Method and apparatus for calibrating a fuel metering system of an internal combustion engine, in particular of a motor vehicle

By varying activation durations and using regression analysis to create a control duration map, the method addresses calibration inaccuracies in fuel metering systems, enhancing precision and reducing learning time while improving noise reduction and emissions.

DE102008002482B4Active Publication Date: 2024-11-28ROBERT BOSCH GMBH
View PDF 5 Cites 0 Cited by

Patent Information

Application Number
DE102008002482
Authority / Receiving Office
DE · DE
Patent Type
Patents
Current Assignee / Owner
Filing Date
2008-06-17
Publication Date
2024-11-28
Estimated Expiration
2028-06-17

AI Technical Summary

Technical Problem

Existing fuel metering systems in internal combustion engines face challenges in accurately calibrating partial injections due to manufacturing tolerances and drift over the engine's lifespan, leading to increased emissions and combustion noise, with current learning methods being time-consuming and prone to extrapolation errors.

Method used

A method that varies the activation duration of injectors to cover a wide range of test injections, using regression analysis to create a control duration map that accurately correlates injection quantity with activation duration, eliminating the need for extrapolation and reducing learning time.

Benefits of technology

This approach enhances calibration precision, reduces learning cycles, and minimizes extrapolation errors, resulting in improved noise reduction and faster adaptation to injector drift, especially in engines with multiple cylinders.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 00000000_0000_ABST
    Figure 00000000_0000_ABST
Patent Text Reader

Abstract

Method for calibrating a fuel metering system of an internal combustion engine (10), in particular of a motor vehicle, in which at least one injector is actuated with a first test injection with a first actuation duration and a first quantity signal resulting therefrom is recorded, wherein a minimum actuation duration is determined, characterized in that the at least one injector is actuated with at least one second test injection with a second actuation duration differing from the first actuation duration and a second quantity signal resulting therefrom is recorded, and wherein a regression calculation is carried out on the basis of the first actuation duration and the second actuation duration as well as the first quantity signal and the second quantity signal.
Need to check novelty before this filing date? Find Prior Art

Description

State of the art

[0001] The invention relates to a method and a device for calibrating a fuel metering system of an internal combustion engine, in particular of a motor vehicle, according to the preambles of the respective independent claims.

[0002] In modern fuel injection systems of the type in question here, for example, in common-rail diesel systems, partial injections with relatively small fuel quantities are implemented before or after the actual main injections to improve mixture preparation. These main injections are usually calculated based on a torque request from the driver. The injection quantities of these partial injections must be as small as possible to avoid emissions disadvantages. At the same time, the injection quantities must be large enough to ensure that the minimum amount required for the combustion process is always delivered, even taking all tolerance sources into account. Such improved mixture preparation enables reduced exhaust emissions and reduced combustion noise.

[0003] The small fuel quantities involved in these partial injections require precise metering of the respective injection quantity. If a partial injection is even completely omitted, for example, because an existing injection component—an injector in common rail injection systems—is not yet injecting due to typical tolerances in an underlying control signal, this has a significant impact on the operation of the internal combustion engine, manifesting itself, for example, in increased noise during combustion.

[0004] In the aforementioned common rail injection systems, pressure generation and injection are decoupled from one another by means of a high-pressure accumulator ("rail"). The injection pressure is generated independently of the engine speed and the injected fuel quantity and is available in the high-pressure accumulator for injection. The injection timing and the injected fuel quantity are calculated in an electronic engine control unit and implemented by the injectors of each cylinder of the internal combustion engine via remotely controlled valves. It must be ensured that the aforementioned partial injections are implemented with the highest possible precision over the service life of the components of the fuel metering system and under all operating conditions occurring during operation of the internal combustion engine, such as the rail pressure prevailing in the high-pressure line (rail) of the common rail injection system or the injector temperature.

[0005] The metering of these minute quantities is based on a zero-quantity calibration, which is described, for example, in DE 199 45 618 A1. In this process, a single injection valve (injector) is activated during the engine's overrun, and the activation duration is gradually increased until a signal change occurs at a "minimum activation duration," for example, a measurable increase in torque in the engine. The activation duration then prevailing corresponds to an operating state in which injection for the respective combustion chamber (cylinder) of the engine is just beginning. This procedure is performed cyclically on all injection valves and cylinders of the engine.The control data obtained in this way are preferably stored in a characteristic map which is used for a subsequent control of the injectors as part of a zero-quantity calibration, whereby the current value of the control duration is converted into a correction value for the fuel quantity.

