DE102007032062B3 - Method for determining the control parameters of a control device and control device operating according to this method - Google Patents

Abstract

By a method for determining the control parameters of a control device, which regulates the operation of the various cylinders of an automotive engine individually, a control model is created and a time grid for a simulation of the control device based on the number of cylinders and the speed of the engine set. The reference variable, the controlled variable, the controller output variable, the manipulated variable and the system deviation are measured for a predetermined reference variable jump, and a simulation of the controller for the same reference variable jump is performed. The results of the simulation and the measurement of the control model are compared. If the results match, the open loop transfer function is created and the phase margin for that loop is entered. The parameters for a given number of engine speeds are calculated and transmitted to the controller.

Description

The The invention relates to a method according to the preamble of claim 1 and a control device according to the preamble of claim 9. The method is used to determine the control parameters of a control device, the operation of the various cylinders of a motor vehicle engine individually regulated.

A such control device comprises: a route whose output represents the controlled variable and an actuator whose time constant is very small. To control the actuator uses a controller that is infinitely fast Is accepted. This actuator is controlled by a PI controller. The algorithm realized as an algorithm is at discrete times calculated using a variable time grid for the control. The time is between two consecutive calculations of the controller substantially over the time constants of actuator and actuator. The to capture the controlled variable required Measuring device has a dead time, so that the recycled or feedback Measured variable of the controlled variable delayed follows. The dead time is in the range of the time grid of the control or a Many of them.

One known simulation model of a piezoactuator (Wang Q .: Piezoactuators for applications in motor vehicles, metrology and modeling, dissertation, Bochum, 2006) simulates the non-linear behavior for large signal operation. It will in a for the automotive industry is expanding relevant temperature range, taking first the complexity the model is gradually reduced by the expected Non-linear behavior through appropriate measures such as low clock frequency and slow current or power pulses gradually excluded were. Thereafter, a test stand from the mechanical structure, the Control and measurement electronics and the computer-aided control further developed. The tests at different temperature are automated, because under these conditions a manual intervention in the mechanical Setting is excluded. By comparison between simulation and measurement results, the quality of the created model is evaluated.

Known are also a modeling and adaptive dosing optimization method for piezoactuators in motor vehicle injection valves (Melbert J .: piezo actuators in Kfz injection valves, MTZ 03/2006, Jg. 67, P. 190-197). One exact model for Piezoactuators enabled the precise one Simulation and optimization of the injector in the early development phase. By the simultaneous use of the piezoelectric element as an actuator and sensor can Disturbances like Temperature, pressure fluctuations or wear compensated become. Also, the continuous observation of the internal mechanical Events in the Injector is possible. The aim is one of the temperature and the mechanical load almost independent Deflection of the piezoactuator.

In a known device for parameter identification of a transmission path, in particular a controlled system in manual operation, ( DE 41 20 796 A1 ) input and output signals are detected. The input signals are fed to a simulation model of the transmission link for determining estimated output signals. By varying the parameters of the simulation model, the parameters are determined with which the deviations between the detected output signals and the output signals estimated with the aid of the simulation model are smallest. The simulation model of the transmission link is a model with an integrating component, and means are also provided for generating a pulse-shaped input signal.

Of the Invention has for its object to provide a method which allows the controller parameters of a controller depending on from the changeable one Time interval of the calculation to determine.

The The object of the invention is achieved by a method according to claim 1 and a device according to claim 9 solved. In the process is create a control model and a time grid for a simulation the control device based on the Zy cylinder number and the speed set the engine; the reference variable, the controlled variable, the Controller output, the Manipulated variable and the Control difference will be for a predetermined guide size jump measured; a simulation of the controller will be for the same Command variable jump carried out; the results of the simulation and the measurement the control model are compared; at agreement the results will be the transfer function created the open loop, the transient response of the controlled system is determined by specifying the phase reserve (PRE); and the parameters for one given number of engine speeds are calculated and sent to the Control device transmitted.

Appropriate further education The invention are laid down in the subclaims.

