DE102021205589B4 - Method and device for operating a fuel injector using machine learning methods - Google Patents

Abstract

The invention relates to a method for operating an injection system (2) of an internal combustion engine (1) by determining an opening or closing time (A) of the injection valve (6) based on a sensor signal with the following steps: - providing (S1, S2) an evaluation point time series Scanning a sensor signal of a sensor (65) of the injector (6), - using (S3, S4) a data-based time determination model (20) with the evaluation point time series (E) to provide a modeled indication of an opening or closing time (A) and at least one System size (S) to obtain, wherein the time determination model (20) is trained to indicate an opening or closing time (A) and the modeled system size (S) depending on the evaluation point time series (E) provided, - determining (S5) a confidence value dependent of a detected system size (S) and using the data-based time determination model (20) depending on the Evaluation point time series (E) determined modeled system variable (S); - Operation (S6, S7) of the internal combustion engine (1) depending on the opening or closing time (A) and depending on the confidence value.

Description

technical field

The invention relates to a method for operating a fuel injection valve of an internal combustion engine using data-based models, in particular for determining an opening time and/or a closing time for estimating an injected fuel quantity.

State of the art

• - betriebspunktabhängiges Bereitstellen eines Sollwerts mindestens eines Verbrennungsmerkmals einer Verbrennung in dem Verbrennungsmotor anhand eines Sollwert-Kennfeldes, wobei das Verbrennungsmerkmal einer die Verbrennung im Verbrennungsmotor charakterisierenden Größe entspricht;
• - Bestimmen eines Wertes einer kennfeldbasierten Stellgröße zur Steuerung des Verbrennungsmotors aus einem Stellgrößen-Kennfeld,
• - Ermitteln eines Wertes einer modifizierten Stellgröße zur Steuerung des Verbrennungsmotors mit Hilfe eines datenbasierten Modells, wobei das datenbasierte Modell abhängig von einem realen Wert des Verbrennungsmerkmals der vorangegangenen Verbrennung, der durch Messen einer Größe während des Betriebs des Verbrennungsmotors ermittelt wird, und abhängig von der kennfeldbasierten Stellgröße den Wert der modifizierten Stellgröße zur Steuerung des Verbrennungsmotors ermittelt, wobei das datenbasierte Modell so ausgebildet ist, dass es abhängig von dem Sollwert des Verbrennungsmerkmals und dem realen Wert des Verbrennungsmerkmals adaptierbar ist;
• - Bereitstellen einer realen Stellgröße an den Verbrennungsmotor, um den Verbrennungsmotors anzusteuern, wobei die reale Stellgröße) auf einen Wert eingestellt wird, der von dem Wert der kennfeldbasierten Stellgröße und/oder dem Wert der modifizierten Stellgröße (SGmod) abhängig ist.

The DE 10 2008 001 081 A1 discloses a method for controlling an internal combustion engine, comprising the following steps:

• - operating point-dependent providing a target value of at least one combustion feature of a combustion in the internal combustion engine based on a target value characteristics map, wherein the combustion feature corresponds to a variable characterizing the combustion in the internal combustion engine;
• - determining a value of a map-based manipulated variable for controlling the internal combustion engine from a manipulated variable map,
• - Determining a value of a modified manipulated variable for controlling the internal combustion engine using a data-based model, the data-based model depending on a real value of the combustion feature of the previous combustion, which is determined by measuring a variable during operation of the internal combustion engine, and depending on the map-based manipulated variable determines the value of the modified manipulated variable for controlling the internal combustion engine, the data-based model being designed in such a way that it can be adapted as a function of the setpoint value of the combustion feature and the real value of the combustion feature;
• - Providing a real manipulated variable to the internal combustion engine in order to control the internal combustion engine, the real manipulated variable) being set to a value that depends on the value of the map-based manipulated variable and/or the value of the modified manipulated variable (SGmod).

