EP1698775B1 - System and method to control the combustion behaviour of an internal combustion engine - Google Patents

Description

The invention relates to a method for characterizing the combustion behavior of an internal combustion engine and to a control system and a method for feedback control of the combustion of an internal combustion engine, which are based on the former method

A method for characterizing the combustion behavior of an internal combustion engine having the features of the preamble of claim 1 is known from DE 103 51 133 A1 known.

Conventional control systems for fuel injection in internal combustion engines such. B. Diesel engines typically operate in an open loop. In other words, the injection time and the pulse width of the injection are taken from permanently predetermined maps or table memories which are stored in the engine controller. While such systems have very fast control performance, they are not very robust to motor tolerances because the control strategy can not be adjusted in the event of disturbances. For example, if the flow characteristic of an injector in a diesel engine changes due to wear, the pulse width of the injector used will no longer provide the engine with the required amount of fuel. The consequence of this may be higher values for engine emissions, fuel consumption and noise, or even engine damage. For these reasons, it is desirable to provide feedback in the context of combustion control in an internal combustion engine.

However, such a feedback control of an internal combustion engine requires the availability of a feedback signal which can characterize the combustion behavior. In this regard, measurements of in-cylinder pressure have been proposed and variously studied. A disadvantage However, such measurements is that they require one sensor per cylinder and therefore are relatively expensive. In addition, currently available in-cylinder pressure sensors exhibit drift behavior and a relatively short life. Last but not least, the direct access to the combustion chambers required by currently available sensors poses a problem.

The combustion information obtained from the feedback signal, essentially the combustion position and intensity, can be used to correct injection and cylinder charge composition control variables. In this way, for. B. life drift of sensors and actuators such as air mass sensor and fuel injector can be compensated.

Against this background, it was an object of the present invention to provide means for a simple yet robust control of the combustion behavior of an internal combustion engine.

This object is achieved by a method having the features of claim 1 or 7 and by a control system having the features of claim 8.

Advantageous embodiments are contained in the subclaims.

Knock sensor signals are routinely detected in many engines to detect premature auto-ignition ("knocking") of the engine and, if necessary, to cause appropriate countermeasures thereto. Various types of knock sensor signals are known in the art and are suitable for the present method. In particular, the knock sensor signal may detect mechanical vibrations of the engine generated by the combustion in the cylinders. The "course" of the knock sensor signal can be described both as a function of time and in particular of the associated crankshaft angle of the internal combustion engine.

As studies have shown, an index X can be obtained with the method according to the invention, which characterizes the combustion behavior with surprising accuracy. It is important that the index X depends on a more or less long course of the knock sensor signal and not only on individual values of a sensor signal. The advantage of the method is, moreover, that the knock sensor signal is relatively simple and robust to determine or is already available in many motor vehicles anyway.

As already mentioned, the knock sensor signal can be detected in particular as a function of the crankshaft angle (determined in parallel). The crankshaft angle is directly related to the state of the engine or the position of the engine cylinder, so that the mutual assignment of crankshaft angle and knock sensor signal can describe the combustion behavior particularly meaningful.

The method described can in principle be carried out with time-continuous signals or analog signals. Typically, however, signal acquisition and processing is time discrete and digitized. In this case, the crankshaft angle is preferably determined with a resolution of less than 1 °, more preferably less than 0.5 °. The resolution of the crankshaft angle present in the raw data is determined primarily by the sampling rate and is typically significantly higher than the above-mentioned values. The required finer resolution of the crankshaft angle is then preferably obtained by interpolation or extrapolation from the existing measurement data.

According to a development of the method, the knock sensor signal is taken into account only in an interval of its definition range, which is characteristic of a selected cylinder of the internal combustion engine. In particular, the knock sensor signal may be considered only at a predetermined angular interval around top dead center between the compression and combustion strokes of the selected cylinder to provide combustion relevant information to detect this cylinder and hide interference from other events.