The Advantages of the invention are in particular that with her the possibility is created, based on a simple modeling one for the above called control device suitable control loop a good calibration for the Controller parameters, ie proportional and integral components, depending on to reach from the time grid of the regulator.

embodiments The invention will be explained below with reference to the drawings. It demonstrate:

1 a cylinder of a motor vehicle engine with piezo-controlled injectors, which are controlled according to the invention cylinder-individually;

2 the model of a control device that realizes the method according to the invention;

3 the locus of an open loop in a Nyquist diagram;

4-1 to 4-4 the program flowchart of a method according to the invention, and

5 a method for post-optimization of the controller parameters, with constant phase margin.

In 1 are components of a motor vehicle engine 1 shown schematically, as far as they are necessary for explaining the method according to the invention: a piston 2 with a connecting rod 3 are in a cylinder 4 guided. In a cylinder head 5 is an injector 6 arranged, which is designed as a piezo-controlled injection valve. In the present example, the motor vehicle engine 1 designed as a diesel engine with piezo-controlled injection valves. An engine control unit 7 contains a core controller (also known as core ECU) 8th and a power amp 9 passing through a control line 10 with the injector 6 connected is.

The final stage 9 is on the control unit 7 integrated and serves as a controller in a control loop. In the core controller 8th An algorithm for calculating an actuator command CUR_CHA expires. The result is through the power amplifier 9 converted into electrical control commands. In the control line 10 for the injector 6 also the current and the voltage can be measured and thus the actual energy value of the actuation of the injection process can be detected.

In the control unit 7 a control algorithm is executed. A measure of the amperage or magnitude of voltage for controlling the injector is the variable CUR_CHA. This variable CUR_CHA is a cylinder-specific variable contained in the software, ie the program, of the control unit 7 is calculated and has a value range of 0-100%. The greater the value of the variable CUR_CHA calculated in the control unit, the stronger the electrical actuation by the output stage.

The one from the power amp 9 discharged electrical current controls the injectors 6 , This control leads to a mechanical expansion of a piezoelectric drive in the respective injector, which is also referred to as an indirect piezoelectric effect. Due to the expansion of the piezo drive fuel is injected into the respective cylinder. In order to match the injectors with respect to the mechanical elongation, the energy of the individual piezo drives is used as a gender equality feature. The energy is measured and returned after reaching the maximum in each cylinder. Corresponding to the returned energy, an energy setpoint is controlled individually for each cylinder by changing the manipulated variable CUR_CHA.

When Time grid for the energy control four segment times are selected in a 4-cylinder engine, since all four segments the same injector with one injection or an injection sequence is actuated becomes. The segment time is defined as half a crankshaft revolution, this means from -90 ° to + 90 ° crankshaft angle. The same consideration applies to all four cylinders.

The time constant of the actuator, that is the electric power amplifier 9 , and the actuator, that is the piezoelectric actuator 6 , are much smaller than the time intervals between the calculations of the controller. The controller is calculated individually for each cylinder. This means that in the case of a 4-cylinder engine, the controller will calculate all four segments. In the above 4-cylinder engine is injected per crankshaft revolution in two cylinders. Thus, one crankshaft revolution corresponds to two segments of equal length. As the speed is changed, the segment times and thus the time slot of the PI controller also change.

In the software architecture used, the control of the injector takes place in one working cycle × (Work is the time until the same cylinder is re-controlled in firing order), and a working game later, x + 1, measures the energy. This means that between the applied manipulated variable - the cause - and the measurement of the resulting energy - the effect - on the manipulated variable ei ne dead time of four segments or a work cycle. For a modeling, this means a deadtime element in the measurement feedback.

The Track behavior changes over the runtime, so that the control parameters no longer fit optimally. Therefor can be an adaptation or self-optimization of the controller in the current Operation are made. For this purpose, the transmission behavior is dependent on the operating point the route determines - as Linear relationship between manipulated variable and controlled variable. outgoing from the route description can with the method according to the invention iteratively also operating point dependent a parameter set for the controller can be determined.

at a control loop that corresponds to the engine control system shown above, is with the method according to the invention very effectively realizes a modeling and also the controller data established. In particular, based on a desired and predetermined phase reserve of the open loop, the control parameters speed dependent in Nyquist plot certainly. This procedure is advantageous in engine controls used with cylinder-specific regulations, since there the dead time is a multiple of the segment time and thus large dead times occur.

