EP0826100B1 - Process for the selective lambda control of a cylinder in a multi-cylinder internal combustion engine - Google Patents

Description

The invention relates to a method for cylinder-selective Lambda control according to a multi-cylinder internal combustion engine the preamble of claim 1.

The lambda control connects to the three-way catalytic converter today the most effective emission control method for Internal combustion engines. One delivers in the exhaust pipe oxygen sensor located upstream of the catalytic converter, usually referred to as a lambda probe, one of oxygen content in the exhaust gas dependent signal that the lambda controller so further processed that by means of a metering device, like injectors or carburetor the cylinders of the Fuel-air mixture supplied to the internal combustion engine almost complete combustion (λ = 1.00).

So-called jump probes are used as lambda probes, whose output signal jumps both at the transition from a fat to a lean, as well as at the transition changes from a lean to a rich exhaust state. Such lambda probes based on zirconium oxide or titanium oxide have response times of around 100 ms and record therefore only the oxygen content in the total exhaust gas resulting from the individual exhaust gas packs of the individual cylinders of the internal combustion engine put together.

Such a regulation of the air ratio λ of the total exhaust gas a multi-cylinder internal combustion engine, also as a global control referred to, leads to λ = 1.00 due to the existing Flow scatter of the injection valves and the different Cylinder fillings for significant deviations the individual cylinder air numbers from the setpoint.

This results in both negative influences on the raw emissions of the internal combustion engine and on the conversion rate of the catalytic converter. The concentration of CO and O ₂ in the total exhaust gas increases with increasing scatter of the individual cylinder air numbers. As a result of the exothermic conversion reactions, the increased O ₂ content in the exhaust gas leads to an additional thermal load on the catalytic converter, which, particularly when the catalytic converter is arranged close to the internal combustion engine, entails the risk of accelerated aging.

In addition, due to the cross sensitivity of the conventional jump probes to hydrogen, increased deviations of the individual cylinder air numbers lead to a drift of the air number of the total exhaust gas into the lean range, which causes a significant deterioration in the NO _x conversion of the catalyst.

In addition, due to the increasing use of variable suction systems (switching suction pipes) or variable Valve timing a balanced cylinder charge in all operating points of the internal combustion engine with the conventional means difficult to realize.

To the exhaust gas from the individual cylinders of a multi-cylinder internal combustion engine on the theoretical air-fuel ratio to hold, it is known from DE 40 40 527 A1, each individual cylinder has its own lambda sensor with jump characteristics assign in a corresponding exhaust pipe. One lambda probe each records the composition of the exhaust gas from the respective cylinder and delivers an output signal to an electronic control unit. This touches the Output signals of the two lambda probes when the respective cylinders in question are in the extension stroke or during a period that is slightly later than this and increases or decreases the fuel supply amount for the individual cylinders.

The use of according to the number of cylinders Internal combustion engine provided lambda sensors and their peripherals (e.g. for diagnosis), especially for internal combustion engines with six or more cylinders leads to one Increasing complexity and increasing the cost of the motor vehicle all in all.

From US 4,766,870 is a method for regulating the air-fuel ratio known in an internal combustion engine, with which cylinder-specific actual lambda values are determined. These values are over several cylinder signals of a work cycle averaged and one with an integral behavior Controller regarding the setpoint for all cylinders together fed. This I control is at certain speeds overlaid with a cylinder-specific regulation. There is a linear lambda sensor in each exhaust line an assigned cylinder group and the control happens in groups.

For cylinder-specific mixture control in an internal combustion engine it is also known to have a single oxygen sensor provide a linear dependency of its Output signal from the air ratio λ and beyond one has a short response time. (SAE Paper 940149 "Automatic Control of Cylinder by Cylinder Air-Fuel Mixture Using a Proportional Exhaust Gas Sensor "and SAE Paper 940376" Individual Cylinder Air Fuel Ratio Feedback Control Using an Observer ").

