual additional injected quantities and deducted from all additional injected quantities. Even in the case of sudden changes in the mean rotational acceleration, this form of compensation avoids the mean value of the balancing quantities from becoming different from zero and giving rise to a deviation of the mean injected quantity from a preset target value of the injection quantities. A Mrifting" of the balancing quantities is thus avoided.

An example of the method of the present invention will now be more particularly described with reference to the accompanying drawings, in which:

Fig. 1 is a schematic functional diagram of an internal combustion engine and associated control system; Fig. 2 is a diagram showing qualitative course of engine rotational speed and rotational acceleration in the case of a four cylinder engine; and Fig. 3 is a flow diagram illustrating the steps of a method exemplifying the invention.

Referring now to the drawings, Fi.g. 1 is a highly schematic functional illustration of a fuel-injected internal combustion engine 1 and associated control, the engine 1 having, by way of example, four cyl inders 3. The fuel injection into the cylinders is regulated by way of suitable control lines 5, which are connected with a control device 7. This evaluates signals passed on from a sensor 9, by way of a feed line 11. The sensor 9 scans a segment wheel 13, which rotates synchronously with the crankshaft of the engine 1.

1 In operation of the engine 1, four segments result during the scanning of the rotating segment wheel 13 in the case of a fourcylinder engine, for which it can be presumed that a segment S1 is bounded by instants T1 and T2, a segment S2 by instants T2 and T3, and 5 so forth In the following, the origin of the non-uniformities of engine rotational speed are outlined in general terms. Different cylinder pressure values arise during combustion due to deviations in the quantities of fuel injected into the cylinders 3 of the engine 1.

JO Consequently, accelerating torques based on the combustion also deviate from each other. The relationship between engine torque M and rotational speed n is given by the following equation:

n A dt M B ML dt eges S (2.1) - - 5 - In this equation, the accelerating moment is denoted by M,, the load moment is denoted by M L and the mass inertia moment referred to the crankshaft is denoted by 9 ges.

Neglecting efficiency influences as well as the influence of the crankshaft angle, the accelerating torque M B is proportional to the mass of injected fuel so that the following equation results:

M B c QE In this equation, the mean quantity of fuel conveyed per operating cycle is denoted by E and a constant is denoted by c. For steady engine operating points, the accelerating moment M B coincides with the load moment M L so that the following equation results for the mean quantity of fuel conveyed per operating cycle:

- ML QE = c (2.2) When the conveyed quantity of fuel of a cylinder m differs by the amount ''Em from the mean quantity of fuel, the following equations result for the individual conveyed quantities, wherein the number of the cylinders of the internal combustion engine is characterised by z:

- AQErm f or i>m and i"-m (2.3a) QE 1 for i = m, (2.3b) FE + LQE,M The following formulae for the effective accelerating torques M B of the individual cylinders result from the named equations:

- A QE,m QE 1 MB, i:.' c MB, i = C.

Gges for i>mand for i<m (2. 4a) QE + AQE, M for Gges (2.4b) Resulting from the equations (2.2) and (2.4a/2.4b) for steady engine operating points is the relationship between rotational accelerations for each cylinder averaged over one operating cycle - and the injected quantities:

C. (_ 6 QE, m) -- - - N - E z - I L eges -c AQE, m 1 = - for i > m and-for < M.

eges z Resulting from this for one cylinder m is the following equation:

c - A QE, In.

Am = Gges (2.5a) (2.5b) (2. 5c) The qualitative course of rotational speed n and rotational acceleration n illustrated in Fig. 2 results from these equations for an internal combustion engine with, for example, four cylinders, wherein the illustrated values have been averaged in each case over one cylinder.