One example for the production according to the invention a control command will now be based on the calibration of a cylinder individual Energy controller for Piezo-controlled injection valves of a diesel engine shown.

Through the control unit 7 As described above, the control algorithm is calculated and converted by a power amplifier into electrical control commands.

A system model for the entire control is now based on 2 explained in detail. It is characterized by its simplicity.

Input variables are a setpoint and a reference variable jump (in 2 as a "bar"), which is fed to a controller limit, output variables is the manipulated variable energy, in 2 represented as "scope 1." A precontrol value is in 2 The intermediate values setpoint, actual value, control deviation, manipulated variable and controller output signal are summarized and displayed in a field "Scope".

The Contains controlled system the complete function chain from the calculation of the manipulated variable CUR_CHA, ie the conversion of the manipulated variable into electric control profile in the electric power amplifier up to on the piezoelectric drive (hereinafter referred to as piezo) resulting energy. This is a transfer characteristic with the Input CUR_CHA and the output energy used. Of delay times is apart, since the time constants of the power amplifier as a controller in the Compared to the time grid are negligible. Also the loading / unloading process of the piezo can be neglected compared to the time grid. This results in a characteristic element in the simulation model. The Characteristic element only applies to given environmental conditions. Influencing factors are among others the temperature response of power amplifier and piezo or the piezo acting forces. For calculating, the characteristic element for a given work area linearized. The input of data of the characteristic element is already on the test bench made in pre-controlled operation. For this, the manipulated variable of 0% pass through to 100% and each measured the resulting energy stationary.

Around Designing the controller dynamics with the model requires a guide variable jump "bridge." Before the jump however, must be ideally pre-controlled to close the controller output to hold zero. This happens with a pre - initialization of the Pre-tax value. The pilot control map corresponds to the inverse characteristic curve of the route model. If the jump takes place and becomes the pre-tax value not changed, the controller must compensate the entire control difference.

It Now a case of a rule situation is simulated by model calculation and compared with a measurement under the same conditions. Of the Setpoint for the energy regulator is raised from 40 mJ to 80 mJ. This energy leap is selected, since he is currently the maximum energy jump depending on the ambient conditions corresponds to the real controller system.

The controller has the following forms of transfer function G (s):

In it are:
T _{R is} a parameter, KI is a control parameter, K _R (KR) the gain of the controller, KS or K the gain of the circuit and F ₀ (s) the transfer function of the open circuit.

The sometimes slightly different spelling of sizes with Subindexes results from the fact that in the programming language a Depression corresponding to the mathematical notation is not possible.

The following numerical values are used:
KR = 0.01%, KI = 0.7% / s> T _R = 1.42E-2 s, K _R = 0.7% / s

One Comparison of the temporal results obtained by measurement and simulation courses the size setpoint EGY_INJ_SP, actual value EGY_INJ_SEL, control difference EGY_INJ_CTL_DIF, Control value CUR_CHA and controller output CUR_CHA_ERR_CTL has a good match shown.

At the Setting the controller parameters, it should be noted that in the synthesis of a controller certain criteria must be met. To the First, the control circuit must be stable, that is, the management and Step response for t must strive towards infinity against a fixed value. For the locus of the open circle this means that they left the -1 point (criterion 1). A pure fulfillment of the stability criterion leads however to no satisfactory result. A jump of the setpoint would of course subside, but if this oscillation takes a long time, the control loop is unusable. Thus, it requires another requirement, namely that a leadership or disturbance size jump in a sufficiently short time the stationary value sufficiently close comes. This means that the control loop must have sufficient damping (criterion 2).

When Further requirement, the controller must have sufficient steady state accuracy exhibit. Is in the transmission link of the open circle contain an integral term, it results automatically stationary Accuracy. If not, the loop gain must be raised accordingly (criterion 3). The fourth requirement is that the regulator has to be sufficiently fast to be able to do this in a sufficiently short time stationary value enough come close (criterion 4). The requirements for stationary accuracy and stability are in opposite directions. So there is the task, the best possible compromise for the Find control parameters to satisfy both criteria sufficiently [see Otto Föllinger, Control Engineering, B. Edition, pp. 226-228, Chapter 7.2].