The solutions proposed there for single-cylinder lambda control require an internal combustion engine because of this necessary matrix operations very high computing power, so that an implementation in series engine control systems very difficult for motor vehicles with reasonable effort is to be realized.

The oxygen sensors for cylinder-specific mixture control are also known as linear lambda sensors and are for example based on strontium titanate (SrTiO3) in Thin-film technology built up (VDI reports 939, Düsseldorf 1992, "Comparison of the response speed of automotive exhaust gas sensors for quick lambda measurement based on selected metal oxide thin films ").

The present invention is based on the object of specifying a method for cylinder-selective lambda control of a multi-cylinder internal combustion engine of the type mentioned at the outset, so that the deviation of the individual cylinder air numbers from the desired value is limited to a minimum at all operating points of the internal combustion engine.
This object is achieved according to the features of patent claim 1. Advantageous further developments can be found in the subclaims.

According to the invention, the single-cylinder lambda control consists of two control loops, an outer control loop for control the global lambda mean and an internal control loop, in which the air ratio is controlled cylinder-selectively. A linear proportional integral controller is used to regulate the mean air ratio (PI controller) used. The controlled system can be with sufficient accuracy by Dead time element and two first order delay elements replicate. With the help of this route model one can Design the controller structure, whose parameters depend on the dead time of the Lambda control loop, the time constants of the delay elements and the speed are dependent. Because these system sizes the effort can be easily determined by measurements reduce significantly for the application of the lambda controller.

To identify the air ratio of the individual cylinders of the internal combustion engine is the slope of the oxygen probe signal evaluated after the expiration cycles. A positive gradient means that the air ratio in the current Extension cycle is leaner than the air ratio in the previous one Work cycle, a negative gradient in the current Extending cycle indicates a richer exhaust package. As this qualitative information about the state of the air ratio represents the single-cylinder exhaust gas, the single-cylinder lambda controller as a two-point controller. As a controller a PI controller is also used for the single-cylinder air figures used, in which the proportional and integral part set depending on the load and speed become.

By using the method according to the invention Deviations of the single-cylinder air figures from the setpoint reduce less than 1%.

In addition, the air ratio amplitude in the total exhaust gas is significantly reduced compared to that of a conventional two-point lambda regulator and the conversion rate for CO and NO _{x of} an aged catalytic converter is significantly increased. In addition, the detection and evaluation of the single-cylinder air figures makes it possible to detect defects in the injection valves which are associated with a change in the dynamic properties of the flow rate, which supports on-board diagnosis (OBD II).

Figur 1

ein Blockdiagramm einer Einrichtung zur zylinderselektiven Lambdaregelung einer Brennkraftmaschine,

Figur 2

den Zusammenhang zwischen Sondenspannung und Luftzahl einer linearen Lambdasonde,

Figur 3

die Lage der Abtastpunkte für die Sondenspannung in Bezug auf die Ausschiebetakte der einzelnen Zylinder,

Figur 4

eine graphische Darstellung einer Hysterese zur Bestimmung von Luftzahlgradienten und

Figur 5

ein Ablaufdiagramm zur Bestimmung von Zustandsgrößen, die angeben, ob das Abgas eines Zylinders zu fett oder zu mager ist.

An embodiment of the invention is explained in more detail below with reference to the schematic drawings. Show it:

Figure 1

1 shows a block diagram of a device for cylinder-selective lambda control of an internal combustion engine,

Figure 2

the relationship between probe voltage and air ratio of a linear lambda probe,

Figure 3

the position of the sampling points for the probe voltage in relation to the push-out cycles of the individual cylinders,

Figure 4

a graphical representation of a hysteresis for determining air gradient and

Figure 5

a flowchart for determining state variables that indicate whether the exhaust gas of a cylinder is too rich or too lean.