• - Bestimmen eines zeitlichen Raildruckverlaufs in einem Hochdruckspeicher des Common-Rail-Einspritzsystems;
• - Bestimmen mindestens einer weiteren physikalischen Verlaufsgröße im Common-Rail-Einspritzsystem;
• - Bestimmen der Einspritzkraftstoffmenge in einer Einspritzsequenz, in der Kraftstoff durch ein Einspritzventil in einen Brennraum des Verbrennungsmotors eingespritzt wird, abhängig von Eingangsgrößen mithilfe eines trainierten Funktionsmodells, insbesondere eines nicht-parametrischen Funktionsmodells oder eines neuronalen Netzes, wobei die Eingangsgrößen für jede Einspritzsequenz mehrere Werte des zeitlichen Raildruckverlaufs und einen oder mehrere Werte von mindestens einer weiteren Verlaufsgröße umfassen;
• - Betreiben des Verbrennungsmotors abhängig von der Einspritzkraftstoffmenge.

The DE 10 2018 212 669 A1 apparently a method for operating an internal combustion engine with a common rail injection system depending on an injection fuel quantity, with the following steps:

• - Determining a rail pressure curve over time in a high-pressure accumulator of the common rail injection system;
• - Determining at least one further physical variable in the course of the common rail injection system;
• - Determination of the injection fuel quantity in an injection sequence in which fuel is injected through an injection valve into a combustion chamber of the internal combustion engine, depending on input variables using a trained functional model, in particular a non-parametric functional model or a neural network, the input variables for each injection sequence having multiple values of the rail pressure curve over time and one or more values of at least one further curve variable;
• - Operation of the internal combustion engine depending on the injection fuel quantity.

The DE 10 2017 217 733 A1 discloses a method and system for testing the output of a neural network. The method has the step of reading in a result and its confidence value, which is generated in a layer of a neural network, and the step of checking the plausibility of the result, taking into account its confidence value, in order to conclude that the object recognition of the neural network is incorrect or correct. on. The step of checking includes comparing the confidence value of the result to a predefined limit.

The DE 10 2020 205 688 A1 discloses a method for evaluating an operating variable of an injection valve, in particular a solenoid valve for an internal combustion engine, in particular a motor vehicle, having the following steps: providing a signal characterizing the operating variable, evaluating the signal using a, in particular artificial, neural convolution network.

The DE 10 2007 060 049 A1 discloses a method for determining the course of injection, in particular the end of injection of an injector for injecting fuel into the combustion chamber of an internal combustion engine, is characterized in that the course of injection, in particular the end of injection of the injector, is determined using a multi-layered artificial neural network, the input values of which determine the operating state of the Internal combustion engine characterizing sizes are.

The DE 10 2019 209 690 A1 discloses a method for controlling fuel metering in an internal combustion engine. A signal curve is detected by means of a sensor, which characterizes the time pressure curve of the fuel in an injector. On the pressure curve over time used a neural network that was trained using a learning process. At least one output of the neural network is used to control the internal combustion engine.

The DE 197 40 608 C2 discloses a method for determining at least one fuel-injection-related parameter for an internal combustion engine with a high-pressure accumulator fuel injection system individually for each engine combustion chamber and each injection process, characterized in that the pressure in the distributor pressure chamber of the high-pressure accumulator fuel injection system that is commonly assigned to the engine combustion chambers over a respective fuel injection process for a respective combustion chamber is recorded with high resolution in its course by means of a pressure sensor of the distributor pressure chamber and an associated pressure course pattern is obtained from the recorded pressure course and the at least one fuel-injection-related parameter is determined individually for each combustion chamber and each injection process from the pressure course pattern obtained by means of pattern recognition.

Technical background

Electromechanical or piezoelectric injectors are used to meter fuel in internal combustion engines. These enable a direct and precisely metered supply of fuel to the cylinders of the combustion engine.

One challenge is to control the combustion process as precisely as possible in order to improve the operating properties of the internal combustion engine, particularly with regard to fuel consumption, efficiency, pollutant emissions and smooth running. It is essential to operate the injectors in such a way that the fuel quantity to be injected can be metered with high repeatability, with varying working pressures and, if necessary, with several injections per working cycle.

