EP2142783B1 - Method and device for controlling an internal combustion engine - Google Patents

Description

State of the art

The invention is based on a method and a device for controlling an internal combustion engine. Such a device and such a method is for example from the DE 33 36 028 known. There, a method for controlling an internal combustion engine is described in which, starting from operating parameters, control variables for at least one actuator are specified. In this case, each cylinder of the internal combustion engine is assigned a controller which adjusts the torque output by the cylinder to a common desired value. For this purpose, in particular the speed signals are adjusted to a common setpoint. Such an approach is commonly referred to as restraint. In this case, based on the deviation of the individual cylinders from a common average value, a correction value for the fuel quantity of the individual cylinders to be injected is formed.

From the US 2002/0134340 a method is known in which it is concluded on the fuel quality. The procedure is as follows. As soon as the internal combustion engine is running stably, the inlet valve, via which the mixture or air flows into the combustion chamber, is opened. After opening the inlet valve, it is checked whether the speed drops. If the speed does not drop, this means that the fuel is of sufficient quality. If the speed drops, the fuel is of poor quality.

This exploits the fact that, when the intake valve closes, the passage area in the intake passage decreases and the negative pressure downstream of the intake valve increases. This leads to a better turbulence of the incoming mixture. This makes a lean burn possible. If the valve is opened, the combustion deteriorates, which leads to a drop in engine rpm if the fuel quality is poor. That is, based on this Dehzahlabfall closed on the fuel quality.

Often, different quality fuels are used in the operation of diesel engines. Among other things, this leads to the internal combustion engine emitting increased or reduced power or increased exhaust emissions occurring. Especially with poor fuels, such an increased exhaust emission occurs.

With the procedure according to the invention, it is possible to recognize such different fuel qualities and react accordingly to these different fuel qualities. According to the invention, it is provided that, based on a variable that characterizes the rough running, the fuel properties are concluded. The size that characterizes the rough running is considered to be a variable caused by stochastic torque fluctuations. According to the invention, it has been recognized that poor fuel qualities cause such stochastic torque fluctuations.

A deteriorated smoothness is detected and appropriate countermeasures are introduced. According to the invention, provision is made for at least one control variable to be corrected if poor fuel quality is detected by the fact that the variable characterizing the rough running exceeds a threshold value.

As a measure, it is provided in particular that the time at which the injection is changed is changed, that the amount of air that is supplied to the internal combustion engine, that the pressure of the fuel is changed and / or that in a diesel internal combustion engine, an annealing process is initiated. These measures are used individually or in combination. In particular, the start of injection is adjusted in the direction of early, the amount of air is corrected in the direction of higher air flow and the rail pressure is adjusted towards larger rail pressures.

It is particularly advantageous if this corrective action is at least partially canceled when certain states are present, that is to say the correction value is set to zero or the correction value is reduced to a value which is smaller in magnitude. These particular operating conditions are present in particular when, for example, a refueling process has been detected. Furthermore, it can be provided that this withdrawal of the correction values takes place at certain intervals, in particular in certain time intervals or after a certain mileage is withdrawn.

Stochastic fluctuations are recognized by the speed increase caused by the combustion in one of the cylinders and / or the difference between successive minima and maxima in the instantaneous speed being evaluated. For evaluation, the difference or the speed increase are normalized.

The stochastic fluctuations are characterized by the fact that they do not occur regularly. They usually occur only once for the same cylinder in successive combustion cycles.

Brief description of the drawings

Embodiments of the invention are illustrated in the drawings and explained in more detail in the following description. Show it FIG. 1 a block diagram of a device according to the invention and FIG. 2 and 3 in each case two flow charts to illustrate the procedure according to the invention.

Embodiments of the invention

In FIG. 1 a control of an internal combustion engine is shown in simplified form on the basis of a block diagram. The elements described below are part of an engine control unit. Such an engine control unit processes various signals and controls various actuators in the area of the internal combustion engine.

A controller 100 is acted upon by a controller 110 with a drive signal S via a node 105. This controller 110 processes various input signals of various sensors 120 or various variables present in an engine controller. Based on these quantities, the controller 110 predefines the drive signal S, with which the actuator 100 is acted upon.

These controllers can be a simple controller in which the actuating signal is predetermined based on the input variables. Furthermore, this may also be a regulation, such as a speed control, in which, based on the comparison between an actual value and a desired value, a corresponding manipulated variable S is specified.

Such controls are provided for various variables in the field of an internal combustion engine. Such a controller becomes, for example, the controller the injection timing, the rail pressure, the internal combustion engine supplied amount of air and / or an annealing of a glow plug used.