For constant mean rotational speed, thus in the steady case, the mean rotational acceleration over z operating cycles can be calculated from the following equations:

1 7 n = - z z 5-11 A i i=1 n =. (- (z Z. eges n = 0 - l) AQE, in +A QE, in) z - 1 (2.6) In the "unstationary case", thus for the case that the mean value of the accelerating torque M, is smaller or greater than the load moment ML the mean value of the individual accelerations per operating cycle results from 10 the following equations:

- 1 A = z - Gges E (z - 1) - (c - + (c (E +A QE,Tn) - 14L) 3 d - AQE, m (5E) - ML) z - 1 The following formula results initially from conversion of this equation: 1 z. E)ges [ ( (z - 1) ' (c QE - ML) + c (2.7a) QE - ML) + (- (z - 1) c QE, M + c 5E, L) 3 (2.7b) z - 1 The formula lets itself be simplified further in the following manner:

T_ 1 n = -- z. eges EZ - (c - -E - ML) 1- 01, (2.7c) The following equation results in the end:

c QE - ML (2.7d) n = eges It is evident from both the equation systems (2.6) and (2.7) that the recognition of the injected quantities fluctuating from cylinder to cylinder, thus the systematic scatters of the injected quantities, is possible also for unsteady operating points with the aid of the method exemplifying the invention. For this purpose, the "mean rotational acceleration", i.e. the rotational acceleration averaged over z operating cycles according to equation (2.6) is to be subtracted from the"instantaneous value" of the rotational acceleration, thus from the rotational acceleration averaged over one operating cycle according to equation (2. 5). if it is presumed that fluctuations of the rotational speed of the engine are due only to differing injected quantities being fed to the individual cylinders, the deviations of the injected quantities let themselves be calculated approximately from the following equation:

-911 A QE, i -",: c - (Ai - n).

In this equation n is determined by the following formula:

(2.8a) z n (2.8b) z i=1 The method of the equalisation of the cylinders shall now be entered into more precisely by reference to Fig. 3 with the aid of the relationships explained below.

Initially, the rotational speed of the engine is detected by way of at least one electrical pulse which is produced for each operating cycle of the engine. A pulse wheel, the output signal of which is evaluated in a rotational speed sensor, can be used for this purpose.

It is presumed for the following considerations that the engine operates on the four-stroke principle and that the ignition spacings are constant. Moreover, it is presumed that exactly one rotational speed pulse, the position of which is unchanged with respect to piston top dead centre (OT) of a cylinder, is produced for each operating cycle.

The generation and detection of the rotational speed pulse for the cylinder (i + 1) is indicated in a step 1 of the flow diagram shown in Fig. 3.

In a step 2 of the flow diagram, the transit time At 1 between two rotational speed pulses, which are associated with the cylinders (i + 1) and (i), is ascertained.

Starting from the time A ti 9 which el apses between two successive pulses, the instantaneous rotational speed n i is obtained from the following equation 2 1 ni = - z Ati (3. 1 a) From this equation, the mean rotational acceleration ni arising between two operating cycles can be computed from the following equation:

n nj ni-1 nj = -- = - At ti (3. 1 b) - If, for example, the derivative of the rotational speed, thus the rotational acceleration, is to be computed in segment S2, the difference between the speed n 1 in segment S1 and the speed n 2 in segment S2 is divided by the width At 2 of the segment S2 according to equation (3.1b). This form of computation is required because the rotational speed can be measured only over a segment and not at a certain instant.

The computations according to the equations (3.1a) and (3.1b) are indicated at a and b in a step 3 of the flow diagram in Fig. 3. A computation according to the equation (2.8b), for determining the mean value of the rotational acceleration, is indicated at c in the flow diagram.

For the exclusion of non-uniformities of rotational speed due to differences in the quantities of injected fuel, it is to be noted that these differences can be due to different conveying rates with constant conveying duration, to different conveying durations at constant conveying rates,or to a combination of these conditions.

For consideration of further steps of the method, it is presumed for simplification that a constant efficiency prevails and that the influence of the crankshaft angle is negligible. On the basis of these presumptions, it can be presupposed that engine rotational acceleration is directly proportional to the injected quantity of fuel.