In Figure 1 with the reference numeral 10 is only schematic shown internal combustion engine BKM with 6 cylinders, 3 cylinders combined to form a cylinder bank are. The cylinders are a first cylinder bank ZB1 1,2,3 assigned, the exhaust gas in a common exhaust line AST1 opens. The cylinders 4,5,6 are a second cylinder bank ZB2 assigned to which an exhaust line AST2 common is.

To determine the air ratio λ in the exhaust line AST1 is the Internal combustion engine 10 has a linear lambda probe LS1 in the exhaust line AST2 a linear lambda probe LS2 is provided. A positioning of the two lambda sensors LS1, LS2 near the internal combustion engine 10 favors the detectability of single-cylinder air deviations, because with increasing distance the installation location of the lambda probes LS1, LS2 from the Internal combustion engine 10 the degree of mixing of each Exhaust gas packs increased and thereby cylinder-selective detection is difficult.

The signals from the two lambda sensors LS1, LS2 become one Circuit block 11 supplied, the signal detection and a Controls linearization of these signals. This is due to Circuit block 11 as further input variables a cylinder identification signal ZID and a time signal, namely the Waiting time TEZ on. The value for the waiting time TEZ is made up of a Map KF depends on one that represents the engine load Size, for example the air mass LM and the Speed N read out.

In Figure 2, the dependence of the probe voltage is linear Lambda probe represented by the air ratio λ. In one narrow range of 0.97 <λ <1.03 results in an almost linear relationship between probe voltage ULS and air ratio λ. The shows in the rich and lean air ratio range Probe characteristic a saturation behavior. The probe voltage is by means of a stored characteristic curve or a one-dimensional one Map converted into an actual lambda value LAM_IST.

A separate map can be made for each of the two lambda sensors are provided, with the aid of which the values of the sensor voltages be converted into air ratio values.

Order from the detected probe voltage values ULS of the two Lambda sensors Information about the air figures of the individual To get cylinders, it is necessary to measure the probe voltages ULS to a defined in terms of crank angle Feel position. As a reference for the temporal position of the samples are the top dead center ignition (ZOT) of the single cylinder used. For this purpose, reference marks, e.g. Teeth on one associated with the crankshaft or the camshaft Encoder wheel evaluated (e.g. tooth 15: ZOT cylinder 5, tooth 35: ZOT cylinder 3, tooth 55: ZOT cylinder 6, tooth 75: ZOT cylinder 2, tooth 95: ZOT cylinder 4, tooth 115: ZOT cylinder 1).

Figure 3 shows the position of the sampling points in the first two lines AP for the sensor signals of the two cylinder banks ZB1, ZB2 in relation to the push-out cycles AT of the individual Cylinder. In the 3rd line of Fig.3 are the push-out cycles AT of cylinders 4, 5 and 6 of cylinder bank ZB 2, in the 4th Row are the extension strokes AT of cylinders 1, 2 and 3 the cylinder bank ZB 1 shown. In addition, in the last line of Figure 3, a cylinder identification signal ZID drawn, on which the respective top dead center ignition (ZOT) cylinders 1 to 6 are marked.

To take into account the exhaust gas runtime from the exhaust valves to the respective lambda probe, the value of the probe signal, which contains the information about the air ratio of a cylinder, is only recorded after a specific waiting time TEZ after the exhaust valve has closed (the end of the push-out cycle). This waiting time TEZ depends on the load and the speed of the internal combustion engine. In the case of an air mass-controlled control of the internal combustion engine, the waiting time TEZ is stored in a map that is spanned over the air mass LM and the speed N. After this waiting time TEZ (time between reference mark and sampling time) has elapsed after an ignition TDC has been exceeded, the values of the sensor signals of the lambda sensors assigned to the two cylinder banks ZB1, ZB2 are queried.
The time interval between the signal acquisition is therefore predefined in relation to a trigger mark (tooth number) fixed to the crankshaft, depending on the load and the speed. A lambda voltage value per cylinder bank is determined for each segment.