With a first parameter set becomes a basic calibration for the model of the control loop. With the model described above arise for the control loop following transfer functions.

In this GR (s) are the PI controller, GS (s) is the controlled system, and GM (s) is the measurement feedback or -coupling.

The transfer function of the open circle F0 (s) is: F 0 (s) = G R (S) * G s (S) * G M (s) (V)

In the following, the locus of the open circle is considered (cf. 3 ). Shown in a Nyquist diagram with the unit circle EK in the complex plane, the locus OK of the open loop, an angle PRO, which limits the phase reserve of the locus OK, and an angle PRE, which limits the desired phase margin. These angles are counted from the negative axis. The boundary line of the angle PRO intersects the unit circle at a point A, that of the angle PRE in one Point B.

The characteristic element for the controlled system is expediently linearized: G _S (s) = 1.5 mJ /%.

It should be noted that the shape of the characteristic limb strongly depends on the ambient conditions of the piezo. In 3 the locus OK of the open circle for KR = 0.01 and KI = 0.7 is shown. For this parameter selection of the controller, the controlled system becomes stable because the locus leaves point -1 on the left.

• [Otto Föllinger Regelungstechnik 8. Auflage S. 241]

For the PI controller, the loop gain is expediently chosen so that there is sufficient damping. As an approximate guideline for sufficient damping and yet sufficiently fast response, the loop gain is chosen as follows, satisfying criteria 1, 2 and 4. V = 0.05 * V G ... 0.1 · V G with V G is stability limit (VI) V = K R · K S (VII)

• [Otto Föllinger Regelstechnik 8th Edition p. 241]

Where K _{R is} the gain of the controller and K _{S is} the gain of the distance [KS ≈ 1.5 (linearized)]. V _G is the gain at which the locus goes through the point -1. The locus shows a value V _G = 9 for this example of power regulation. Thus, K _R = 0.3 ... 0.6.

For the election of the parameter TR, note that there are no denominator time constants in the transmission behavior the distance, and thus the time constants of the Regulator not made on the basis of the dominant track time constant can be. It will now be with the described controller model of the parameters KR increased from 0.01 to 0.6.

The control loop becomes relatively unstable with increasing KR, as it does not leave the point -1 in the complex plane left. KR has always been very small in all cylinder-specific controllers. Theoretically it can be confirmed that with increasing KR the diameter of the spiral OK ( 3 ) of the deadtime member increases rapidly and thus the stability criterion is violated very quickly. This represents a fundamental fact with regard to the controller parameterization. In this context, it is important to note the special significance of the dead time element for the energy back measurement. The associated feedback of the energy after a change in the manipulated variable is always a working game later. Therefore, zero is chosen for the first time for further test results KR, since small KR values already generate great unrest in the control.

To the final Determining the controller parameters based on the controller model according to the invention become basic data for the model with the results obtained above certainly. If the controller parameter KI is kept constant above the rotational speed, Thus, a tendency to oscillate occurs at low speeds.

This is on the bigger dead time attributed and means the reinforcement KR or - at another form of transfer function representation - the gain KI over the Speed changed must become.

Criteria for a controller design based on the locus OK in the Nyquist diagram ( 3 ) are set as follows. A measure of the dynamics and the transient response of the controlled system is the phase reserve PRE. The phase margin is a suitable qualitative measure for the attenuation of the control jump response of the control loop. The phase reserve should not fall below 40 °. A normal transient with not too long transient times can be expected with a phase margin of 50 ° to 70 °. An aperiodic transient is achieved from a phase reserve greater than 80 ° to 90 °. Furthermore, the magnitude characteristic in a not too small environment of the phase reserve ω0 should have an attenuation of 20 dB. [Otto Föllinger Regelstechnik 8th Edition, p. 229, chapter 7.2].