In the method according to the invention, the index X is calculated from a "signal energy" E (θ) dependent on the crankshaft angle θ of the knock sensor signal K (θ) (also dependent on the crankshaft angle θ or τ), where E (θ) is calculated according to the following formula is defined: $$e\left(\theta \right)=\underset{{\mathrm{<figure-callout\; id="0"\; label="\theta "\; filenames="imgf0001.png"\; state="\{\{state\}\}">\theta </figure-callout>}}_0}{\overset{\theta }{\int }}K(\mathit{\tau }{)}^2\mathit{d\tau }\mathrm{,}$$

Here, θ _{0 means} a predetermined (lower) integration limit. If the knock sensor signal K is considered only at one interval as described above, θ ₀ typically corresponds to the lower limit of this interval. Furthermore, it is understood that equation (1) is intended to include the corresponding discretized formulation in the case of discrete-time processing of the signals.

The above-defined signal energy E (θ) is preferably bandpass filtered and / or normalized before being further used to calculate the index X.

Based on the signal energy introduced above, a suitable index X of the combustion performance of the internal combustion engine for a given value p (0≤p≤100) is defined by the formula $$X=X_p=\min \left\{\theta |e\left(\theta \right)\ge \frac{\mathrm{<figure-callout\; id="100"\; label="p"\; filenames="imgf0002.png"\; state="\{\{state\}\}">p</figure-callout>}}{100}\underset{\mathit{\tau }}{\mathrm{Max}}\left(e\left(\mathit{\tau }\right)\right)\right\}\mathrm{,}$$

In this case, the index X = X _p thus corresponds to the crankshaft angle at which the signal energy E exceeds p per cent of its maximum value for the first time. The formed for p = 50 Index X ₅₀ represents it as so. "Energy focus on" a particularly meaningful size represents.

The invention further relates to a method for the feedback control of combustion in an internal combustion engine, in which a feedback signal is formed by an index X according to the method according to the invention. As has been explained, on the one hand, the index X is easy to win and on the other hand very meaningful in terms of combustion, so that it allows a simple and robust control of the operation of the internal combustion engine. In this case, control signals influenced by the method may in particular be the time or points in time, the number, the pulse width (s) of the fuel injections, the ignition timing, the valve opening and closing times, the exhaust gas recirculation, the position of the throttle valve or the like.

The invention further relates to a control system for an internal combustion engine, which contains an input for the signal of a knock sensor and is designed to carry out the method according to the invention. That is, the control system may calculate an index X based on the history of a knock sensor signal and use it as a feedback signal for feedback control of the combustion. The knock sensor is a structure-borne sound acceleration sensor, such. B. a piezo pressure sensor which is mounted on the engine block to record mechanical vibrations. Incidentally, the control system can be realized in a known manner, for example by a microprocessor with associated components such as memory and interfaces as well as with suitable software.

Fig. 1

schematisch das Zusammenwirken einer Brennkraftmaschine mit einem erfindungsgemäßen Regelungssystem;

Fig. 2

die Abhängigkeit dreier verschiedener charakteristischer Größen für das Verbrennungsverhalten einer Brennkraftmaschine bei variierendem Beginn der Injektor-Aktivierung (BOA) für 35% Abgasrückführung;

Fig. 3

den Verlauf der Wärmeabgabe in einem Zylinder bei der Versuchsserie von Figur 2;

Fig. 4

die Abhängigkeit dreier verschiedener charakteristischer Größen entsprechend Figur 2 bei einer Abgasrückführung von 20%, und

Fig. 5

den Verlauf der Wärmeabgabe in einem Zylinder bei der Versuchsserie von Figur 4.