The following relationship for the injected quantities of fuel thus results: On deviation of the rotational acceleration, caused by one cylinder, from the mean value, an additional injected quantity AQ e,il which is proportional to this deviation, is fed to this cylinder during the next injection phase so as to provide compensation. The additional injected quantity is computed according to the following equation:

1 1 it QE, i:: Copt (ni- - n 1). (4.1) In this equation, the additional quantity of fuel to be fed to the cylinder i is denoted by AQ e,il the mean rotational acceleration over two crankshaft rotations by n, the rotational acceleration caused by the cylinder i by 1 and a constant by C Opt' The individual quantities of fuel to be supplied additionally are added continuously during the performance of the method,wherein the resulting sum value is denoted by A Q zu,i and is determined by the following equation:

Q z U QE, (4.2) If the equation (4.1) is compared with the equation (2.8a), it becomes evident that the constant C Opt is to be selected in dependence on the mass inertia moment of-the engine..

Comparison of the equations (4.A.) and (4.2) with equation (2.5c) shows that computation of the balancing. quantities has a PT1 behaviour. From the equations (4.1), (2.5c) and (2.2), it can be derived that it is ture for C Opt that:

Copt = eges = G ges O-E C MB The interpretation would compensate for a non-uniformity in rotational speed by the first computation of the associated balancing quantity. A prerequisite, however, is the validity of the linearisation of the relationship between injected quantity and delivered torque.

- - 12 - It must be true in every case that CO-Pt < 2 - E)ges (2E MB This condition marks the limit of stability. If this is exceeded, this has the consequence of compensating quantities which, in the next injection phase, cause equal or greater non-uniformities in rotation of opposite sign.

The determination, which serves for equalisation between the cylinders, of the additional injected fuel quantity AQ E,i is illustrated in the first part of a step 4 of the flow diagram of Fig.

3, where the equation (4.1) is shown. The addition of the compensating quantity results from the equation in the second part of step 4, and a mean value formation is performed in the third part of step 4.

All summated compensating quantities A Q zu,i are compensated for in respect of this mean value in a step 5 of the flow diagram of Fig.

3:

z = E AQZU, K K=1 (4.3a) AQZu'j = AQZu'i &Q zu (4. 3b) for j = 1...z.

Through observance of this coupling condition, a drifting of the compensating quantities is avoided and it is ensured that the actual mean injected quantity over all cylinders is equal to the required target quantity value.

z? Alternatively to the coupl ing condition introduced by the equations (4. 3a) and (4.3b), it is possible to compute the compensating quantities Q zu corresponding to the equation (4.3b) on each determination of QE, i according to equation (4.1) in the following manner:

Qzui AQzu,i + AQE,i 1 -LQzUri = AQzUri - Z'_1 QE i (4.4a) (4. 4b) for j = 1. Z and j J i The additional injected quantity, determined according to the steps stated here, for a particular cylinder i is added to the mean injected quantity which is preset by a target value Q E,Sol 1 ' this target value being determined by way of, for example, the accelerator pedal of a vehicle fitted with the engine. The individual target val ue Q So] l 'i for the cylinder i can be computed by the following equation:

QSoll,i = QE,Soll + AQzu,i (4.5) Apart from these two mentioned methods, the possibility exists of maintaining the compensation with respect to the mean value of the compensating quantities in the following manner: Initially, an arbitrarily selected one of the cylinders of the engine is denoted by k and its compensating quantity is computed by the following equation:

AQzu,k = -E;4 QZU, i f or i > k nd i < k For all cylinders i K, the computation of AQzu'i takes place according to equations (4.1) and (4.2).

From what has been said above, in particular from the flow diagram of Fig. 3, it is evident that the computation of the additional injected quantity should preferably be concluded before the next injection phase occurs. The reason for this is that a compensating quantity, which must be taken into consideration for a cylinder with the immediately following injection phase, is influenced in every case in the consideration of the coupling condition according 10 to equation (4.4).

This results from the following method steps having to take place after occurrence of a rotational speed pulse for the cylinder i: Initially, the computation of the value A Q zu,i must be performed Subsequently, the is admetering of fuel for the cylinder (i + 1). can take place.

Thereafter, the conveying of fuel is activated. Then combustion in the cylinder (i +1) can begin.