Based on the program schedule of the 4-1 to 4-4 An algorithm will now be described with which the control parameters for the modeling according to the invention are calculated when the phase reserve ω0 is specified. As mentioned, the controller parameterization must be performed for different speeds. This means that the algorithm also has to be processed for different speeds. Furthermore, the described transfer function of the open circuit F0 (s) only applies to one parameter set of the controller. This must be varied so that F0 (s) intersects the unit circle in the complex plane so that the phase reserve is formed according to the desired information.

• S1 – die Modellierung der zylinderindividuellen Regelung gemäß den beschriebenen Kriterien;
• – die Festlegung des Zeitrasters der Simulation, indem die Segmentzeit anhand der Zylinderanzahl und der Drehzahl bestimmt werden;
• – eine Darstellung der Messrückführung, indem ein Totzeitglied anhand der Zylinderanzahl und Segmentzeiten bestimmt wird;
• – ein einfache Streckenmodellierung, indem das Kennlinienglied von Stellgröße CUR_CHA zu Regelgröße EGY_INJ aus Messungen identifiziert werden;
• – eine Festlegung der Vorsteuerungsstruktur für den Regler, das heißt des Übertragungsverhaltens Führungsgröße EGY_INJ_SP zu Stellgröße CUR_CHA. Hierfür wird das inverse Kennlinieglied verwendet. Dann erfolgt in einem Schritt
• S2 – eine Verifikation des Modells durch
• – Messung von Führungsgröße EGY_INJ_SP, Regelgröße EGY_INJ, Reglerausgangsgröße CUR_CHA_ERR_CTL, Stellgröße CUR_CHA und Regeldifferenz EGY_INJ_SEL für einen definierten Führungsgrößensprung,
• – eine Simulation desselben Führungsgrößensprungs, und
• – ein Vergleich von Simulation und Messung. Danach erfolgt in einem Schritt
• S3 – eine Abfrage, ob die Ergebnisse von Simulation und Messung im Rahmen der Mess- und Simulationsgenauigkeit gleich sind. Ist die Antwort nein, so wird in einem Schritt
• S4 – festgestellt, dass das Verfahren nur begrenzt einsetzbar ist und es wird abgebrochen. Ist die Antwort ja, so wird in einem Schritt
• S5 – die Phasenreserve F0(j·ω) bestimmt, indem
• – das validierte Modell wird zum Aufstellen der Übertra gungsfunktion des offenen Kreises F0(j·ω) eingesetzt wird,
• – die Phasenreserve für die Übertragungsfunktion F0(j·ω) des offenen Kreises eingegeben wird, und
• – der Schnittpunkts B der Phasenreserve mit dem Einheitskreises der komplexen Ebene berechnet wird.

After starting the program in one step ( 4-1 )

• S1 - the modeling of the cylinder-specific control according to the described criteria;
• The determination of the simulation time frame by determining the segment time based on the number of cylinders and the number of revolutions;
• A representation of the measurement feedback by determining a deadtime element based on the number of cylinders and the segment times;
• A simple route modeling, by identifying the characteristic element of control variable CUR_CHA to control variable EGY_INJ from measurements;
• - A determination of the feedforward structure for the controller, that is, the transfer behavior of reference variable EGY_INJ_SP to manipulated variable CUR_CHA. The inverse characteristic element is used for this purpose. Then done in one step
• S2 - a verification of the model
• - Measurement of reference variable EGY_INJ_SP, controlled variable EGY_INJ, controller output CUR_CHA_ERR_CTL, manipulated variable CUR_CHA and control difference EGY_INJ_SEL for a defined reference variable jump,
• A simulation of the same reference variable jump, and
• - a comparison of simulation and measurement. This is done in one step
• S3 - a query as to whether the results of simulation and measurement are the same in terms of measurement and simulation accuracy. If the answer is no, it will be in one step
• S4 - found that the procedure is limited and it is canceled. If the answer is yes, it will be in one step
• S5 - the phase reserve F0 (j · ω) determined by
• The validated model is used to set up the transmission function of the open circuit F0 (j ·ω),
• - the phase reserve for the transfer function F0 (j · ω) of the open circuit is entered, and
• - the point of intersection B of the phase reserve is calculated with the unit circle of the complex plane.