In the following the invention will be explained in more detail by way of example with reference to the figures. Show it:

Fig. 1

schematically the interaction of an internal combustion engine with a control system according to the invention;

Fig. 2

the dependence of three different characteristic quantities for the combustion behavior of an internal combustion engine with varying Initiation of injector activation (BOA) for 35% exhaust gas recirculation;

Fig. 3

the course of heat release in a cylinder in the pilot series of FIG. 2 ;

Fig. 4

the dependence of three different characteristic quantities accordingly FIG. 2 at an exhaust gas recirculation of 20%, and

Fig. 5

the course of heat release in a cylinder in the pilot series of FIG. 4 ,

In FIG. 1 is schematically shown an internal combustion engine 10 with (at least) a cylinder 13 and a piston 12 movable up and down therein. When the internal combustion engine is z. As a diesel engine, but without the invention would be limited thereto. The piston 12 is connected in a known manner via a connecting rod with the crankshaft 11, wherein a crankshaft angle sensor 18 measures the crankshaft angle θ. The cylinder further includes an intake valve 14 and an exhaust valve 16 for fresh air and exhaust gases, respectively, and a fuel injector 15 for direct injection of fuel into the combustion chamber.

At the engine block, a knock sensor 17 is arranged, which may be formed for example as a pressure sensor with piezo pickups. By the knock sensor 17 vibrations of the engine block caused by the combustion are detected. Preferably, the signal of the knock sensor 17 is immediately low-pass filtered in order to avoid aliasing effects (cf. Ch. Vigild, A. Chevalier, E. Hendricks: "Avoiding signal aliasing in event-based engine control", SAE Paper No: 2000-01-0268 ).

The - possibly low-pass filtered - signal K of the knock sensor 17 and the crankshaft angle θ from the sensor 18 are sampled by a gain and filter module 20, amplified and filtered. The sampling of the signals can be done either in the time domain or in the crankshaft angle range. When scanning in the time domain is a fixed time interval, in the scan in the crankshaft angle range a fixed crankshaft angle between the sampling points. Of course, the sampling can also be carried out according to other schemes and the sampling rate, for example, vary (in the angular range or in the time domain). In the latter case, a high signal resolution can be achieved, in particular in certain signal areas of interest.

For the desired characterization of the combustion behavior by the knock sensor signal K, a correct synchronization of this signal K with the crankshaft angle θ is of great importance. As a rule, however, the crankshaft angle θ is detected by a toothed disk on the flywheel, in which - due to the tooth spacing - only angular resolutions of typically 3 °, 5 °, 6 ° or 10 ° are obtained. In contrast, in the present case higher resolutions up to 0.1 ° or less are needed. The crankshaft angle θ is therefore determined in the module 20 by interpolation or extrapolation with the required fineness from the raw data. An interpolation can be used if the crankshaft angle is not needed immediately and can therefore be calculated as an intermediate value of two consecutive sampling points. If, on the other hand, an immediate use of the crankshaft angle θ takes place, then it must be extrapolated from the preceding sampling points.

The amplified and filtered signals θ, K (θ) of the crankshaft angle and the knock sensor are forwarded to a combustion profile module 21 for estimating the combustion profile or for determining characteristic indices X for this purpose. The signals or indices calculated by the module 21 are used by the subsequent control module 22 as feedback signals for the feedback control of the internal combustion engine 10. In the following, a preferred method implemented in module 21 for calculating an index X is explained in more detail:

In this method, first of all, a region of interest is extracted from the knock sensor signal K (θ) provided by the module 20 for a selected cylinder of the internal combustion engine, which can be achieved, for example, by limiting the Signal on a (angle) window J = _start ; θ _end ] can be done according to the formula: $$K_J\left(\mathrm{\theta }\right)=\{\begin{array}{ccc}K\left(\mathrm{\theta }\right)& .& \mathrm{\theta }\in J=\left[{\mathrm{\theta }}_{\mathit{begin}};{\mathrm{\theta }}_{\mathit{<figure-callout\; id="0"\; label="end"\; filenames="imgf0001.png"\; state="\{\{state\}\}">end</figure-callout>}}\right]\\ 0& .& \mathit{otherwise}\end{array}$$