If the time required for the admetering 9f the fuel is not taken into consideration, the actually conveyed compensating quantities according to equations (4.2) and (4.3) or (4.4).

AQ can have a mean value differing from zero in spite of the zu, i coupling condition according to equation (4.4).

To that extent, the method for observing the coupling condition, in which a single cylinder k is drawn upon for making the sum of the compensating quantities equal to zero, has the disadvantage that the coupling condition is maintained merely each two crankshaft revolutions. Consequently, the transient times of a method performed in such a manner increase slightly by comparison with both the other methods for the maintenance of the coupling condition.

It is also to be noted that rounding-off errors, by reason of which the mean value ultimately becomes unequal to zero, can arise through computation of the value AQE,i /(z - 1) with integral number arithmetic in the case of the method for compensation in respect of the mean value of all compensating quantities, which was mentioned at the second place.

By reason of these considerations, the method in step 5 illustrated in Fig. 3 is preferably to be used: After each new computation of a compensating quantity A Q zu'll the mean value of all compensating quantities for all cylinders can be computed and subtracted from all compensating quantities.

Because of the numerous successive method steps which must be performed after the occurrence of the rotational speed pulse for the cylinder i, too great a spacing between the pulse and the piston top dead centre may become necessary, in particular when taking into consideration the mass inertia of setting members, for controlling fuel injection. In that case the compensation of the injected fuel quantity for a cylinder may no longer be able to take place in the immediately following injection phase. This is indicated in a step 6 of the flow diagram of Fig. 3 by the compensation possibly not being capable of being performed for the cylinder (i + 1), but only for the cylinder (i + 2).

The cost and complication in setting an injection system may be able to be reduced substantially by such a method for adaptive equalisation between cylinders. The described method may be effective over the entire operating range of the engine, thus even in unsteady operating states.

Finally, it is also possible to separately detect extreme values arising during the addition or integration of the individual values in order to note errors of the overall system. It is thus evident that this method can also be utilised in diagnosis of the engine condition - 5 and functioning 9 1 1 1.

- 17

Claims

1. A method of equalising fuel injection between the cylinders of a fuelinjected multicylinder internal combustion engine, comprising the steps of measuring the rotational speed acceleration for each combustion process, comparing the obtained individual measurement val ues with each other to recognise deviations therebetween, and in the case of recognised deviations correspondingly varying the quantities of injected fuel until the deviations are elimiated.

2. A method as claimed in claim 1, wherein the step of measuring the rotational acceleration comprises detecting rotational speed in each of a plurality of segments, which are respectively associated with the cylinders, of a revolution, determining the difference between the detected rotational speed for two successive segments and dividing the difference by the transit time of the latter one of those two segments.

3. A method as claimed in claim 1 or claim 2, comprising the step of ascertaining the mean value of said measurement values.

4. A method as claimed in claim 3, wherein the mean value is a sliding mean over all cylinders.

5. A method as claimed in "claim 3 or claim 4, comprising the step of modifying, in response to detection of a deviation of the rotational speed acceleration measurement value for one combustion process from the mean value, the quantity of fuel injected into the associated cylinder in a subsequent injection phase for that cylinder.

6. A method as claimed in claim 5, wherein the subsequent injection phase is the phase next following that in which the deviation occurred.

7. A method as claimed in claim 5 of claim 6, wherein the step of modifying comprises adding to or subtracting from said quantity an amount substantially proportional to said detected deviation.

8. A method as claimed in any one of claims 5 to 7, comprising the step of summating amounts of fuel added to or subtracted from said injected fuel quantity by the step of modifying and determining an injection target value for the respective cylinder in dependence on the value obtained by the summation.

9. A method as claimed in any one of the preceding claims, wherein the step of varying the quantities of injected fuel comprises so controlling addition to and subtraction from the injection quantities for the individual cylinders that the sum of the addition and subtraction amounts is equal to zero.

t t

10. A method as claimed in claim 1 and substantially as hereinbefore described with reference to the accompanyings drawings.