DE102020006255A1 - METHOD FOR CALIBRATING A SIMULATION MODEL FOR CONTROLLING AN COMBUSTION ENGINE - Google Patents

Abstract

The present invention relates to a method for calibrating a simulation model (100) with several independent maps, characteristic curves or scalars, with the steps of determining (S101) a reference map for each of the independent maps, characteristic curves or scalars of the simulation model (100); a selection (S102) of a functional relationship with at least one functional parameter for the independent characteristic maps, characteristic curves or scalars, based on the determined reference characteristic fields; defining (S103) at least one initial functional parameter for the functional context; calculating (S104) the function values of the at least one functional relationship of the independent characteristic diagrams, characteristic curves or scalars; calculating (S105) the output value of the simulation model (100) on the basis of the at least one calculated function value; calculating (S106) a deviation value between the calculated output value of the simulation model (100) and a measured target value; and minimizing (S107) the deviation value by varying the at least one functional parameter and repeating the calculation steps (S104, S105, S106).

Description

The present invention relates to a method for calibrating a simulation model with a plurality of independent characteristic diagrams, characteristic curves or scalars and a computer program for performing the method.

The pamphlet DE 102 19 797 A1 describes a method for optimizing a simulation model for controlling an internal combustion engine, the optimization taking place with regard to the accuracy of the simulation model and the smoothness of the characteristic field or characteristic fields used. The smoothness is defined by the sum of the second derivatives, which must be minimized accordingly.

The calibration of characteristic diagrams, characteristic curves and scalars of a model by means of an optimization process, when using a method that only takes into account the accuracy of the simulation model, often leads to courses of the characteristic diagrams and characteristic curves to be optimized, which, within the given model structure, can be described as implausible.

In addition, areas of characteristic diagrams and characteristic curves for which no measurement data are available are not accessible for optimization when purely considering the accuracy of the simulation model.

The smoothness of a map or a characteristic does not always reflect the plausibility of the data within a given model structure. Occasionally, courses with large gradients are even explicitly required. An extrapolation of the course of a characteristic curve or a characteristic field into areas for which no measurement data is available for optimization with regard to the accuracy of the simulation model is only possible to a limited extent.

This object is achieved by subject matter according to the independent claims. Advantageous embodiments are the subject matter of the dependent claims, the description and the figures.

According to one aspect, this object is achieved by a method for calibrating a simulation model with several independent maps, characteristic curves or scalars, with the steps of determining a reference map in each case for the independent maps, characteristic lines or scalars of the simulation model, the reference map also for maps to be optimized is a map, for characteristic curves to be optimized the reference map is a characteristic curve and for scalars to be optimized the reference map is also a scalar; a selection of a functional relationship with at least one functional parameter for the independent characteristic maps, characteristic curves or scalars, based on the determined reference characteristic maps; defining at least one initial functional parameter for the functional context; a calculation of the functional values of the at least one functional relationship of the independent characteristic diagrams, characteristic curves or scalars; calculating the output value of the simulation model on the basis of the at least one calculated function value; calculating a deviation value between the calculated output value of the simulation model and a measured target value; and minimizing the deviation value by varying the at least one functional parameter and repeating the calculation steps. This has the technical advantage that the output value of the model can be adapted quickly to measured values and, at the same time, plausible values and curves result for the independent characteristic diagrams, characteristic curves or scalars.

In a technically advantageous embodiment of the method, the interpolation points of the independent characteristic maps, characteristic curves or scalars are determined by shifting the interpolation points of a reference characteristic field by applying the functional relationship. This has the technical advantage, for example, that a predictable change in the support points is achieved.

In a further technically advantageous embodiment of the method, the functional relationship is described by a parametric function that changes both the support points and the support values of the independent characteristic diagrams, characteristic curves or scalars. In this way, for example, the technical advantage is achieved that the accuracy of the adaptation of the characteristic maps, characteristic curves or scalars is increased.

In a further technically advantageous embodiment of the method, the functional relationship is represented by a spline or a rational Bezier curve. This has the technical advantage, for example, that smooth and continuous characteristic fields, characteristic curves or scalars can be generated.

In a further technically advantageous embodiment of the method, the permitted value range for the independent characteristic diagrams, characteristic curves or scalars is restricted by an upper and / or lower limit. As a result, for example, the technical advantage is achieved that technically meaningful and realizable values are obtained.

In a further technically advantageous embodiment of the method, the functional parameters of the functional relationship are changed only within a predetermined value range. This also has the technical advantage, for example, that technically meaningful and realizable values are obtained.

In a further technically advantageous embodiment of the method, the simulation model has several output values which are each compared with a measured target value and the respective deviation values are weighted. This has the technical advantage, for example, that cross-effects are taken into account within the simulation model for several output values.