Next, the included signal energy E (θ) is calculated therefrom according to the formula: $$e\left(\mathrm{\theta }\right)=\underset{{\mathrm{<figure-callout\; id="0"\; label="\theta "\; filenames="imgf0001.png"\; state="\{\{state\}\}">\theta </figure-callout>}}_0}{\overset{\mathrm{\theta }}{\int }}{K}_J(\mathrm{\tau }{)}^{2}d\mathrm{\tau }\approx e_{N\left(\mathrm{\theta }\right)}={\displaystyle \underset{n=N\left(\mathrm{\theta }\mathrm{0}\right)}{\overset{N\left(\mathrm{\theta }\right)}{\Sigma }}}{K}_{J.n}^2{\mathrm{\Delta \theta }}_n$$

Here, θ _{0 is} a predetermined starting angle of the integration, which is typically equal to the lower interval limit: θ ₀ = θ _start . In the discretized form of this formula, the variable Δθ _n defines the sampling interval between the crankshaft angle samples number (n-1) and n, N (θ ₀ ) and N (θ), respectively, are the numbers of sampling for the crankshaft angle θ ₀ or θ, and K _{J, n} is the knock sensor signal K of the n-th sample. In the simplest case of constant sampling distances, Δθ _n = 1. The discretized form of equation (4) is used as a basis for further consideration, although all considerations apply analogously to the continuous version.

The energy signal E _{N (θ)} from equation (4) is then advantageously digitally bandpass filtered and normalized according to the formula $$e_{N\left(\mathrm{\theta }\right)}^F=\frac{F_{\mathit{BP}}\left(e_{N\left(\mathrm{\theta }\right)}\right)}{\underset{\mathrm{\tau }\in J}{\mathrm{Max}}\left[F_{\mathit{BP}}\left(e_{N\left(\mathrm{\tau }\right)}\right)\right]}$$

Herein, the function F _{BP represents} the band-pass filtering, which may be either the forward type or the forward / backward type. Forward type filters filter a signal only in the forward direction, that is, the angle θ grows at one such filter incrementally. For this reason, forward filters require less computation and can be used for online calculations, for example, for calculations of current events. Due to the nature of these filters, however, these lead to a phase shift of the input signal. On the other hand, forward / reverse type filters filter a signal in both the forward and reverse directions so that these phase shifts can be compensated. However, they usually require more computational effort than corresponding forward filters and can only be used offline, eg. In calculations between combustion events.

The maximum value in the denominator of equation (5) is formed over the entire considered crankshaft angle interval J. Hereinafter, for simplicity, the superscript F indicating the bandpass filtering and normalization will be omitted, and further only the energy symbol E will be used, but it will be understood that this is (also) to designate a bandpass filtered and normalized signal.

wobei N(θ_end) - N(θ_start) die Gesamtzahl der Abtastpunkte ist.From the energy signal values in the sampled angular interval J, a signal energy vector can be defined in accordance with $$\overrightarrow{e}=\left(e_{N\left(\mathrm{\theta }\mathit{begin}\right)}.e_{N\left(\mathrm{\theta }\mathit{begin}\right)+1}....e_{N\left(\mathrm{\theta }\mathit{end}\right)}\right).$$

where N (θ _end ) - N (θ _start ) is the total number of sampling points.

Based on the variables explained above, indices are now defined which characterize the combustion behavior of the engine. According to their definition, these indices are also referred to as "energy focus" indexes. An important feature of the indices is that they focus on the distribution of the signal energy in the given signal window J, rather than on individual signal values or points, such as abrupt changes in signal energy (which would be intuitively close to estimating the maximum pressure gradient in the cylinder). , Another benefit of the energy-balance indexes This is because they rely on signal integration and are therefore less susceptible to noise problems.

The general formula for the definition of the energy center indices X _p can be written as follows (this corresponds to the discretized formulation of equation (2)): $$X=X_p=\min \left\{\theta |e_{N\left(\theta \right)}\ge \frac{\mathrm{<figure-callout\; id="100"\; label="p"\; filenames="imgf0002.png"\; state="\{\{state\}\}">p</figure-callout>}}{100}\underset{\mathit{\tau }\in J}{\mathrm{Max}}\left(e\left(\mathit{\tau }\right)\right)\right\}\mathrm{,}$$

X₁₀,

bei welchem 10% der gesamten Signalenergie erreicht werden;

X₅₀,

bei welchem 50% der gesamten Signalenergie erreicht werden und welcher daher auch als "Schwerpunkt" der Signalenergie bezeichnet wird;

X₉₀,

bei welchem 90% der gesamten Signalenergie erreicht werden; und

E_max =

max(E_N(θ)), d. h. der Wert der gesamten Signalenergie.