Fuel injectors may include an electromagnetic actuator or piezo actuator that actuates a valve needle to lift it from a needle seat and open an injector orifice to discharge fuel into the combustion chamber. Due to structural differences and different operating conditions such as temperature, fuel pressure, fuel viscosity, there is an uncertainty in determining the exact opening time, i. H. the point in time at which fuel enters the combustion chamber of the cylinder through the injector and the exact point in time when the injector closes, d. H. the point in time up to which fuel enters the combustion chamber of the cylinder through the injector.

Disclosure of Invention

According to the invention, a method for operating a fuel injection valve according to claim 1 and a device according to the independent claims are provided.

Further developments are specified in the dependent claims.

• - Bereitstellen einer Auswertungspunktzeitreihe durch Abtasten eines Sensorsignals eines Sensors des Einspritzventils,
• - Verwenden eines datenbasierten Zeitpunktbestimmungsmodells mit der Auswertungspunktzeitreihe, um eine modellierte Angabe zu einem Öffnungs- oder Schließzeitpunkt und mindestens eine Systemgröße zu erhalten, wobei das Zeitpunktbestimmungsmodell trainiert ist, um abhängig von der bereitgestellten Auswertungspunktzeitreihe einen Öffnungs- oder Schließzeitpunkt und die modellierte Systemgröße anzugeben,
• - Ermitteln eines Konfidenzwerts abhängig von einer erfassten Systemgröße und der mithilfe des datenbasierten Zeitpunktbestimmungsmodells abhängig von der Auswertungspunktzeitreihe bestimmten modellierten Systemgröße;
• - Betreiben des Verbrennungsmotors abhängig von dem Öffnungs- oder Schließzeitpunkt und abhängig von dem Konfidenzwert.

According to a first aspect, a method is provided for operating an injection system of an internal combustion engine by determining an opening or closing time of the injection valve based on a sensor signal, with the following steps:

• - providing an evaluation point time series by scanning a sensor signal of a sensor of the injector,
• - Using a data-based point in time determination model with the evaluation point time series in order to obtain a modeled indication of an opening or closing point in time and at least one system variable, the point in time determination model being trained to indicate an opening or closing point in time and the modeled system variable depending on the evaluation point time series provided,
• - Determining a confidence value as a function of a detected system variable and the modeled system variable determined using the data-based point in time determination model as a function of the evaluation point time series;
• - Operation of the internal combustion engine depending on the opening or closing time and depending on the confidence value.

Although an injection valve is activated according to a predetermined profile of an activation signal, the resulting opening and closing movements of the injection valve vary, so that the actual opening and closing times for starting and ending fuel injection cannot be precisely predetermined. The reason for this lies in the complex dependencies of the valve movement on the current operating point.

In order to monitor the valve movement, a piezo sensor is provided in the injectors, which is designed as a pressure sensor in order to detect the changes in fuel pressure triggered by the activation of the injector and to provide a corresponding sensor signal. The measured sensor signal can now be evaluated to determine the actual opening and closing times of the injection valve in order to adjust the activation of the injection valve accordingly.

However, the sensor signal is also subject to noise and depends in particular on the actual fuel pressure in the fuel supply and the duration of the actuation to be measured.

The evaluation of the sensor signal to determine an opening and/or closing time of the injection valve can be carried out with the help of a data-based time determination model, which can be embodied in the form of a classification model, for example.

The use of a data-based point in time determination model for determining an opening or closing point in time is advantageous since suitable modelling, in particular with the aid of a physically motivated model, cannot be mapped due to the diverse influences and interactions. However, in safety-critical applications, the use of purely data-based models is questionable, since it cannot be ensured for every operating point that the output of the evaluation model will not lead to undesired system behavior.

The data-based point in time determination model can be designed and trained to output an output vector, which represents a modeled indication of the opening or closing point in time, depending on an input vector that indicates the evaluation point time series.

For example, an output class value of "1" might indicate that the time corresponds to a time associated with that output class. Similarly, an output class value of "0" may indicate that the open or close time does not correspond to a time associated with that output class. Such a classification model outputs a value for each output class, which indicates a probability with which the time assigned to the corresponding output class is the opening or closing time to be determined.

The index value of the element with the maximum element value in the output vector therefore indicates the opening and closing times to be determined in the evaluation period. The assignment of the index value to a corresponding point in time is specified by the training of the point in time determination model and specified in particular relative to a crankshaft angle or with reference to the start point in time of triggering the injector by the triggering signal for the actuator unit of the injector.