[0006] Furthermore, EP 1 388 661 A2 discloses a method for zero-quantity calibration in which a learning value is calculated based on test injections. However, this learning method does not take into account all quantity signals from the test injections; at least two activation durations are tested in order to cross a signal threshold and thus achieve a learning result. However, even in a steady state, approximately three to four iterations per learning result are necessary to achieve usable learning, with the resulting learning value additionally having to be filtered to achieve usable results. The learning process requires driving up to approximately 3,000 km, and therefore zero-quantity compensation is not available at the beginning of the vehicle's service life.

[0007] DE 10 2006 009 920 A1 describes a method for determining cylinder-specific correction values ​​for the injection quantity of an internal combustion engine. For this purpose, a characteristic curve is determined that indicates the relationship between the activation duration of the injection valve and the injected fuel quantity. To determine the correction value, the gradient and the Y-axis intercept of the determined characteristic curve are determined. DE 60 2004 003 390 T2 also describes a similar method.

[0008] EP 0 947 686 A2 discloses a zero-quantity calibration with an interpolation of small map values.

[0009] The learning of the injection quantity error or the zero quantity correction is carried out in the state of the art at a defined injection quantity, typically at a quantity value of approximately 1.2 mm 3 / H. However, since the requested pre-injection quantities are variable during driving, typically in a range of approximately 0.5 mm 3 / H - 5.0 mm 3 / H, the learned value determined at the calibrated injection quantity is transferred through extrapolation. However, this extrapolation reduces the accuracy of the correction.

[0010] Manufacturing tolerances that occur during the manufacture of the aforementioned injectors or other components of the respective fuel metering system result in differences in the operating parameters of the individual injectors of a fuel metering system. These differences only become apparent over the service life of the injectors or the fuel metering system or are even intensified during this service life, with the drift behavior of an injector representing a significant source of tolerance for the quantity accuracy of the partial injections. In addition, the injectors of a fuel metering system usually have different quantity maps, i.e. different dependencies between the injection quantity, the rail pressure and the actuation duration. This means that the different injectors fill the combustion chamber with different quantities of fuel despite precise actuation. Summary of the invention

[0011] The present invention is based on the idea of ​​improving the learning method described above for zero-quantity calibration so that the time required for learning a calibration value is reduced.

[0012] According to the method according to the invention, the actuation duration of an injector is varied such that the typical quantity range for test injections mentioned above is traversed as closely as possible. For each actuation duration value, the resulting injection quantity, which is determined according to the zero-quantity calibration mentioned above, is stored in a actuation duration map. The relationship between the value pairs of the actuation duration resulting from the zero-quantity calibration and the respective injection quantity is evaluated using regression calculations, preferably using linear regression, but alternatively also using a non-linear regression approach.

[0013] It should be emphasized that the term “injector” is understood here to include all types of injection valves or injectors, such as the injectors used in direct-injection gasoline engines, and is not limited to the injectors known in diesel engines.

[0014] The proposed calibration procedure makes it possible to adapt the aforementioned control duration map for the low-flow range over the lifetime of injectors or an underlying fuel metering system. The aforementioned control duration map plays a central role in metering accuracy.

[0015] The method according to the invention takes into account the quantity signals of all calibration sequences, whereby the aforementioned regression approach already achieves very high precision and good noise suppression. The filtering of the learning value described above can therefore be completely eliminated.

[0016] The signal evaluation according to the invention results in a significant acceleration of the zero-quantity calibration and the learning process underlying it. Furthermore, the method and device according to the invention enable the learning of a larger quantity range than just a single calibration quantity. This reduces possible extrapolation errors and thus increases the correction accuracy of the zero-quantity calibration.

[0017] As a result, the learning behavior during zero-quantity calibration is significantly improved, and the learning cycles are correspondingly shortened. With just a few calibration sequences with variable control durations, a straight line can be adjusted to the measurement points, for example, which further shortens the learning time.