• S6 – eine Festlegung der Größe des Kennfeldes für den Regelparameter KI, wobei
• – die Regelparameter drehzahlabhängig gestaltet werden, um eine optimale Güte der Regelung zu erreichen,
• – im Falle der Energieregelung ein Kennfeld für die Regelparameter mit sieben Stützstellen über der Drehzahl vorliegt;
• – dies bedeutet, dass die Regelparameter für sieben verschiedene Drehzahlen bestimmt und auch später kalibriert werden können.

Subsequently, in a step ( 4-2 )

• S6 - a determination of the size of the characteristic diagram for the control parameter KI, where
• - The control parameters are speed-dependent designed to achieve optimum control quality,
• - in the case of the energy regulation, a characteristic field for the control parameters with seven interpolation points above the rotational speed is present;
• - This means that the control parameters for seven different speeds can be determined and also calibrated later.

• S7 – eine Initialisierung Drehzahl für einen ersten Durchlauf der Regelparameterbestimmung.

This is then done in one step

• S7 - an initialization speed for a first pass of the control parameter determination.

• S8 – Schleifenbildung über die Anzahl der Stützstellen der Drehzahl. Anschließend wird in einem Schritt
• S9 – eine Initialisierung des Regelparameters KI für einen ersten Durchlauf der Regelparameterbestimmung durchgeführt. Danach erfolgt in einem Schritt
• S10 – eine Schleifenbildung über den Regelparameter KI.

After that takes place in one step

• S8 - Looping over the number of nodes of the speed. Subsequently, in one step
• S9 - carried out an initialization of the control parameter KI for a first pass of the control parameter determination. This is done in one step
• S10 - a loop formation via the control parameter KI.

• S11 – wird die Übertragungsfunktion des offenen Kreises F0(j·ω) als Funktion der Kreisfrequenz ω (für die entsprechende Drehzahl) berechnet. Dann wird in einem Schritt
• S12 – Betrag |F₀(j·ω)| der Übertragungsfunktion des offenen Kreises F₀(j·ω) gebildet. In einem Schritt
• S13 – wird der Index oder das Elements des Vektors |F₀(j·ω)| gesucht, für den |F₀(j·ω)| = 1 gilt;
• – dies stellt den Schnittpunkt der Übertragungsfunktion F₀(j·ω) mit dem Einheitskreis dar (Punkt A in 3). In einem Schritt
• S14 – wird geprüft, ob der Punkt A die Bedingung der Phasenreserve (Punkt B) erfüllt. Ist die Antwort nein, wird der Regelparameter KI dekrementiert und es erfolgt ein Rücksprung zu dem Schritt S10. Ist die Antwort ja, wird in einem Schritt
• S15 – der Regelparameter KI für die jeweilige Drehzahl gespeichert.

In one step ( 4-3 )

• S11 - the transfer function of the open circuit F0 (j · ω) is calculated as a function of the angular frequency ω (for the corresponding speed). Then in one step
• S12 - magnitude | F ₀ (j · ω) | the transfer function of the open circuit F ₀ (j · ω) is formed. In one step
• S13 - becomes the index or element of the vector | F ₀ (j · ω) | searched for | F ₀ (j · ω) | = 1 holds;
• - This represents the intersection of the transfer function F ₀ (j · ω) with the unit circle (point A in 3 ). In one step
• S14 - it is checked whether the point A meets the condition of the phase reserve (point B). If the answer is no, the control parameter KI is decremented and there is a return to step S10. If the answer is yes, it will be in one step
• S15 - the control parameter KI stored for the respective speed.

• S16 – wird geprüft, ob der Regelparameter KI für alle Drehzahlen bestimmt worden ist. Ist die Antwort nein, wird er für eine nachfolgende Drehzahl bestimmt, und es erfolgt dazu ein Rücksprung zu dem Schritt S8. Ist die Antwort ja, so wird in einem Schritt
• S17 – der Regelparameter KI, der die Verstärkung des Reglers steuert, abhängig von der Drehzahl ausgegeben.

In one step ( 4-4 )

• S16 - checks whether the control parameter KI has been determined for all speeds. If the answer is no, it is determined for a subsequent speed, and a return is made to step S8. If the answer is yes, it will be in one step
• S17 - the control parameter KI, which controls the gain of the controller, depending on the speed output.