The data-based evaluation model can be a multi-layer neural network, for example, which has multiple layers with respective activation functions. In particular, the evaluation model can be designed and trained in such a way that, in addition to the opening or closing time, model values of at least one system variable are also modeled.

In particular, the data-based time determination model can be trained based on training data sets, which each assign an evaluation point time series to an opening or closing time and at least one system variable, the time determination model being a neural network with an input layer with a plurality of neurons, with one or more intermediate layers each with a plurality of neurons and a Output layer with several neurons for outputting the opening or closing time and a further output layer for outputting the at least one system variable.

The model values of the modeled system variable can be compared with actual values of the system variable, for example obtained from a measurement or derived from a corresponding activation of an actuator or determined from a calculation model. With the help of a monitoring function, which is e.g. executed in a control unit, the difference between the provided/recorded system size and the modeled system size can now be determined and depending on the difference, the opening or closing time determined using the evaluation model can be used or discarded. In particular, if there is a difference of more than a predetermined threshold value, the determined opening or closing time can be discarded and instead a predetermined substitute value, such as the last determined or another opening or closing time determined, can be assumed.

Based on the opening or closing time, an actually injected quantity of fuel is determined in a manner known per se and the internal combustion engine is operated accordingly.

The injection valve can be operated as a function of the opening and/or closing time, with the injection valve being operated in particular in such a way that an opening duration of the injection valve, which is determined by the determined opening and/or closing time, falls to a specified target -Opening time is set.

• - Trainieren des neuronalen Netzes mit Trainingsdatensätzen, die eine Auswertungspunktzeitreihe auf einen Öffnungs- oder Schließzeitpunkt abbilden, um Modellparameter der Neuronen der Eingangsschicht, der Neuronen der einen oder der mehreren Zwischenschichten und der Neuronen der Ausgabeschicht zu bestimmen;
• - Anschließendes Trainieren des neuronalen Netzes mit den Trainingsdatensätzen, die jeweils weiterhin die Auswertungspunktzeitreihe auf mindestens eine Systemgröße abbilden, wobei die zuvor erhaltenen Modellparameter der Neuronen der Eingangsschicht und der Neuronen der einen oder der mehreren Zwischenschichten beibehalten werden und lediglich die Modellparameter der Neuronen der weiteren Ausgabeschicht trainiert werden.

According to a further aspect, a method for training a neural network that is used as a time determination model for determining an opening and closing time and for determining a confidence value is provided, the neural network having an input layer with a plurality of neurons, with one or more intermediate layers with each with a plurality of neurons and an output layer with a plurality of neurons for outputting the opening or closing time and a further output layer for outputting the at least one system variable, with the following steps:

• - training the neural network with training data sets that map an evaluation point time series to an opening or closing time in order to determine model parameters of the neurons of the input layer, the neurons of the one or more intermediate layers and the neurons of the output layer;
• - Subsequent training of the neural network with the training data sets, which each continue to map the evaluation point time series to at least one system variable, with the previously obtained model parameters of the neurons of the input layer and the neurons of one or more intermediate layers being retained and only the model parameters of the neurons of the further output layer be trained.

The training method can in particular be carried out on a computer and the method can therefore be understood as a computer-implemented method.

According to a further aspect, a device for carrying out one of the above methods is provided.

character list

• 1 eine schematische Darstellung eines Einspritzsystems für die Einspritzung von Kraftstoff in den Zylinder eines Verbrennungsmotors;
• 2 ein Flussdiagramm zur Darstellung eines Verfahrens zur Anwendung eines trainierten, datenbasierten Zeitpunktbestimmungsmodells zum Bestimmen einer eingespritzten Kraftstoffmenge;
• 3 eine schematische Darstellung des Zeitpunktbestimmungsmodells als neuronales Netz.

Embodiments are explained in more detail below with reference to the accompanying drawings. Show it:

• 1 a schematic representation of an injection system for injecting fuel into the cylinder of an internal combustion engine;
• 2 a flow chart for representing a method for using a trained, data-based time determination model for determining an injected fuel quantity;
• 3 a schematic representation of the time determination model as a neural network.