[0018] A shortened learning phase during zero-quantity calibration has the following additional advantages: a) Previous injector / nozzle generations were characterized by the fact that drift was a slow phenomenon, which the zero-flow calibration could easily follow, despite the considerable learning time. In cases where significant short-term drift occurs, shortening the time until the correction is activated or accelerating the learning time also enables accurate calibration at this early stage of the injector's life. b) The zero-quantity calibration is carried out individually for each cylinder, whereby doubling the number of cylinders, for example from 4 to 8 cylinders, doubles the learning time of the zero-quantity calibration, which can be considerably reduced by the present invention. c) Transmission applications intended for the US market in particular have relatively few overrun phases. Due to the large number of cylinders typical in the US, zero-quantity calibration tends to learn even more slowly in US applications, which is why the shortened learning process according to the invention is particularly advantageous in this application area. d) By covering a control duration range or injection quantity range, the need for extrapolation is completely eliminated, which reduces calibration effort. Furthermore, eliminating the additional extrapolation tolerance offers the possibility of reducing the tolerance margin in the setpoint specification for the pilot injection quantity, which has a positive effect on emissions while simultaneously achieving very low combustion noise. e) The proposed method allows for improved noise suppression. According to the prior art, the learning values ​​are smoothed with a filter, which requires a number of learning results. The method proposed by the invention converges to a final value with just a few points and changes only slightly with increasing numbers of measured values ​​while improving noise suppression. drawing

[0019] The invention is described in more detail below using preferred embodiments, from which further features and advantages emerge.

[0020] In detail: Fig. 1 is a schematic representation of a fuel metering system of an internal combustion engine known in the prior art; Fig. 2 a detailed representation of the calculation of the control durations of an electrically operated valve according to the state of the art; Fig. 3a shows a typical control duration map with associated operating ranges according to the prior art and the present invention in comparison; Fig. 3b two variants of the regression approach according to the invention; Fig. 4 is an overview block diagram of a device according to the invention; and Fig. 5 a flowchart of an embodiment of the calibration method according to the invention. Description of implementation examples

[0021] The Fig. Figure 1 shows a block diagram of the essential elements of a fuel metering system of an internal combustion engine, preferably a compression-ignition or direct-injection motor vehicle engine. The internal combustion engine 10 receives a specific amount of fuel metered at a specific time by a fuel metering unit 30. Various sensors 40 record measured values ​​15 that characterize the operating state of the internal combustion engine and transmit them to a control unit 20. Various output signals 25 from further sensors 45 are also transmitted to the control unit 20. These sensors record operating variables that characterize the state of the fuel metering unit and / or environmental conditions. One such variable is, for example, the driver's command. Based on the measured values ​​15 and the further variables 25, the control unit 20 calculates control pulses 35 that are applied to the fuel metering unit 30.The fuel metering unit 30 can be designed in various ways. For example, a distributor pump can be used as the fuel metering unit, in which a solenoid valve determines the timing and / or duration of fuel injection.

[0022] Furthermore, the fuel metering unit can be designed as a common rail system. In this system, a high-pressure pump compresses fuel in a reservoir. From this reservoir, the fuel then flows into the combustion chambers of the internal combustion engine via injectors. The duration and / or start of fuel injection is controlled by the injectors. The injectors preferably contain a solenoid valve or a piezoelectric actuator.

[0023] The control unit 20 calculates the amount of fuel to be injected into the internal combustion engine in a known manner. This calculation is performed depending on various measured values ​​15, such as the engine speed n, the engine temperature, the actual start of injection, and possibly other variables 25 that characterize the operating state of the vehicle. These additional variables include, for example, the position of the accelerator pedal or the pressure and temperature of the ambient air. Furthermore, it can be provided that a torque request is specified by other control units, such as the transmission control.

[0024] The control unit 20 then converts the desired fuel quantity into control pulses. These control pulses are then applied to the quantity-determining element of the fuel metering unit. The electrically actuated valve serves as the quantity-determining element. This electrically actuated valve is arranged such that the amount of fuel to be injected is determined by the opening or closing duration of the valve.

