DE102019200493A1 - Method and device for checking the software of a control device - Google Patents

Abstract

Method (10) for checking the software of a control device, characterized by the following features: - First operands and second operands are selected from the interval (11) according to predetermined criteria, - value pairs are formed (12) from the operands, - for each of the value pairs a function value is calculated using the linkage and an approximation value is calculated using the approximation (13), - a deviation of the approximation value from the function value is output (14), - the first operands are selected from the approximation values and the second operands from the interval according to the criteria (11) and - forming the value pairs (12), calculating (13) the function and approximation values, outputting (14) the deviation and selecting (11) the operands are repeated several times.

Description

The present invention relates to a method for checking the software of a control device. The present invention also relates to a corresponding device, a corresponding computer program and a corresponding storage medium.

State of the art

Numerical mathematics, also referred to in technical terms as "numerics", deals with the construction and analysis of algorithms for continuous mathematical problems as a branch of mathematics. A main application of numerics is the approximate calculation of solutions with the help of computers of all kinds, for example in control engineering.

Of particular technical importance is the numerical error analysis of algorithms that replace a continuous mathematical or physical problem with a discrete, i.e. finite, problem. In this respect, different types of errors come into play in a calculation, for example in an electronic control unit: Rounding errors inevitably occur when calculating with floating point (floating point, hereinafter "float") or integer numbers (hereinafter "integer"). In the case of floating point arithmetic in accordance with the IEEE 754 standard, these errors can be contained but not completely eliminated by increasing the calculation accuracy. Such so-called discretization errors are evaluated according to the prior art in the context of a consistency, their propagation by means of a stability analysis. The so-called approximation theory, in particular the investigation of the approximation of links or other functions, proves to be an important tool, since it can accelerate or make possible computer-aided solution methods.

DE 10 2008 004361 A1 For example, a method for regulating an internal combustion engine is disclosed, in which the setpoint value of a combustion position feature of the combustion process and a related manipulated variable are determined by means of a predictive control based on Greybox modeling of the combustion position feature as a function of the manipulated variable in the combustion process, a value being used as the manipulated variable is determined with which the difference between the target value of the combustion situation feature and the model-based predicted combustion situation is minimized.

Disclosure of the invention

The invention provides a method for checking the software of a control device, a corresponding device, a corresponding computer program and a corresponding machine-readable storage medium according to the independent claims.

The proposed method is based on the knowledge that physical models are mapped in software in control units. In order to reduce computational effort and required storage space, the calculations are carried out with either integer or float values. As described above, this results in deviations from the physical calculation. In the case of vehicle electronic applications, these deviations can lead to incorrect behavior in the vehicle.

The approach according to the invention also recognizes that such deviations can be determined by simulating a physical model and a model of the control device. A conventional simulation of this type has several disadvantages: First, the test is carried out as a black box test at the system level; It is not possible to determine whether deviations occur in the individual calculation steps in the control unit. Accordingly, it is difficult to find the cause of a deviation if it is found. Secondly, the test is only carried out for a configuration or for an operating mode of the control device. This configuration must be determined by system constants before the simulation. It is time-consuming to simulate several times for different configurations. Third, due to the large number of input variables, it is not possible to test all combinations of input variables. It is therefore desirable to specifically test critical combinations of the input variables. In the case of a black box test, however, it is not possible to predict which values at the inputs lead to large deviations at the outputs. In practice, operating conditions can occur that lead to critical deviations and were not considered in the simulation.

The aim of the method proposed below is therefore to systematically check all critical combinations of input variables for deviations between the physical model and the implementation in the control unit. System constants that represent different configurations of the software of the control unit are treated like other input variables. The test is carried out step by step, with an intermediate result available after each calculation step, which makes troubleshooting easier in the event that deviations occur.

An advantage of this solution is that the test is carried out as a whitebox test, which makes troubleshooting easier. The test is also carried out simultaneously for all conceivable configurations of the software of the control unit, since system constants are treated as normal input variables.

The measures listed in the dependent claims allow advantageous developments and improvements of the basic idea specified in the independent claim. It can be provided that the smallest possible value, the smallest possible value plus a predetermined increment, the largest possible value, the largest possible value minus the increment and the mean value plus and minus the increment are selected as operands. In this way, combinations of input variables are systematically tested that potentially lead to large deviations at the output.

Figure list

• 1 das Flussdiagramm eines Verfahrens gemäß einer ersten Ausführungsform.
• 2 schematisch einen Arbeitsplatzrechner gemäß einer zweiten Ausführungsform.