Description of Embodiments

1 shows an arrangement of an injection system 2 for an internal combustion engine 1 of a motor vehicle, for which a cylinder 3 (of in particular a plurality of cylinders) is shown as an example. The internal combustion engine 1 is preferably designed as a diesel engine with direct injection, but can also be provided as an Otto engine.

The cylinder 3 has an inlet valve 4 and an outlet valve 5 for supplying fresh air and for discharging combustion exhaust gas.

Furthermore, fuel for operating the internal combustion engine 1 is injected via an injection valve 6 into a combustion chamber 7 of the cylinder 3 . For this purpose, fuel is supplied to the injection valve via a fuel supply line 8, via which fuel is made available under a high fuel pressure in a manner known per se (eg common rail).

The injection valve 6 has an electromagnetically or piezoelectrically controllable actuator unit 61 which is coupled to a valve needle 62 . When the injector 6 is closed, the valve needle 62 sits on a needle seat 63. When the actuator unit 61 is activated, the valve needle 62 is moved in the longitudinal direction and releases part of a valve opening in the needle seat 63 in order to inject the pressurized fuel into the combustion chamber 7 of the to inject cylinder 3.

The injection valve 6 also has a piezo sensor 65 which is arranged in the injection valve 6 . The piezo sensor 65 is deformed by pressure changes in the fuel guided through the injection valve 6 and is generated as a sensor signal by a voltage signal.

The injection takes place under the control of a control unit 10 which specifies a fuel quantity to be injected by energizing the actuator unit 61 . The sensor signal is time-sampled with the aid of an A/D converter 11 in the control unit 10, in particular with a sampling rate of 0.5 to 5 MHz.

Furthermore, a pressure sensor 8 is provided in order to determine a fuel pressure upstream of the injection valve 6 .

During operation of internal combustion engine 1, the sensor signal is used to determine the correct opening or closing time of injection valve 6. For this purpose, the sensor signal is digitized using A/D converter 11 into an evaluation point time series and evaluated by a suitable evaluation model, from which an opening time of injection valve 6 and according to a turned on injected fuel quantity can be determined depending on the fuel pressure and other operating variables. In particular, an opening time and a closing time are required to determine the opening time in order to determine the opening time as the time difference between these variables.

An opening time and/or a closing time can be determined by considering the sensor signal curve. In particular, the opening time and/or the closing time can be carried out using a data-based evaluation model.

2 uses a flowchart to illustrate a method for determining an opening and/or closing time of an injection valve 6 of a cylinder 3 in an engine system 1. The method is used to perceive control functions for the internal combustion engine 1 and in particular to determine the injected fuel quantity.

3 shows a block diagram of the data-based time determination model and the monitoring unit for determining a confidence value of the modeled opening and closing time.

The method is executed in the control unit 10, in which it can be implemented in the form of software and/or hardware.

The method is described based on the consideration of an injection valve 6 for an individual cylinder 3 of the internal combustion engine 1 . Of course, the method can be used in parallel for all injection valves 6 of the internal combustion engine 1, i. H. also for multi-cylinder internal combustion engines.

A sensor signal is detected using the piezo sensor 65 in step S1. This signal is typically a voltage signal generated as a result of pressure changes in the fuel being supplied.

In step S2, the sensor signal can be sampled using the A/D converter in order to determine an evaluation point time series E within an evaluation period. The evaluation period can be defined in relation to a control time window of injector 6 . The control time window is defined by the start of the control of the actuator unit 61 and a fixed period of time, which is a maximum period of time in which the control signal for the actuator unit 61 specifies a valve opening. The control time window thus has a defined time reference for which an evaluation point time series Z is provided, which represents the basis for the further determination of an opening or closing time A. In particular, the evaluation point time series E can be determined by downsampling the previously oversampled sensor signal.

The evaluation period can be provided with a fixed time reference to the working strokes of the internal combustion engine 1; in particular, the evaluation period can begin at a predetermined crankshaft position, preferably within the compression stroke. The evaluation period can be selected in such a way that the entire opening time window of the injection valve can be mapped in it.