[0025] Often, a small amount of fuel is metered into the cylinder shortly before the actual injection. This can significantly improve the noise level of the internal combustion engine. This injection is called the pre-injection, and the actual injection is called the main injection. Furthermore, a small amount of fuel can be metered after the main injection. This is then called a post-injection. Furthermore, the individual injections can be divided into further partial injections.

[0026] The problem with such fuel metering systems is that the electrically operated valves can meter different amounts of fuel with the same control signal. In particular, the control duration at which fuel is metered depends on various factors. This minimum control duration is also referred to as the minimum control duration AD0. This minimum control duration results in an injection; control durations shorter than the minimum control duration do not result in an injection. This minimum control duration depends on various factors, such as temperature, fuel type, service life, rail pressure, manufacturing tolerances of the injectors, and other influences. In order to achieve precise fuel metering, this minimum control duration must be known.

[0027] A device for controlling the fuel metering in an internal combustion engine is described in Fig. 2. Already in Fig. The elements described in Figure 1 are designated by corresponding reference numerals. The signals 25 from sensors 45 and other sensors not shown are fed to a quantity specification 110. This quantity specification 110 calculates a fuel quantity QKW that corresponds to the driver's request. This quantity signal QKW is fed to a node 115, at whose second input the output signal QKM of a second synchronization unit 155 is present. The output signal of the first node 115 is fed to a second node 130, which in turn feeds a control duration calculation 140. The signal QKO of the zero-quantity correction 145 is present at the second input of the second node. At the two nodes 115 and 130, the quantity signals are preferably additively combined. Based on the output signal of the node 130, the control duration calculation 140 calculates the control signal for feeding the fuel metering unit 30.The control duration calculation calculates the control duration with which the electrically operated valves are actuated.

[0028] Various markings are arranged on a sensor wheel 120, which are scanned by a sensor 125. In the illustrated embodiment, the sensor wheel is a so-called segment wheel, which has a number of markings corresponding to the number of cylinders—four in the illustrated embodiment. This sensor wheel is preferably arranged on the crankshaft. This means that a number of pulses corresponding to twice the number of cylinders is generated per engine revolution. The sensor 125 delivers a corresponding number of pulses to a first synchronization unit 150.

[0029] The first synchronization 150 controls a first controller 171, a second controller 172, a third controller 173, and a fourth controller 174. The number of controllers corresponds to the number of cylinders. The output signals of the four controllers then reach the second synchronization 155. Furthermore, the output signals of the controllers reach the zero-quantity correction 142. Alternatively, the output signal of the second synchronization can also be fed to the zero-quantity correction 142. This alternative is shown with a dashed line.

[0030] The Fig. The device shown in Figure 2 operates as follows. Based on various signals, such as a signal indicating the driver's request, the quantity specification 110 determines the desired fuel quantity signal QKW, which is required to provide the torque desired by the driver. Due to tolerances, particularly in the fuel metering unit 30, deviations arise between the desired injection quantity and the actually injected fuel quantity. The individual cylinders of the internal combustion engine generally meter different fuel quantities with the same control signal.

[0031] These variations between the individual cylinders are usually compensated by a quantity compensation control (MAR). Such a quantity compensation control is shown schematically in the upper part of the Fig. 2. For quantity compensation control, each cylinder of the internal combustion engine is assigned a controller. The first controller 171 is assigned to the first cylinder, the second controller 172 to the second cylinder, the third controller 173 to the third cylinder, and the fourth controller 174 to the fourth cylinder. It can also be provided that only one controller is provided, which is alternately assigned to the individual cylinders. By means of the sensor 125 and the sensor wheel 120, the first synchronization 150 determines a setpoint value and an actual value for each individual controller. In this case, it is provided that a special filtering of the signal from sensor 125 takes place to compensate for tolerances of the sensor wheel and torsional vibrations. The output signals of the controllers 171 to 174 are fed to a second synchronization 155, which provides a correction quantity QKM with which the desired quantity QKW is corrected.