The accuracy of the subsequent calculation of the mean lambda value Increasing all cylinders will result in an additional one Sample AP between two push-out cycles AT recorded.

Serves as a global lambda controller for controlling the total exhaust gas a proportional integral controller (PI controller) with the proportional component LAM_P and the integration component LAM_I (Circuit block 14 in Fig. 1) Lambda mean LAMMW_IST and the target value LAM_SOLL this Controller shares calculated. The setpoint LAM_SOLL is one Map dependent on the load, for example on the air mass LM and the speed N of the internal combustion engine filed.

To calculate the lambda mean LAMMW_IST_i (i = 1.2 for the two lambda probes) are 6 for each exhaust line Lambda measured values LAM_IST_i per work cycle, corresponding to 2 Crankshaft revs recorded and saved:

LAM_IST_i

n-6 n-5 n-4 n-3 n-2 n-1 n = number of the measured value

The lambda mean value LAMMW_i is calculated segment-synchronously for each lambda probe using the following formulas: LAM_SUM_i (n) = LAM_SUM_i (n-1) - LAM_IST_i (n-6) + LAM_IST_i (n) LAMMW_i (n) = LAM_SUM_i (n) / 6

This calculation is carried out in circuit block 12 (FIG. 1).

The input variable for the global lambda controller is the control deviation LAM_DIF_i (n), which is defined as the difference between the setpoint value LAM_SOLL (n) taken from the map mentioned above and the average lambda value LAMMW_IST (n): LAM_DIF_i (n) = LAM_SOLL (n) - LAMMW_IST_i (n)

LAM_KPI_FAK =

Regelverstärkungsfaktor (zB.0-2)

P_FAK_LAM_GR =

Applizierbare Konstante (zB.0-2)

I_FAK_LAM_GR =

Applizierbare Konstante (zB.0-2)

T_LS =

Applizierbare Zeitkonstante (zB.0-0.043)[sec]

TN =

Segmentdauer [sec]

The lambda controller components LAM_P_i and LAM_I_i of the global lambda controller are calculated as follows: LAM_P_i (n) = LAM_KPI_FAK (n) * P_FAK_LAM_GR * (T_LS + TN) * LAM_DIF_i (n) LAM_I_i (n) = LAM_I_i (n-1) + LAM_KPI_FAK (n) * I_FAK_LAM_GR * 2 * TN * LAM_DIF_i (n) With:

LAM_KPI_FAK =

Control gain factor (e.g. 0-2)

P_FAK_LAM_GR =

Applicable constant (e.g. 0-2)

I_FAK_LAM_GR =

Applicable constant (e.g. 0-2)

T_LS =

Applicable time constant (e.g. 0-0.043) [sec]

TN =

Segment duration [sec]

The control gain factor LAM_KPI_FAK is selected depending on a dead time LAM_TOTZ_GR in the lambda control loop, which results from the fuel storage period, the Duration of the intake, compression, work and extension cycle as well as the gas running time for the respective lambda probe. This dead time LAM_TOTZ_GR is load- a map and taken depending on the speed.

The influence of the global lambda controller results from the sum of the controller components LAM_P_i and LAM_I_i: LAM_GR_i (n) = LAM_P_i (n) + LAM_I_i (n)

This controller output of the global lambda controller is preferably limited to ± 25% of the basic injection time, ie -0.25 <LAM_GR_i <0.25. The integral component can also be limited to ± 25% of the basic injection time, ie - 0.25 <LAM_I_i <0.25.

A gradient method is used to identify the individual cylinder air numbers used. The slope behavior of the lambda probe signal after the expiration cycle a qualitative assessment of the individual cylinder air numbers carried out, i.e. it is determined whether the exhaust gas of the current Cycle is richer or leaner than that exhaust gas from the previous cycle.