In order to a program run is finished. The program is repeated cyclically.

A Self-optimization of the controller according to the inventive method can be done as follows become.

For the transfer function of the open circuit, the following applies:

If the controller is implemented in a control unit and provided with output data according to the method described above, the controller parameters KR and TR of the open-circuit transfer function are known. Furthermore, the parameters of the measurement feedback KM and TM are known. The transmission behavior of the track can change - for example due to aging and temperature effects. It thus requires a Nachoptimierung the controller parameters when a constant constant phase margin is desired. Such an adaptation of the controller parameters is carried out during operation according to the following methods ( 5 ).

at the transfer function the open circle is the only unknown the transmission behavior the way. Furthermore the controlled system is specified as linear element. Since the manipulated variable is specified by the control unit and therefore is known, and the control variable is measured back, can for each Operating point the transmission behavior the route can be identified. Based on this route behavior are all parameters of the transfer function known. Now, the control parameters are each iteratively optimized, that for himself the desired operating point sets the desired phase reserve. Depending operating point Now the parameter set for the controller is stored or adapted.

Claims (10)

A method for determining the control parameters of a control device, which regulates the operation of the various cylinders of an automotive engine individually, characterized in that - a control model is created and a time grid for a simulation of the control device based on the number of cylinders and the speed of the engine is set; - that the reference variable (EGY_INJ_SP), the controlled variable (EGY_INJ), the controller output CUR_CHA_ERR_CTL), the manipulated variable (CUR_CHA) and the system deviation (EGY_INJ_SEL) are measured for a given reference variable jump, - that a simulation of the control device for the same reference variable jump is performed; - that the results of the simulation and the measurement are compared to the control model; If the results agree, the transfer function (F0 (j · ω)) of the open loop is established and the transient response of the controlled system is determined by specifying the phase margin (PRE), and - the parameters are calculated and calculated for a given number of different engine speeds the control device are transmitted. Method according to claim 1, characterized in that that a distance model is created by the characteristic element as a ratio of manipulated variable CUR_CHA to rule size EGY_INJ is determined from measurements. A method according to claim 1, characterized in that a pre-control structure for the controller as a transfer behavior reference variable (EGY_INJ_SP) to manipulated variable (CUR_CHA) using the Inverse characteristic member is set. Method according to claim 1, characterized in that that a model of the controlled system by measuring reference variable (EGY_INJ_SP), Controlled variable (EGY_INJ), Controller output variable (CUR_CHA_ERR_CTL), manipulated variable (CUR_CHA) and control difference (EGY_INJ_SEL) for a defined reference variable jump is verified. Method according to claim 1, characterized in that - that the Size of the map for the Control parameter (KI) is set, the control parameters designed speed-dependent become; - that in the case of energy regulation, a characteristic diagram for the control parameter with a given number of nodes over the Speed is provided, and - that the control parameter for the given Number of speeds is determined and calibrated. Method according to claim 5, characterized in that that an initialization of the control parameter (KI) for a first pass of the control parameter determination and then looping over the control parameter (AI) become. Method according to claim 1, characterized in that - the transfer function of the open circuit (F0 (j · ω)) is calculated as a function of the angular frequency (ω); - that the amount | F ₀ (j · ω) | the transfer function of the open circuit is formed; - that the index of the vector (| F ₀ (j · ω) |), for | F ₀ (j · ω) | = 1 holds, is determined; where this represents the intersection (A) of the transfer function with the unit circle, and that it is checked whether this intersection (A) meets the condition of the phase reserve. Method according to claim 7, characterized in that that in the case that this condition is met, the control parameter (AI) for the respective speed is stored. Control device which controls the operation of the various cylinders of a motor vehicle engine ( 1 ) is individually controlled, characterized in that it is designed in such a way that the regulation of the cylinders is carried out according to a method according to one of the preceding claims. Control device according to claim 9, characterized in that the energy supply to piezoelectric injection valves ( 6 ) in the cylinders ( 4 ) of the motor ( 1 ) is controlled.