[0032] This fuel quantity compensation control is designed so that the controllers regulate the amount metered to the individual cylinders to a common average value. If a cylinder meters an increased fuel quantity due to tolerances, a negative fuel quantity QKM is added to the driver's desired fuel quantity QKW for this cylinder. If a cylinder meters too little fuel, a positive fuel quantity QKM is added to the driver's desired fuel quantity QKW. Such fuel quantity errors result in rotational irregularity. This has the effect that oscillations are superimposed on the speed signal whose frequency corresponds to the camshaft frequency and / or multiples of the camshaft frequency. These components in the speed signal with camshaft frequency characterize the rotational irregularity and are corrected to zero by the fuel quantity compensation control. Fuel quantity average value errors cannot be corrected with this fuel quantity compensation control.In particular, errors resulting from the fact that no fuel is metered below a minimum control duration cannot be corrected with such a quantity compensation control.

[0033] If the vehicle is in overrun mode, i.e. no injection is taking place, the internal combustion engine is by definition equal with regard to the fuel quantities injected into the individual cylinders. Therefore, there are no or only small components of the engine speed with camshaft frequency. If the control duration of the injector is slowly increased for a cylinder N, injection into cylinder N takes place above a minimum control duration AD0(N). This leads to combustion irregularities, which in turn result in engine speed irregularities. In particular, oscillations with multiples of the camshaft frequency occur in the engine speed signal. These camshaft frequency components are detected by the fuel quantity compensation control.

[0034] The controller corresponding to cylinder N determines a correction value. When the correction value of the quantity compensation control is present, the zero-quantity correction 142 detects the activation duration AD0(N) at which an injection quantity that is just distinguishable from the zero quantity is injected. The corresponding value AD0(N) is stored and used for subsequent metering to correct the activation duration of cylinder N. Fig. 2 this is shown by the fact that the value AD0(N) is used to form the correction value QKO.

[0035] In the Fig. 3a, the current and future available application and operating ranges for the control duration and the injection quantity are compared in a diagram of injection quantity in the unit [mm3 / H] versus control duration in the unit [µs]. The current application range 200 for pre-injections is significantly larger than the current operating range 205, with the possible operating range 210 significantly exceeding the current operating range 205. The Fig. The (measured) calibration or characteristic curve, also shown in Figure 3a, shows linear behavior only within a limited range of values ​​for both variables and behaves more parabolically over the entire range covered here. In addition, the typical signal threshold 215 of an injector is 1.00 mm. 3 / H is shown. When the working range is enlarged, for example, to the possible working range 210, this non-linear behavior becomes more apparent during zero-quantity calibration.

[0036] In the Fig. 3b shows two different regression approaches using measured parameters (as in Fig. 3a) in the time window between 250 and 520 µs activation duration. The first approach is a regression line 300 fitted to the measured data, whereas the second approach corresponds to a fitting of a parabola (e.g., a quadratic curve). As can be seen from the Fig. As can be seen in Figure 3b, the parabolic approach provides a significantly higher quality of fit in this range of values.

[0037] The Fig. 4 shows a simplified block diagram of a device according to the invention for controlling a fuel metering system. In the upper half of the diagram, separated by the dotted line, a control duration calculation unit 400 is shown, which, by means of interpolation based on the equation AD = f(target quantity, pRail), determines or sets a value AD for the control duration of the respective injector from a target quantity for the pre-injection and the currently prevailing rail pressure pRail. The input variables of this functional unit 400 thus represent the target quantity of the pre-injection 405, the current rail pressure 410, and the request for a pre-injection 415 itself. As an output variable, the functional unit 400 provides the value of the control duration under load 420 for the requested pre-injection.

[0038] In the lower part of the Fig. 4 shows additional functional units to be provided according to the present invention, which are used in particular during the learning phase described in more detail below. These are, on the one hand, a functional unit 425, by means of which the quick learning of the injector characteristic curve described here takes place, and a memory element 430, preferably an EEPROM, in which the respectively learned injector characteristic curve is stored in a non-volatile manner. The input variables for the quick learning unit 425 are the injection quantity (ZFC) 435 resulting from the zero-quantity calibration, the current rail pressure 440, and the control duration ZFC 445, which is variably specified according to the invention. The so-called 'Zero Fuel Quantity Calibration' (ZFC) is another term for the zero-quantity calibration for pre-injections described here.An additional input variable for the quick learning unit 425 is a status signal or status bit 450, which signals a push phase and by means of which the quick learning unit is started.