In step S3, the evaluation point time series E is supplied to a point in time determination model 20 in order to obtain an indication of an opening or closing point in time A. The time determination model 20 can be trained accordingly in order to specify an opening or closing time A in a suitable manner depending on the evaluation point time series 20 . The opening and closing time A can be suitably provided as an output vector.

For example, the evaluation point time series E can be specified as a vector, with each element of the vector being assigned to a time segment. The value of an element of the vector then indicates the measured value of the sensor signal at the relevant time segment.

The output vectors denoting the opening or closing time A are in the form of a logit, with the index of the elements of the output vectors having a corresponding opening or closing time A, i. H. the time specification. For example, given a number of N evaluation points in the evaluation point time series E, the output vector can correspondingly comprise a number of N elements. The index values of the elements of the output vector are assigned to consecutive points in time within the evaluation period considered. In particular, the points in time assigned to the elements/index values of the output vector can correspond to the evaluation points in time, which are evenly spaced apart in time, or indicate successive points in time in some other way.

The data-based point in time determination model 20 can be designed in the form of a neural network 20, as is the case, for example, in 3 is specified. The neural network 20 can have an input layer 21 with a plurality of neurons 22, one or more intermediate layers 23 with each a predetermined number of neurons 22 and an output layer 25 with a number of neurons 22 corresponding to the number of elements of the output vector. Each neuron 22 can have an activation function for non-linearization. In this case, the last intermediate layer 23 forms a feature layer.

The neural network 20 can have a further output layer 25 in order to output model values of one or more system variables S. For this purpose, the further output layer 25 uses the outputs of the preceding intermediate layer 23, which serves as a feature layer.

In step S4, the model values of the system variables S are also determined with the aid of the neural network 20. For this purpose, the neural network 20 of the time determination model is used up to the last intermediate layer 23 and the corresponding further output layer 25 is provided for the system variables S to be determined, which evaluates the results of the last intermediate layer 23 of the neural network 20 with the corresponding further output layer 25. Instead of the last intermediate layer 23, another intermediate layer 23 can also be used as the feature layer.

The training data for training the neural network includes training data sets that include realistic evaluation point time series, each of which is assigned an output vector as a label.

For example, the evaluation point time series of the training data sets can be specified as a vector, with each element of the vector being assigned to a time segment. The value of an element of the vector then indicates the measured value of the sensor signal at the relevant time segment.

In order to create the output vector as a label, the period of time in which the opening or closing time A to be determined lies can be subdivided into time segments. The output vectors may be in the form of a logit, with the index of the elements of the output vectors indicating a corresponding opening and closing time A, i. H. indicating the time as described above.

The time determination model can then be trained on the training data created in this way using conventional training methods, such as backpropagation, in order to determine the weights and, if applicable, bias values of the neurons 22 in the intermediate layers 23 and the output layer 24 .

Furthermore, a value of at least one system variable S is assigned to each of the training data sets. The system variables S can include a fuel pressure and a control time at which the injection valve 6 is controlled to open. The fuel pressure can be measured with the pressure sensor 8 and the activation time is known in the control device for operating the internal combustion engine 2 . These system variables S are also available as measured variables or internal variables of injection system 2 . Thus, both a measured system variable S′ and a modeled system variable can be provided for each injection cycle.

The further output layer 25 for outputting the model values of the respective system variables is trained in a corresponding manner after training the time determination model, with the intermediate layers 23 and the input layer 21 being identical. When training the model for the system variables, only the weights and model parameters of the output layer 25 for the system variables are adjusted in order to obtain the system variable to be trained. The model parameters (weights and bias values) of the neurons 22 of the input layer 21 and the intermediate layers 23 remain unchanged after the training for outputting the opening or closing time A.

In step S5, the system variables S, which are modeled by the time determination model 20 in addition to the opening or closing time A, are recorded by measuring or querying an internal state of the control unit.