In a further technically advantageous embodiment of the method, the simulation model is a charge detection model, a torque model, an exhaust gas temperature model, a fuel temperature model or an oil temperature model. This has the technical advantage, for example, that the method is used for the calibration of particularly suitable technical models.

According to a further aspect, this object is achieved by a computer program with instructions which, when the program is executed by a computer, cause the computer to carry out the method according to the first aspect. As a result, the same technical advantages are achieved as by the method according to the first aspect.

• 1 ein Blockdiagramm des Verfahrens; und
• 2 eine Kennlinie vor und nach der Anpassung durch das Verfahren, sowie ein Referenzkennfeld

Embodiments of the invention are explained in more detail with reference to the following figures. It shows

• 1 a block diagram of the process; and
• 2 a characteristic before and after the adaptation by the method, as well as a reference map

1 shows a block diagram of the method for calibrating a simulation model 100 with several independent maps, curves or scalars. The simulation model 100 is, for example, a simulation model for controlling an internal combustion engine, such as a gasoline or diesel engine in a motor vehicle.

The simulation model 100 contains, for example, characteristic curves or maps that assign an output variable to one or more input variables and scalars that output an output variable. These output variables are linked by mathematical operations and result in the output value of the model. The output value of the model is used to control the internal combustion engine as a function of the input values of the model and the independent characteristic diagrams, characteristic curves and scalars.

In addition to the accuracy of the simulation model 100 the deviation of the characteristic diagrams, characteristic curves and scalars to be optimized from a selected reference characteristic diagram is also taken into account. For characteristic diagrams to be optimized, the reference characteristic diagram is likewise a characteristic diagram, for characteristic curves to be optimized the reference characteristic diagram is a characteristic curve and for scalars to be optimized the reference characteristic diagram is likewise a scalar. Each reference map represents a plausible calibration for the given model structure with regard to the value range and the course.

In the first step S101 becomes a reference map for the independent maps, characteristic curves or scalars 101 of the simulation model 100 certainly. After that, in step S102 a functional relationship with at least one functional parameter is selected for the independent maps, characteristic curves or scalars based on the specific reference maps. In step S103 at least one initial functional parameter is defined for the functional relationship. Then in step S104 the functional values of the at least one functional relationship of the independent characteristic diagrams, characteristic curves or scalars 101 calculated. In step S105 becomes the output value of the simulation model 100 calculated on the basis of the at least one calculated function value. In step S106 becomes a deviation value between the calculated output value of the simulation model 100 and a measured target value. Finally, in step S107 the deviation value is minimized by varying at least one functional parameter and the calculation steps S104 , S105 and S106 are repeated.

The method can be used to efficiently determine characteristic curves, maps or scalars for controlling an internal combustion engine. The characteristic diagrams, characteristic curves or scalars obtained have a plausible value range and a plausible course within the given model structure, since these have been obtained taking reference characteristic diagrams into account.

Furthermore, with this method, areas for which no measurement data can be used to minimize the deviation of the simulation model 100 can be calibrated by simply looking at the reference maps.

In addition, the calibration can be carried out with significantly less expenditure of time and resources, since manual post-processing of the data resulting from the optimization process is less necessary. Due to the significantly reduced effort, cost and time savings are possible.

2 shows a characteristic 101 before the adjustment by the method, the associated reference map in the form of a characteristic curve 102 and the characteristic curve after the adaptation by the method 103 . The characteristic 101 is part of a simulation model 100 with further independent maps, characteristic curves and scalars. For the sake of simplicity, however, it is assumed for the following explanations that the simulation model 100 only this characteristic 101 includes.

The characteristic 102 is determined on the basis of a consideration of the given model structure, the input and output variables as well as existing calibrations of the same model for other real systems. The course of the characteristic 102 is through several support points 104 Are defined. These bases 104 are value pairs from interpolation points 105 (Values for the input variable) and corresponding support values 106 (Values for the output variable).

The course of a changed characteristic 103 can now with the help of a functional relationship based on the support points of the characteristic 102 To be defined. This functional relationship can be given by a parametric representation using rational Bäzier functions, for example:

$$\begin{array}{c}\text{St}\ddot{\text{u}}\text{tzwert Kennlinie 103}=\\ \text{f2}\left(\text{St}\ddot{\text{u}}\text{tzwert Kennlinie 102},\text{ p4},\text{ p5},\text{ p6}\right).\end{array}$$