Embodiments of the invention are shown in the drawings and explained in more detail in the following description. It shows:

• 1 the flowchart of a method according to a first embodiment.
• 2nd schematically a workstation according to a second embodiment.

Embodiments of the invention

1. 1. der kleinste mögliche Wert,
2. 2. der größte mögliche Wert,
3. 3. der kleinste mögliche Wert zuzüglich eines Inkrements,
4. 4. der größte mögliche Wert abzüglich eines Inkrements,
5. 5. der Mittelwert zuzüglich eines Inkrements sowie
6. 6. der Mittelwert abzüglich eines Inkrements.

1 illustrates the basic sequence of a method according to the invention ( 10th ). As the loop structure of the flow chart shows, the calculation is carried out step by step: Each step of the calculation (process 13 ) has two input variables and one output variable. Six different values are selected as input variables (process 11 ) where large deviations can occur:

1. 1. the smallest possible value,
2. 2. the greatest possible value,
3. 3. the smallest possible value plus an increment,
4. 4. the largest possible value minus an increment,
5. 5. the mean plus an increment as well
6. 6. the mean minus an increment.

Then all possible combinations of the two input variables are formed (process 12th ) and calculated both physically (preferably with the greatest possible accuracy of the computer used) and analogously to the implementation in the control unit (e.g. as an integer or float) (process 13 ). With this calculation ( 13 ) there are 36 different output variables. The deviations of these quantities between physical and implemented calculation are determined. In the next step of the calculation ( 13 ), the previously determined output variable is offset against a new input variable. The values of the new input variable are determined as shown above. To prevent the number of necessary calculations ( 13 ) potentially increases in the ratio 6 ⁿ , where n denotes the respective iteration, the 36 initial values are reduced to six values. The six values are selected from the 36 initial values that meet or largely correspond to the criteria mentioned above.

These calculations ( 13 ) are repeated over several iterations, with the deviation being determined and output in each iteration (process 14 ). The result of the calculation ( 13 ) is a list of the deviations for each step, considering six input and 36 output values. In practice, the test described above is preferably carried out automatically by a computer program.

The program can use files that are required in the course of ECU development - for example, C source code, the proprietary PaVaSt specification or an XML model description, such as the open product family ETAS ASCET - to determine the required input variables and calculation steps. A possible alternative to the procedure described above is to reduce or increase the number of values used per input variable. In one embodiment, it would be conceivable to consider further intermediate values in addition to the minimum, maximum and mean values, without leaving the scope of the invention.

This method ( 10th ) can, for example, in software or hardware or in a mixed form of software and hardware, for example in a workstation computer ( 20 ) be implemented as the schematic representation of the 2nd clarifies.

QUOTES INCLUDE IN THE DESCRIPTION

This list of documents listed by the applicant has been generated automatically and is only included 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 102008004361 A1 [0004]

Claims (10)

Method (10) for checking the software of a control device, which implements an approximation of an inner two-digit link on a real interval, characterized by the following features: - First operands and second operands are selected from the interval (11) according to predetermined criteria, - from the Operands are formed value pairs (12), - for each of the value pairs, a function value is calculated using the link and an approximation value is calculated using the approximation (13), - a deviation of the approximation value from the function value is output (14), - according to the criteria, the first Operands among the approximate values and the second operands from the interval are newly selected (11) and - forming the value pairs (12), calculating (13) the function and approximate values, outputting (14) the deviation and selecting (11) the operands repeated several times. Method (10) according to Claim 1 , characterized by the following feature: - the criteria are specified in such a way that the operands each comprise the smallest possible value. Method (10) according to Claim 2 , characterized by the following feature: - the criteria are specified such that the operands each comprise the smallest possible value plus a specified increment. Method (10) according to Claim 3 , characterized by the following feature: - the criteria are specified in such a way that the operands each comprise the largest possible value. Method (10) according to Claim 4 , characterized by the following feature: - the criteria are specified in such a way that the operands each include the largest possible value minus the increment. Method (10) according to one of the Claims 3 to 5 , characterized by the following feature: - the criteria are specified in such a way that the operands each include an average plus the increment. Method (10) according to Claim 6 , characterized by the following feature: - the criteria are predefined in such a way that the operands each include an average minus the increment. Computer program which is set up, the method (10) according to one of the Claims 1 to 7 to execute. Machine-readable storage medium on which the computer program is based Claim 8 is saved. Device (20) which is set up to process (10) according to one of the Claims 1 to 7 to execute.

Images