This identification of the single cylinder air numbers is in the Circuit block 13 (Fig. 1) carried out in the following manner:

The air ratio gradients are calculated segment-synchronously cylinder-selective from the actual lambda values LAM_IST_i, whereby only every second measured value per cylinder bank for the gradient calculation is taken into account.

x:

Zylindernummer 1...6

i:

Sondennummer 1,2

The general calculation formula is: LAM_GRD_ZYL_x = LAM_IST_i (n) - LAM_IST_i (n-2) With

x:

Cylinder number 1 ... 6

i:

Probe number 1.2

Depending on the crankshaft tooth number with which the scanning of the probe signals is triggered, the following values for x and i result: Tooth point scanning point: Gradient for cylinder No.x Probe No.i 15 2nd 1 35 4th 2nd 55 1 1 75 5 2nd 95 3rd 1 115 6 2nd

From this table it can be seen that e.g. the measured value of Lambda probe No. 1, its scanning by crankshaft tooth No. 15 was triggered to calculate the air ratio gradient of cylinder # 2 is used.

The evaluation of the air ratio gradients provides so-called state variables as a result:
LAM_ZST_ZYL_i with i = 1.2.

If the exhaust gas of a cylinder is detected as too rich, the state variable LAM_ZST_ZYL_i = 0 is set, if the exhaust gas of a cylinder is too lean, the state variable becomes
LAM_ZST_ZYL_i = 1 set.

To suppress interference, especially in small ones Air ratio gradients can lead to false detections introduced a hysteresis LAM_ZST_HYS, the width of which can be applied is.

This hysteresis is shown graphically in FIG. Lies is the air ratio gradient calculated using formula (1) LAM_GRD_ZYL_x within the range ± LAM_ZST_HYS, so is the result of the gradient evaluation from the previous state dependent in the relevant exhaust line. To the procedure two more state variables are easier to design VOR_ZST 1, VOR_ZST 2 introduced.

The state variable VOR_ZST 1 saves the previous one Condition in the exhaust system of the first cylinder bank with the Probe 1, the state variable VOR_ZST 2 the previous state in the exhaust line of the second cylinder bank with the probe 2. Depending on the values of these state variables VOR_ZST 1,2 there is a sequence for determining the values (1 or 0) for LAM_ZST_1.2, as shown in Figure 5.

In a first step S1, a query is made as to whether the state variable VOR_ZST_i = 0. If the result of this query is positive, it is checked in step S2 whether the using the Formula (1) calculated value of the air ratio gradient LAM_GRD_ZYL_x is less than the hysteresis value + LAM_ZST_HYS. If this is the case, the state variable LAM_ZST_i = 0 is set (step S3), otherwise LAM_ZST_i = 1 is set (Step S4).

If the query in step S1 yields a negative result, then it is checked in step S5 whether the value of the air ratio gradient LAM_GRD_ZYL_x is less than the hysteresis value - LAM_ZST_HYS. If this is the case, the state variable becomes LAM_ZST_i = 0 is set (step S6), otherwise LAM_ZST_i = 1 set (step S7).

These state variables LAM_ZST_i are used to control the individual cylinder air numbers used. They serve as input variables for a single cylinder lambda controller (circuit block 15 in Fig. 1), which is used as a proportional integral controller (PI controller) is trained.

The circuit blocks 11-15 in Fig. 1 are preferably in one known electronic control device 16 integrated, as used for control in modern motor vehicles anyway and control of various operating parameters such as e.g. Injection time calculation, ignition control, diagnosis, etc. is used. Also those mentioned in the description Characteristic maps are stored in memories of the control device 16.