Function value = f (t, p1, p2, p3). Here t is the curve parameter and the parameters p1, p2, p3 are the function parameters. The changed bases 107 (defined by changed support points 108 and changed support values 109 ) of the changed characteristic 103 can then be described, for example, by the following functional context: $$\begin{array}{c}\text{St.}\ddot{\text{u}}\text{<figure-callout id="103" label="point characteristic" filenames="DE102020006255A1\_0003.png,DE102020006255A1\_0004.png" state="{{state}}">point characteristic</figure-callout> 103}=\\ \text{f1}\left(\text{St.}\ddot{\text{u}}\text{<figure-callout id="102" label="point characteristic" filenames="DE102020006255A1\_0004.png" state="{{state}}">point characteristic</figure-callout> 102},\text{p1},\text{p2},\text{p3}\right);\text{and}\end{array}$$

$$\begin{array}{c}\text{St.}\ddot{\text{u}}\text{tzwert <figure-callout id="103" label="characteristic curve" filenames="DE102020006255A1\_0003.png,DE102020006255A1\_0004.png" state="{{state}}">characteristic curve</figure-callout> 103}=\\ \text{f2}\left(\text{St.}\ddot{\text{u}}\text{tzwert characteristic 102},\text{p4},\text{p5},\text{p6}\right).\end{array}$$

The functions f1 and f2 are each one-dimensional, rational Bezier curves with different function parameters p1, p2, p3 and p4, p5, p6.

By defining initial values for the function parameters p1, p2, p3 and p4, p5, p6, an initial form of the characteristic curve can be determined using the calculated support points 103 determine. On the basis of this characteristic, output values of the simulation model can then be obtained 100 be calculated. In this example, the input value corresponds to the simulation model 100 the input variable of the characteristic curve and the output value of the simulation model 100 corresponds to the output variable of the characteristic.

Corresponding values of the mass flow are calculated for different values of the flow cross section. In a next step, the output values of the simulation model are compared with measured values, for example a mass flow as a function of the cross section, measured on the real system, and a deviation value between the calculated output values of the simulation model 100 and calculated from the measured target values.

By varying the function parameters p1, p2, p3, p4, p5, p6, the shape of the characteristic curve is changed in order to reduce this deviation value, so that an adapted and improved characteristic curve is obtained 103 results, which leads to a higher model accuracy.

The method described can be carried out, for example, by dedicated hardware or by a computer program that is carried out on a computer with a processor and a data memory.

All the features explained and shown in connection with individual embodiments of the invention can be provided in different combinations in the subject matter of the invention in order to simultaneously realize their advantageous effects.

All method steps can be implemented by devices that are suitable for carrying out the respective method step. All functions that are carried out by objective features can be a process step of a process.

QUOTES INCLUDED IN THE DESCRIPTION

This list of the documents listed by the applicant was generated automatically and is included solely for the better information of the reader. The list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions.

Patent literature cited

• DE 10219797 A1 [0002]

Claims (10)

Method for calibrating a simulation model (100) with several independent characteristic diagrams, characteristic curves or scalars (101), with the steps: - Determination (S101) of a respective reference characteristic diagram for the independent characteristic diagrams, characteristic curves or scalars (101) of the simulation model (100); - Selecting (S102) a functional relationship with at least one function parameter for the independent maps, characteristic curves or scalars (101), based on the determined reference maps; - Specifying (S103) at least one initial functional parameter for the functional relationship; - Calculating (S104) the function values of the at least one functional relationship of the independent characteristic maps, characteristic curves or scalars (101); - Calculating (S105) the output value of the simulation model (100) on the basis of the at least one calculated function value; - calculating (S106) a deviation value between the calculated output value of the simulation model (100) and a measured target value; and - Minimizing (S107) the deviation value by varying the at least one functional parameter and repeating the calculation steps (S104, S105, S106). Procedure according to Claim 1 , the interpolation points of the independent characteristic diagrams, characteristic curves or scalars (101) being determined by shifting the interpolation points of a reference characteristic diagram through an application of the functional relationship. Procedure according to Claim 2 , the functional relationship being described by a parametric function that changes both the interpolation points and the interpolation values of the independent characteristic maps, characteristic curves or scalars (101). Method according to one of the preceding claims, the functional relationship being represented by a spline or a rational Bezier curve. Method according to one of the preceding claims, wherein the permitted range of values for the independent characteristic diagrams, characteristic curves or scalars (101) is restricted by an upper and / or lower limit. Method according to one of the preceding claims, wherein the functional parameters of the functional relationship are changed within a predetermined value range. Method according to one of the preceding claims, wherein the simulation model (100) has a plurality of output values which are each compared with a measured target value and the respective deviation values are weighted. Method according to one of the preceding claims, wherein the simulation model (100) is a charge detection model, a torque model, an exhaust gas temperature model, a fuel temperature model or an oil temperature model. Device for carrying out the method according to one of the Claims 1 to 8th . Computer program with instructions which cause a computer to execute the computer program, the method according to one of the Claims 1 to 8th execute.

Images