For a given percentage p with 0 ≦ p ≦ 100, the index X _p thus corresponds to the minimum crankshaft angle at which p percent of the total signal energy is reached. Of particular importance in this regard are the four indices

X ₁₀ ,

at which 10% of the total signal energy is achieved;

X ₅₀ ,

at which 50% of the total signal energy is reached and which is therefore also called the "center of gravity" of the signal energy;

X ₉₀ ,

where 90% of the total signal energy is reached; and

E _max =

max (E _{N (θ)} ), ie the value of the total signal energy.

If M ≥ 1 knock sensors are used, the indices explained above can be replaced by the individual indices X _p ⁽ⁱ⁾ weighted with the respective total signal energy E _max ⁽ⁱ⁾ according to the following formula $${\tilde{X}}_p=\frac{{\displaystyle \underset{i=1}{\overset{M}{\Sigma }}}{e}_{\mathrm{Max}}^{\left(i\right)}{X}_p^{\left(i\right)}}{{\displaystyle \underset{i=1}{\overset{M}{\Sigma }}}{e}_{\mathrm{Max}}^{\left(i\right)}}$$

In summary, in module 21 of FIG. 1 from the amplified and filtered knock sensor signals K (θ) calculates the combustion characteristic indices X _p , the latter implicitly describing the profile of the diesel combustion or the profile of the heat output in the combustion chambers. The indices X _p may then be used in the control module 22 to affect fuel injection via injection pressure, injection pulse width, and / or injection time (s), exhaust gas recirculation, boost pressure, and / or another suitable amount.

In the FIGS. 2 to 5 experimental results for the application of the method explained above are shown. These results were obtained for cylinder # 2 of a 2.7L V6 diesel engine at various load conditions between approximately 0.5 and 6 bar indicated mean pressure (IMEP) and at engine speeds between 1500 and 3400 rpm. The crankshaft angle _interval J used in equation (3) was selected to be 80 ° wide symmetrically about top dead center (TDC), that is, the limits θ _start = -40 °, θ _end = + 40 °.

In a first series of experiments, an exhaust gas recirculation rate (EGR) of approximately 35% was set, while the beginning of activation of the fuel injectors (BOA: Begin Of Activation) was adjusted between 6 ° and 18 ° before top dead center (BTDC). The data of the experiments are listed in detail in the following table: BOA = 6 ° 8 ° 10 ° 12 ° 14 ° 16 ° 18 ° Of test. 73 72 69 62 65 68 75 Engine speed [rpm] 1995 1995 1995 1995 1994 1994 1994 Inlet pressure [bar] 1210 1211 1215 1226 1220 1209 1207 IMEP [cash] 5.96 6:01 6:03 5.98 6.17 6.18 6.23 Standard. Air / fuel ratio 0559 0564 0597 0665 0607 0605 0600 AGR rate [%] 29.5 30.6 34.0 41.8 37.1 34.7 33.8 Rail pressure [bar] 1008 1008 1008 1002 1010 1015 1007 Injector pulse width [μs] 470 470 470 470 470 470 470 Start of combustion [° BTDC] -5 -3 -1 1 3 4 6 Standard. Combustion time (20% to 80%) [%] 15 17 17 19 20 21 22 AVL noise [dB] 94.9 95.4 95.3 93.8 96.0 96.8 97.5 Standard. Energy focus [%] 82 80 78 77 72 69 68 Standard. Energy width [%] 23 25 25 27 30 32 33 Standard. Signal intensity [%] 68 73 74 70 83 88 91

FIG. 2 shows in this regard the functional relationship between the knock sensor signal based combustion characterization and selected combustion parameters. The left diagram shows the behavior of the center of gravity energy index X ₅₀ with increasing BOA. The normalized energy width or "knock energy duration", X ₉₀ - X ₁₀ , is plotted against the normalized main burn duration (ie the time to get from 20% to 80% of the total energy release within the data window) in the center graph. The right diagram finally shows the relationship between the value of a sound pressure level meter (in dB) and the average normalized knock signal energy (in dB) accumulated over the data window.