Injection timing control determines the time at which injection begins. This quantity has a significant influence on the fuel burning behavior of a diesel engine. The amount of air that is supplied to the internal combustion engine is specified depending on different sizes and can be adjusted by means of various controllers. As such a controller, for example, an exhaust gas recirculation valve is provided. Another major influence on combustion is the rail pressure, which corresponds to the fuel pressure during metering. In addition to these sizes, other sizes can be controlled in a corresponding manner.

A second sensor 130 provides a signal N indicative of random torque fluctuations. Such a signal is provided for example by a speed sensor. This signal reaches a rough-motion detection 140. This rough-motion detection is designed such that it detects stochastic torque fluctuations and outputs a corresponding signal IS to a correction-value determination 150. If such stochastic torque fluctuations detected, the correction value determination 150 is a corresponding correction signal K to the node 105. In the node 105, the signal K and the signal S of the controller 110 is preferably additively linked and then used to control the actuator 100.

A corresponding procedure is in FIG. 2 illustrated by a flow chart.

In a first step 200, a signal is evaluated which indicates a stochastic torque fluctuation. For this purpose, in particular the signal of a speed sensor is used. Usually, incremental wheels with a 6 ° KW resolution are used in the vehicle. On the circumference of a increment wheel 60 minus 2 teeth are arranged. The evaluation evaluates the sequence of these teeth and thereby receives a speed signal with an angular resolution of 6 ° crankshaft. By a suitable evaluation, eg a segment-synchronous Speed detection, stochastic torque fluctuations are detected from this signal.

The query 210 checks whether the intensity IS of these stochastic torque fluctuations are greater than a threshold SW. If this is not the case, then step 200 is repeated. If this is the case, then in step 220 a corresponding reduced fuel quality is detected and corresponding countermeasures are initiated. In this embodiment, the quantity IS may also be referred to as a fuel quality index.

As countermeasures it is provided, inter alia, that a correction value K is specified, with which the corresponding manipulated variables are corrected. After the correction has been successful, a new evaluation of the speed signal is performed in step 230 in order to detect corresponding stochastic torque fluctuations. The query 240 again checks whether the intensity IS of these stochastic torque fluctuations is greater than a threshold value. If so, the correction is maintained in step 250. If this is not the case, it is detected in step 260 that the stochastic torque fluctuations are based on a different cause and not on a reduced fuel quality.

According to the invention, it is thus provided that the stochastic torque fluctuations are detected and, if these exceed a certain level, a correction value K for the correction of a suitable manipulated variable is specified. If this correction of the manipulated variable results in a reduction of the stochastic fluctuations, then the correction values are retained and in the following the manipulated variable is corrected with the corresponding correction value K.

The test of whether torque fluctuations are present preferably takes place at idle, since these are recognized there particularly reliably and easily. The correction of the manipulated variables by means of the correction value K is active in all operating states.

It is particularly advantageous that the correction value K or other variables, from which the correction value is determined, stored in a memory is, which does not lose its content when you turn off the controller, or the internal combustion engine. Preferably, an EEPROM is used for this purpose. As the quantity from which the correction value is determined, in particular the intensity IS of the stochastic fluctuations or the fuel quality index are stored. When restarting the engine, these variables are immediately available for controlling the internal combustion engine.

If this measure does not lead to success, further measures, which are not the subject of the invention, must be carried out. A success is recognized, for example, if, after the correction of a manipulated variable, the intensity IS of the stochastic fluctuations is smaller than before the correction.

In this simplified embodiment, only one manipulated variable is corrected when the stochastic fluctuations exceed a certain intensity.

In an improved embodiment, it is provided that the correction value K is predefined as a function of the threshold value SW and / or that, depending on the threshold value SW, it is also determined which subset of said manipulated variables will be corrected. In this case, a plurality of thresholds are provided, which reacts differently when the respective threshold values are exceeded. Furthermore, it can be provided that, depending on the intensity IS of the fluctuations, it is determined which value the correction value assumes or which manipulated variables are corrected.

If the case now occurs that better fuel was refueled in a subsequent refueling process, a corresponding correction is no longer necessary or counterproductive. Therefore, the invention provides that is checked at certain intervals, if this correction is still necessary. For this purpose, it is checked in a step 300 whether a particular condition exists. Thus, for example, it can be checked here whether a specific time condition exists. That is, the review is done at specific intervals. Alternatively, it can also be provided that the check is carried out according to a specific driving performance of the vehicle and / or a specific number of engine revolutions. Furthermore, it can be provided be that the check is performed after each start of the internal combustion engine and / or after each refueling operation. It is particularly advantageous if, after the refueling process, a certain amount of time is left to wait.