[0039] It should be noted that the method and device described herein can be applied not only to the aforementioned pre-injections, but also to any form of partial injections such as main injections or post-injections, since the described learning method is independent of the injection type as long as these injections are of the same order of magnitude in terms of quantity.

[0040] Based on the Fig. Figure 5 describes an embodiment of a routine for implementing the learning method according to the invention during zero-quantity calibration in a fuel metering system of a motor vehicle's internal combustion engine. It should be emphasized that the invention can, in principle, also be applied to other internal combustion engines with the advantages described herein.

[0041] After the start 500 of the routine, a loop is first checked 505 to determine whether the internal combustion engine is in overrun mode, because, as already mentioned, the inventive determination of the learning values ​​for the zero-quantity calibration takes place exclusively in overrun mode.

[0042] Test injections are then triggered on one cylinder at a time, the activation duration or resulting injection quantity of which is not fixed but is varied so that a typical quantity range of, for example, 0.5 mm 3 / H - 5.0 mm 3 / H is covered.

[0043] If an overrun operation of the internal combustion engine is detected in step 505, the value of the test injection quantity ME TE initially to a starting value, in the present embodiment to a minimum value ME TE _min, is determined (step 510). The following steps 520 - 550 are performed for each cylinder and each injector arranged on a cylinder.

[0044] In step 520, the value of the activation duration for the test injection AD(ME TE ) to a corresponding starting value AD(ME TE_ min). In step 525, a test injection with the specified activation duration AD(ME TE ). For each of the subsequent test injections, a quantity signal ME is calculated (step 530) and the activation duration AD, starting from the stated starting value AD(ME TE_min), if necessary even in both temporal directions, until a specified quantity signal range has been traversed.

[0045] The termination criterion for the variation or iteration mentioned is, in the present embodiment, the undershooting / exceeding of an empirically specified min / max threshold of the resulting quantity signal, in the present embodiment the exceeding of the max threshold ME TE_ max. Alternatively, the flow rate can be varied within a corresponding flow map between the outer limits of the map. It can also be specified that the flow rate range to be covered is determined by the permitted values ​​of the control duration.

[0046] If the quantity signal ME calculated in step 530 according to test step 535 has the specified max threshold ME TE_ max is not yet reached, the control duration of the then existing value pair (ME TE / AD TE) and in step 545 the current value of the control duration AD is set to the value AD neu = AD TE + ΔAD, whereby the magnitude of the change value ΔAD can be empirically specified. The system then returns to step 525 and performs another test injection, but with the aforementioned increased value of the control duration AD. neu .

[0047] For each tested activation duration AD, the resulting quantity signal ME is recorded in the manner described, and a regression calculation is performed in step 550 based on the then stored value pairs (ME / AD). In the present embodiment, the regression calculation is a linear regression approach, by means of which a correlation coefficient for all value pairs (ME / AD) is calculated in a manner known per se.

[0048] As already mentioned, according to the invention, the corresponding quantity signal is determined and stored for each activation period according to the zero-quantity calibration principle described above. The injection quantity according to the zero-quantity calibration principle can, as is known, be determined from the engine speed or an oxygen or ion current signal.

[0049] During driving, the adapted straight line (quantity = f(control duration)) can be evaluated inversely (i.e. control duration = f(target quantity), whereby the appropriate control duration for each individual injector can be determined for any quantity in the adapted range. The need for extrapolation is thus completely eliminated, but can be easily achieved using the straight line equation approach without additional computational or application effort.

[0050] As can be seen from the Fig. 3a and Fig. As can be seen in Figure 3b, currently available injectors exhibit, to a first approximation, linear behavior within the range of the specified injection quantities of pre-injections. Therefore, using methods known from statistical mathematics, a regression line can be calculated through the measured points, quantity signal = f(actuation duration). The respective measurement points "quantity signal as a function of the actuation duration" can be measured during one or more overrun phases of the internal combustion engine. A large number of measurement points is advantageous in order to achieve high accuracy / noise suppression. The correlation coefficient offers the possibility of monitoring the process for learning quality and detecting potential errors in the calibration process.