1. Fall:
LAM_ZST_i = O   (Abgas eines Zylinders ist zu fett) LAM_P_EZ_x(n) = -LAM_P_EZ(n) LAM_I_EZ_x(n) = LAM_I_EZ_x(n-1) - LAM_I_EZ(n) - LAMMW_I_EZ_i(n)

2. Fall:
LAM_ZST_i = 1   (Abgas eines Zylinders ist zu mager) LAM_P_EZ_x(n) = LAM_P_EZ(n) LAM_I_EZ_x(n) = LAM_I_EZ x(n-1) + LAM_I_EZ(n) - LAMMW_I_EZ_i(n)

The calculation of the controller components - proportional component LAM_P_EZ_x and integral component LAM_I_EZ_x - of the single cylinder lambda controller takes place depending on the value (1 or 0) that the status variable LAM_ZST_i has:

1st case:
LAM_ZST_i = O (exhaust gas from a cylinder is too rich) LAM_P_EZ_x (n) = -LAM_P_EZ (n) LAM_I_EZ_x (n) = LAM_I_EZ_x (n-1) - LAM_I_EZ (n) - LAMMW_I_EZ_i (n)

2nd case:
LAM_ZST_i = 1 (exhaust gas from a cylinder is too lean) LAM_P_EZ_x (n) = LAM_P_EZ (n) LAM_I_EZ_x (n) = LAM_I_EZ x (n-1) + LAM_I_EZ (n) - LAMMW_I_EZ_i (n)

The calculation of the mean value of the I components of the cylinders of a cylinder bank LAMMW_I_EZ_i takes place segment-synchronously alternating with i = 1 or i = 2 as follows: LAM_I_SUM_EZ_i (n + 1) = LAM_I_SUM_EZ_I (n) - LAM_I_EZ_i (n-2) + LAM_I_EZ_x (n)

The value LAM_I_EZ_x (n) is entered in a memory LAM_I_EZ_i.

LAM_I_EZ_i n-2 n-1 n LAMMW_I_EZ_i (n + 1) = LAM_I_SUM_EZ_i (n + 1) / 3

The values LAM_P_EZ and LAM_I_EZ are each in a map filed over the load size LM and the speed N of the Internal combustion engine are clamped.

The integration component LAM_I_EZ_x of the single cylinder lambda controller is, for example, ± 10% of the basic injection time TI_B limited, i.e. -0.1 <LAM_I_EZ_x <0.1.

für x = 1,2,3: i = 1

für x = 4,5,6: i = 2

When calculating the cylinder-specific injection time TI_x, the output variable of the global lambda regulator and the output variable of the single cylinder lambda regulator are taken into account: TI_x = TI_B * ........ (1 + TI_LAM_x) with TI_LAM_x = LAM_GR_i + LAM_P_EZ_x + LAM_I_EZ_x

for x = 1,2,3: i = 1

for x = 4,5,6: i = 2

The invention was explained on the basis of an exemplary embodiment in which the internal combustion engine has 6 cylinders and in each case 3 cylinders are combined to form a group (cylinder bank ZB1, ZB2). Each group or cylinder bank is assigned an exhaust line containing a linear lambda probe.
It is also possible within the scope of the invention, for example, to provide a single exhaust line in a 4-cylinder internal combustion engine, in which a single linear lambda probe is arranged, or to form 2 groups of 4 cylinders in an 8-cylinder internal combustion engine, or in a 12 Cylinder internal combustion engine to form 3 groups of 4 cylinders or 4 groups of 3 cylinders. The number of exhaust gas lines and thus the number of linear lambda sensors are then determined in accordance with the number of groups.