If the query 300 detects that one of these conditions is present, an evaluation is carried out in step 310 as to whether stochastic torque fluctuations exist. If the query 320 recognizes that the intensity IS of the fluctuations is greater than a threshold value, then it is detected in step 330 that low-quality fuel continues to be used. On the other hand, if the query 320 recognizes that the intensity of the fluctuations is less than the threshold value SW, it is detected in step 340 that the fuel quality has changed. Therefore, in step 340, the correction values are canceled.

That is, depending on the embodiment, the correction values are set zero or reduced by a certain amount or a certain factor. Subsequently, in step 350, a new evaluation takes place whether fluctuations occur. If the query 360 detects that the intensity of the fluctuations IS is smaller than a threshold value SW, then it is detected in step 370 that the fuel quality is good again. If the query 360 recognizes that the intensity IS of the fluctuations is greater than the threshold value, a new correction is performed in step 380 and it is determined that the fuel quality is still poor.

This means that it is checked at certain intervals whether a withdrawal of the corrections causes the stochastic fluctuations to recur. If this is the case, i. if the correction is withdrawn, the fluctuations recur, it can be assumed that the fuel quality has not improved. In this case, a correction of the corresponding manipulated variable continues to take place. If a withdrawal of the correction value does not result in any fluctuations, then it can be assumed that refueling has improved the fuel quality again. In this case, it can be assumed that the fuel quality has returned to its normal quality.

Depending on the embodiment, it can be provided that the return of the correction value takes place in one step, ie that the correction value K is set to zero becomes. In one embodiment, it may also be provided that the withdrawal takes place in several steps or with another function.

The check on this has improved the fuel quality is preferably carried out at certain intervals or after a certain period of operation of the internal combustion engine and / or a certain driving distance of the vehicle. Furthermore, it can be provided that the check is carried out after each refueling operation, preferably after the refueling process a certain time condition must be fulfilled.

According to the invention, the check is made as to whether the fuel quality has improved if at least one of the above conditions is fulfilled. In an advantageous embodiment, it is provided that all or several conditions are checked and, when a condition is present, the described procedure is carried out. In a simplified embodiment, only one of the conditions is checked.

Furthermore, this check is carried out, whether after withdrawal of the corrections again fluctuations occur, preferably only at idle after the conditions, such as refueling, or distance since the last review are met.

It is particularly advantageous that the determined fuel quality is permanently stored in the engine control in order to be available again at the next engine start.

The following describes the detection of stochastic fluctuations. According to the invention misfires are detected by a misfire detection. The number of detected dropouts is used as the intensity IS of the stochastic fluctuations. Alternatively, a fuel quality ratio can be determined using the procedure described below. This can then be like in FIG. 2 and 3 described instead of the intensity IS of the fluctuations are processed. The intensity IS of the fluctuation can also be referred to as the fuel quality index.

In the case of known misfire detection, speed increases due to combustion are determined and evaluated in the region close to the idling. The speed rise dn are averaged over the past working cycle and subtracted from the current value dnk. $$d{\tilde{n}}_k=\frac{1}{\mathit{Zyl}}{\displaystyle \underset{k=0}{\overset{\mathit{Zyl}-1}{\Sigma }}}{\mathit{dn}}_k\left(\mathrm{average\; increase\; over\; a\; work\; game}\right)$$

Dnk falls below an applicable threshold x * d n _k , a misfire is detected (0 <x <1).

Stochastic dropouts are recognized with the following configuration. There is no comparison with the mean increase d n _k , but from the current value, the average increase is subtracted and multiplied by the average speed divided by a normalization factor. $${\mathit{dn}}_{\mathit{MISF}.k}=\left(d{n}_k-d\tilde{n}.k\right)*\frac{n}{\mathit{standardization}}\mathrm{,}$$

Negative values indicate delays. If a predetermined negative threshold is undershot, there is a misfire. This change allows misfire detection throughout the RPM range as well as cylinder assignment.

In a further embodiment, the cubic sum is formed via a work cycle. This is done with the following formula: $$y_k=\mathrm{\Sigma }{{\mathit{dn}}^3}_{\mathit{MISF}.k}$$

These measures suppress noise and emphasize high fluctuations due to dropouts.

Furthermore, statistics about cylinder-specific speed / torque fluctuations can be formed by taking into account the value of the last working cycle of the respective cylinder. The difference between the current value and that of the last working cycle is formed. $${{\mathit{dn}}^{*}}_k={\mathit{dn}}_{\mathit{MISF}.k}-{\mathit{dn}}_{\mathit{MISF}.k-\mathit{Zyl}}$$

Using conventional statistical methods, the fluctuation can be calculated from the cylinder-specific values k = 0..Zyl-1. (moving standard deviation or amount formation and PT1 filtering).