[0051] The following additional variants are possible. If the shape of the injector characteristic curve deviates significantly from a linear relationship, a non-linear approach can be chosen instead of a straight line approach. Fig.Figure 3b shows the straight-line and parabolic approaches. Approaches that can be easily linearized are particularly advantageous. Instead of learning a characteristic curve for each rail pressure stage, the rail pressure stages can be selected so closely that a map region can be learned. With map learning, it would also be possible to learn injectors with plateau formation.

[0052] In step 555, a check is now made to determine whether the currently available correlation coefficient exceeds a lower threshold value, Correlation_coeff_min, which indicates whether sufficient measurement points are available based on different activation durations. It should be noted that a sufficiently high correlation quality can be achieved due to the large value range for the input variables, especially since the activation duration is varied over a sufficiently large range. If this criterion 555 is not met, the memory is initialized (step 560), and a return to the initial step 500 is performed.

[0053] Only when condition 555 is fulfilled is the current routine aborted and in step 565 the then existing regression line is released or enabled for further operation of the internal combustion engine.

[0054] Using the described method, it is possible to adapt the mentioned control duration map for the small quantity range over the lifetime of an injector.

Claims

[1] Method for calibrating a fuel metering system of an internal combustion engine (10), in particular of a motor vehicle, in which at least one injector is activated with a first test injection with a first activation duration and a resulting first quantity signal is detected, wherein a minimum activation duration is determined, characterized by that the at least one injector is controlled with at least one second test injection with a second control duration that differs from the first control duration, and a resulting at least second quantity signal is detected, and wherein a regression calculation is carried out on the basis of the first control duration and the at least second control duration as well as the first quantity signal and the at least second quantity signal. [2] Method according to claim 1, characterized bythat the first and at least the second quantity signal are determined according to the zero quantity calibration method. [3] Method according to claim 1 or 2, characterized by that by means of the at least first test injection and the at least second test injection, an injection quantity range available for partial injections is covered as far as possible. [4] Method according to claim 3, characterized by that the injection quantity range mentioned is 0.5 mm 3 / H up to 5.0 mm 3 / H. [5] Method according to one of the preceding claims, characterized by that a linear regression is carried out on the basis of the first control duration and the at least second control duration as well as the first quantity signal and the at least second quantity signal. [6] Method according to one of the preceding claims, characterized bythat during operation of the internal combustion engine, a characteristic curve calculated by means of the said regression calculation is applied inversely, wherein an associated control duration for at least one injector is determined for injection quantities. [7] Device for calibrating a fuel metering system of an internal combustion engine (10), in particular of a motor vehicle, with control means for controlling at least one injector with a first test injection with a first control duration and for detecting a first quantity signal caused by the control, as well as with means for determining a minimum control duration, characterized bythat the said control means are designed such that the at least one injector is controlled with at least one second test injection with a second control duration that differs from the first control duration and a resulting at least second quantity signal is detected, and that computing means are provided which carry out a regression calculation on the basis of the first control duration and the at least second control duration as well as the first quantity signal and the at least second quantity signal. [8] Device according to claim 7, in particular for use in a common rail injection system, characterized byan actuation duration calculation unit for determining an actuation duration under load for an injector on the basis of an interpolation from a target quantity for a test injection and a currently existing rail pressure and by a learning unit for quickly learning an injector characteristic curve as well as by a data memory for storing the learned injector characteristic curve. [9] Device according to claim 8, characterized by that the input variables of the learning unit are the injection quantity resulting from a zero-quantity calibration and the control duration as well as a status signal which signals a coasting phase of the internal combustion engine.

Citation Information

Patent Citations

  • determination of cylinder-specific correction values ​​for the injection quantity of an internal combustion engine

    DE102006009920A1

  • Control method for fuel injection system in internal combustion engine by storing drive period at which change in signal occurs as minimum drive period

    DE19945618A1

  • PROCEDURE FOR REAL-TIME DETERMINATION OF A FUEL INJECTION FLOW CHARACTERISTIC

    DE602004003390T2

  • Fuel-injection system for engine

    EP0947686A2

  • Method and device for controlling the fuel quantity injected into an internal combustion engine, in particular a diesel engine equipped with a common rail injection system

    EP1388661A2