In a subsequent step S6, a check is made as to whether a relative or absolute deviation (difference) between the detected and modeled system variable is greater than a correspondingly predefined threshold value. If this is the case (alternative: yes), the confidence of the determined model value for the opening or closing time A is considered to be too low and the opening and closing time A determined by the time determination model is discarded. Instead, an opening or closing time that was previously determined or determined in some other way is assumed to be the correct opening and closing time and is used in step S7 to operate the internal combustion engine 1 . In particular, the fuel quantity actually injected can be determined based on the opening or closing time to be determined, and the engine system can be operated accordingly.

If it is determined that the deviation/difference between the measured and modeled system variable S is less than the specified threshold value (alternative: no), the determined opening or closing time is trusted and used in a subsequent step S8 to operate the internal combustion engine 1. esp In particular, the actually injected fuel quantity can be determined based on the opening or closing time to be determined and the engine system can be operated accordingly.

Claims (9)

Method for operating an injection system (2) of an internal combustion engine (1) by determining an opening or closing time (A) of the injection valve (6) based on a sensor signal, the method comprising the following steps: - Providing (S1, S2) an evaluation point time series by scanning a sensor signal of a sensor (65) of the injection valve (6), - Use (S3, S4) of a data-based time determination model (20) with the evaluation point time series (E) to obtain modeled information about an opening or closing time (A) and at least one system variable (S), the time determination model (20) being trained is to specify an opening or closing time (A) and the modeled system size (S) depending on the evaluation point time series (E) provided, - determining (S5) a confidence value as a function of a detected system variable (S) and the modeled system variable (S) determined using the data-based point in time determination model (20) as a function of the evaluation point time series (E); - Operation (S6, S7) of the internal combustion engine (1) depending on the opening or closing time (A) and depending on the confidence value. procedure after claim 1 , wherein the data-based point in time determination model (20) is designed and trained to output an output vector, which represents a modeled indication of the opening or closing point in time (A), depending on an input vector that indicates the evaluation point time series (E). procedure after claim 1 or 2 , wherein the data-based point in time determination model (20) is trained based on training data sets, which each assign an evaluation point time series (A) to an opening or closing point in time (A) and at least one system variable (S), the point in time determination model (20) as a neural network with an input layer with a plurality of neurons, with one or more intermediate layers (23) each with a plurality of neurons and an output layer (24) with a plurality of neurons for outputting the opening or closing time (A) and a further output layer (25) for outputting the at least one system variable (p ) is trained.. Procedure according to one of Claims 1 until 3 , the internal combustion engine (1) being operated as a function of an opening or closing point in time (A) that was determined beforehand or determined in some other way if the confidence value as a difference specification between the at least one detected system variable (S) and the at least one modeled system variable (S) exceeds a predetermined threshold. Procedure according to one of Claims 1 until 4 , the operation of the injection valve (6) being carried out as a function of the opening and/or closing time, the operation of the injection valve (6) being carried out in particular in such a way that an opening duration of the injection valve (6) which is determined by the opening and / or closing time (A) is determined, is set to a predetermined target opening time. Method for training an artificial neural network, which is used as a time determination model for determining an opening and closing time (A) and for determining a confidence value, the neural network having an input layer (21) with a plurality of neurons (22), with one or more Intermediate layers (23) each having a plurality of neurons (22) and an output layer (24) having a plurality of neurons for outputting the opening or closing time (A) and a further output layer (25) for outputting the at least one system variable (S), comprising following steps: - Training of the neural network with training data sets that map an evaluation point time series to an opening or closing time in order to model parameters of the neurons (22) of the input layer (21), the neurons of the one or more intermediate layers (23) and the neurons of the output layer (24 ) to determine; - Subsequent training of the neural network with the training data sets, which each continue to map the evaluation point time series to at least one system variable, with the previously obtained model parameters of the neurons of the input layer (21) and the neurons of the one or more intermediate layers (23) being retained and only the Model parameters of the neurons of the further output layer (25) are trained. Device, which is set up, one of the methods according to one of Claims 1 until 6 to perform. Computer program product, comprising instructions which, when the program is executed by a computer, cause the latter to carry out the steps of the method according to one of Claims 1 until 6 to execute. A machine-readable storage medium comprising instructions which, when executed by a Computer cause this, the steps of the method according to one of Claims 1 until 6 to execute.

Images