If one compiles a corresponding statistic not for the individual but over all dn * _k one receives a measure SI for stochastic torque fluctuations of the entire engine.

From the DE 10 2006 018 958 For example, a misfire detection is known that does not rate the speed increases but the differences of successive minima and maxima. For the difference of the successive minima and maxima of the speed, the formula applies: $${\mathit{dn}}_k=\left(n_k-n_{k-2}\right)\mathrm{,}$$

To suppress dynamic problems by a mean acceleration becomes a mean difference dn _k formed over the past work cycle and deducted from the current value dnk. Loss is detected when the value thus formed falls below a certain negative threshold.

$$\overline{{\mathrm{dn}}_{\mathrm{k}}}=\frac{1}{2*\mathit{Zyl}}{\displaystyle \sum _{k=0}^{2*\mathit{Zyl}-1}}{\mathit{dn}}_k$$

$${{\mathit{dn}}^{*}}_k={\mathit{dn}}_k-{\overline{\mathit{dn}}}_{\mathit{k}}$$

In one embodiment, the difference of the minima or maxima is normalized according to the following formula: $${\mathit{dn}}_k=\left(n_k-n_{k-2}\right)\frac{n}{\mathit{standardization}}$$

$$\widetilde{{\mathrm{dn}}_{\mathrm{k}}}=\frac{1}{2*\mathit{Zyl}}{\displaystyle \underset{k=0}{\overset{2*\mathit{Zyl}-1}{\Sigma }}}{\mathit{dn}}_k$$

$${{\mathit{dn}}^{*}}_k={\mathit{dn}}_k-{\widetilde{\mathit{dn}}}_{\mathit{k}}$$

A cylinder assignment is carried out either according to the in the DE 102006018958 described method or by displacement and subsampling. Dropouts are detected by means of a threshold underflow.

Shift by a certain number of segments s with fractional part t: $${\mathit{dnseg}}_k=\left(1-t\right)*{\mathit{dn}}_{k-s}^{*}+t*{\mathit{dn}}_{k-s-1}^{*}$$

Downsampling:

Every second value of dnseg k for even k is stored in a matrix dak (m, n), for a certain number of past cycles m and for the number of cylinders n (k = 0 ... 2zyl-1).

On the basis of the matrix dak, statistical analyzes already mentioned above are carried out individually for the cylinder and for the entire engine. In an advantageous extension, the application of an upstream correction of tooth pitch errors and / or a low-pass filtering of the speed signal in order to avoid aliasing effects is provided. This increases the signal quality and thus the quality of the statistics.

Furthermore, stochastic fluctuations can be detected based on regulations that perform a cylinder equalization. The presence of stochastic torque fluctuations is recognized by the fact that the control deviation can not be controlled to 0, but permanent cylinder-specific fluctuations in the control deviation are present.

A statistical analysis of the control deviations of all cylinders is a measure of stochastic torque fluctuations. Stochastic dropouts can not be assigned to each cylinder individually, but for large fluctuation values in the statistics.

According to the invention, statistical analyzes are possible where torque-proportional features are formed.

Claims (8)

1. Method for controlling an internal combustion engine, in which control variables for at least one actuator (100) are predefined on the basis of operational characteristic variables, wherein a variable is determined which characterizes unsmooth running, wherein the fuel properties are inferred on the basis of the variable which characterizes the unsmooth running, characterized in that poor fuel quality is detected if the intensity of stochastic torque fluctuations is greater than a threshold value.
2. Method according to Claim 1, characterized in that a fuel quality characteristic number is determined on the basis of the intensity of the stochastic torque fluctuations.
3. Method according to Claim 1 or 2, characterized in that at least one control variable is corrected if poor fuel quality is detected.
4. Method according to one of the preceding claims, characterized in that if the intensity of stochastic torque fluctuations is greater than the threshold value, the fuel quality characteristic number of further variables are stored permanently.
5. Method according to one of the preceding claims, characterized in that the correction is at least partially cancelled if certain states are present.
6. Method according to Claim 5, characterized in that the certain states are present if the internal combustion engine is started and/or a refuelling process has been detected.
7. Method according to Claim 5, characterized in that the correction is at least partially cancelled at certain intervals.
8. Device for controlling an internal combustion engine which predefines control variables for at least one actuator on the basis of operational characteristic variables and which determines a variable which characterizes unsmooth running, characterized in that means are provided which infer the fuel properties on the basis of the variable which characterizes the unsmooth running, and detect poor fuel quality if the intensity of stochastic torque fluctuations is greater than a threshold value.

Images