FIG. 3 shows the curves associated with the experiments described above, the heat release in the cylinder for each set values of the BOA.

In a second series of experiments, an inlet pressure of about 1.0 bar and an EGR rate of about 20% was set, deviating from the first series. The detailed data of the test series are shown in the following table: BOA = 6 ° 8 ° 10 ° 12 ° 14 ° 16 ° 18 ° Of test. 74 71 70 63 66 67 76 Engine speed [rpm] 1995 1995 1994 1995 1995 1995 1995 Inlet pressure [bar] 1035 1033 1031 1033 1031 1030 1029 IMEP [cash] 5.89 5.95 6:00 am 6:09 6.13 4:52 6.11 Standard. Air / fuel ratio 0573 0568 0575 0581 0582 0580 0591 AGR rate [%] 20.6 20.6 20.7 20.6 20.8 20.8 20.9 Rail pressure [bar] 1007 1008 1008 1008 1010 1014 1018 Injector pulse width [μs] 470 470 470 470 470 470 470 Start of combustion [° BTDC] -6 -3 -1 0 2 4 6 Standard. Combustion time (20% to 80%) [%] 13 15 17 19 20 18 20 AVL noise [dB] 95.8 96.7 97 97.5 98.0 98.5 98.9 Standard. Energy focus [%] 84 82 79 76 74 73 69 Standard. Energy width [%] 20 23 25 28 32 34 36 Standard. Signal intensity [%] 72 81 86 91 95 98 97

The FIGS. 4 and 5 show the results of the second series of experiments in an analogue representation like the Figures 2 and 3 ,

Claims (8)

1. Method for characterizing the combustion behavior of an internal combustion engine (10), wherein an index X characterizing the combustion behavior is calculated on the basis of the profile of a knocking sensor signal K characterized in that the index X is calculated from the signal energy E(θ), dependent on the crankshaft angle θ, of the knocking sensor signal K(θ), wherein $$E\left(\mathrm{\theta }\right)=\underset{{\mathrm{\theta }}_0}{\overset{\mathrm{\theta }}{\int }}K(\mathit{\tau }{)}^2\mathit{d}\mathrm{\tau },$$

   

   and the index X is defined for a given value 0 ≤ p ≤ 100 by $$X=X_p=\min \left\{\theta |E\left(\theta \right)\ge \frac{p}{100}\underset{\mathrm{\tau }}{\max }\left(E\left(\mathit{\tau }\right)\right)\right\}\mathrm{.}$$

   
2. Method according to Claim 1, characterized in that the knocking sensor signal K characterizes mechanical vibrations of the internal combustion engine (10).
3. Method according to Claim 1 or 2, characterized in that the knocking sensor signal K is sensed as a function of the crankshaft angle θ.
4. Method according to Claim 3, characterized in that the crankshaft angle θ is determined with a resolution of less than 1°, preferably less than 0.5°.
5. Method according to at least one of Claims 1 to 4, characterized in that the knocking sensor signal K is taken into account only in an interval which is characteristic for a selected cylinder of the internal combustion engine (10).
6. Method according to at least one of Claims 1 to 5, characterized in that the signal energy E(θ) is bandpass-filtered and/or normalized for the calculation of the index X.
7. Method for feedback control of the combustion in an internal combustion engine (10), characterized in that a feedback signal is formed by an index X according to a method according to at least one of Claims 1 to 6.
8. Control system for an internal combustion engine (10), containing an input for the signal of a knocking sensor (17), characterized in that the latter is designed to carry out a method according to at least one of Claims 1 to 6.

Images