Claims (15)

1. Method for cylinder-selective control of the air/fuel ratio of an internal combustion engine (10) comprising several cylinders (x),

   having a control unit (16) which calculates a basic injection signal (TI_B) depending on a variable (LM) representing the load on the engine (10) and on the speed (N) of the engine (10)

   having a lambda control system with at least one exhaust gas stream (AST1, AST2), where each exhaust gas stream (AST1, AST2) has allocated to it an oxygen sensor (LS1, LS2) which delivers a sensor signal (ULS1, ULS2) representing the oxygen content of the overall exhaust emission resulting from the individual exhaust gas packets of the individual cylinders (x), and where

   for each value of the sensor signal (ULS1, ULS2) the associated actual value of lambda (LAM_IST_i(n)) is determined from a characteristic curve,

   from these values (LAM_IST_i(n)) a mean lambda value (LAMMW_IST_i(n)) being formed for each oxygen sensor (LS1, LS2), and where

   the difference LAM_DIF_i(n) between a target lambda value (LAM_SOLL_i(n)) determined by the load on the engine (10) and the mean actual lambda value (LAMMW_IST_i(n)) is utilized as the input quantity for a global lambda controller (14) and is fed to a global lambda controller (14) of the lambda control system for correction of the basic injection signal (TI_B), so that a theoretical air/fuel ratio (λ = 1) can be set, and where

   a proportional-plus-integral controller is used as the global lambda controller (14), having a proportional component LAM_P_i(n) = LAM_KPI_FAX(n) • P_FAX_LAM_GR • (T_LS + TN) • LAM_DIF_i(n) and an integral-action component LAM_I_i(n) = LAM_I_i(n-1) + LAM_KPI_FAX(n) • I_FAK_LAM_GR • 2 • TN • LAM_DIF_i(n) where:

   LAM_KPI_FAX = Controller gain factor

   P_FAK_LAM_GR = Variable constant

   I_FAK_LAM_GR = Variable constant

   T_LS = Variable time constant [sec]

   TN = Segment duration [sec]

   where, furthermore, the lambda control system comprises an individual-cylinder lambda controller (15) for controlling the individual air/fuel ratio of the individual cylinders (x), and

   where the cylinder-selective output quantity (LAM_P_EZ_x, LAM_I_EZ_x) of said individual-cylinder lambda controller (15) is superimposed on the output quantity (LAM_GR_i) of the global lambda controller (14), and

   where the value (TI_LAM_x) obtained therefrom is used to correct the basic injection signal (TI_B) on a per-cylinder basis.
2. Method in accordance with Claim 1, characterized in that

   the cylinders (x) are combined into at least one group (ZB1, ZB2),

   each group (ZB1, ZB2) of cylinders (x) has allocated to it one exhaust gas stream (AST1, AST2)

   a linear oxygen sensor (LS1, LS2) is provided in each exhaust gas stream (AST1, AST2), delivering a signal (ULS1, ULS2) which corresponds to the oxygen content of the exhaust gas emission of the individual cylinders (x),

   the signals (ULS1, ULS2) from the sensors (LS1, LS2) are sampled in particular positions (AP) defined with respect to the crank angle.
3. Method in accordance with Claim 2, characterized in that the ignition top dead centre points (ZOT) are used as reference points for the position in time of the sampling points (AP) and that the sensor signals (ULS1, ULS2) are sampled at the end of a waiting time (TEZ) following the transition of the ignition top dead centre point (ZOT).
4. Method in accordance with Claim 3, characterized in that the waiting time (TEZ) is chosen depending on a parameter (LM) representing the load on the internal combustion engine (10) and on the speed (N) of the internal combustion engine (10).
5. Method in accordance with Claim 1, characterized in that the controller gain factor (LAM_KPI_FAK) is chosen depending on a dead time (LAM_TOTZ_GR) which is determined by the fuel pre-feed time, the duration of the induction, compression, power and exhaust strokes together with the gas travelling time to the relevant oxygen sensor, and is taken from a characteristics map as a function of load and speed of rotation.
6. Method in accordance with Claim 1, characterized in that the values of the output quantity (LAM_GR_i) of the global controller (14) and of the integral-action component (LAM_I_i) of the global controller (14) are limited to ± 25% of the basic injection signal (TI_B).
7. Method in accordance with Claim 1, characterized in that the air ratio gradients (LAM_GRD_ZYL_x) are calculated on a cylinder-selective basis from the measured actual lambda values (LAM_IST_i) by taking the differences between actual lambda values (LAM_IST_i), with only every second actual lambda value per group (ZB1, ZB2) being used for calculating the air ratio gradient, and that when positive air ratio gradients (LAM_GAD_ZYL_x) are measured in the current cycle leaner emission from the respective cylinder (x) than in the preceding cycle is assumed to exist, whilst when negative air ratios (LAM_GRD_ZYL_x) are measured in the current cycle richer emission from the respective cylinder (x) than in the preceding cycle is assumed to exist.
8. Method in accordance with Claim 7, characterized in that depending on the sign of the individual air ratio gradients (LAM_GRD_ZYL_x), said gradients have state variables (LAM_ZST_ZYL_i) allocated to them which take the value I or 0.
9. Method in accordance with Claim 8, characterized in that the allocation of the state variables (LAM_ZST_ZYL_i) takes place via a hysteresis (LAM_ZYST_HYS) the width of which can be set, and that in the event of the calculated air ratio gradient (LAM_GRD_ZYL_x) falling within double the width of the hysteresis (± LAM_ZST_HYS), it is decided that the result of the gradient evaluation is dependent on the preceding state in the emission stream concerned (AST1, AST2) and that said state is taken account of in the allocation of the state variable (LAM_ZST_ZYL_i).
10. Method in accordance with Claim 9, characterized in that the integral-action controller component (LAM_I_EZ_x) and the proportional controller component (LAM_P_EZ_x) of the individual-cylinder lambda controller (15) are calculated separately, depending on the value of the state variable (LAM_ZST_x).
11. Method in accordance with Claim 10, characterized in that when the state variable (LAM_ZST_x) has the value 0, the proportional controller component (LAM_P_EZ_x) is formed in accordance with the following rule LAM_P_EZ_x(n) = -LAM_P_EZ(n) and the integral-action controller component (LAM_I_EZ_x) is formed in accordance with the following rule LAM_I_EZ_x(n) = LAM_I_EZ_x(n-1) - LAM_I_EZ(n) - LAMMW_I_EZ_i(n) where n = running index of the measured value
       LAMMW_I_EZ_i(n) = mean value of lambda
12. Method in accordance with Claim 10, characterized in that when the state variable (LAM_ZST_x) has the value 1, the proportional controller component (LAM_P_EZ_x) is formed in accordance with the following rule LAM_P_EZ_x(n) = LAM_P_EZ(n) and the integral-action controller component (LAM_I_EZ_x) is formed in accordance with the following rule LAM_I_EZ_x(n) = LAM_I_EZ_x(n-1) + LAM_I_EZ(n) - LAMMW_I_EZ_i(n) where n = running index of the measured value
       LAMMW_I_EZ_i(n) = mean value of lambda.
13. Method in accordance with Claims 11 or 12, characterized in that the mean lambda value (LAMMW_I_EZ_i) of the integral-action components (LAM_I_EZ) of a group (ZB1, ZB2) is formed in accordance with the following rule LAMMW_I_EZ_i(n+1) = LAM_I_SUM_EZ_i(n+1)/3 where LAM_I_SUM_EZ_i(n+1) = LAM_I_SUM_EZ_i(n) - LAM_I_EZ_i(n-2) + LAM_I_EZ_x(n) with
    LAM_I_SUM_EZ_i(n+1) = new total
    LAM_I_SUM_EZ_i(n) = old total
14. Method in accordance with Claim 10, characterized in that the integral-action controller component (LAM_I_EZ_x) of the individual-cylinder lambda controller (15) is limited to ± 10% of the basic injection signal (TI_B).
15. Method in accordance with Claim 10, characterized in that the integral-action controller component (LAM_I_EZ_x) and the proportional controller component (LAM_I_EZ_x) of the individual-cylinder lambda controller (15) are stored in characteristics maps as a function of engine load and speed.

Images