Patent Application
20190265128
The disclosure relates to a method for determining support points of a test plan for measuring or simulating pre-defined test variables of a test machine.
Test plans are used in the measuring of certain characteristics of test machines, for example internal combustion engines, transmissions, or entire vehicles, but test plans are also used in the measuring of certain characteristics of electric motors, generators and comparable machines. With the aid of the test plans, test values of the pre-defined test variables are meant to be measured and detected in a large variation range of machine parameters, which influence the test variables, in as short a time as possible. In addition, test plans permit the repetition of a test, in order to, for example, review and validate the already-determined test results. Test plans can also be used to establish characteristics of the test machines, through a simulation, on the basis of a mathematical model of the test machine.
Any parameters of the test machine come into consideration as test variables. These parameters can either be directly detected by measurement or can be established on the basis of metrologically detected variables via mathematical models. If, for example, a motor vehicle is to be measured, the test variable can be the fuel consumption of the motor vehicle or, for example, the damage level of different components of the motor vehicle, as well.
The test plans used for measuring comprise a sequence of support points for each machine parameter of the test machine to be varied in the test. The test machine is controlled or regulated with the aid of these support points such that the different operating parameters, at time points linked with the support points, assume machine parameter values likewise linked with the support points. Each support point thus contains a time point information, as well as a machine parameter value. The support points can, for example, relate to speeds of a motor vehicle which this vehicle is to reach at certain time points. The sequence of the support points here forms a speed profile to be driven by the motor vehicle, or a route, provided that the transverse dynamics of the motor vehicle is taken into account.
Different standardized test plans are known, in particular for determining the fuel consumption of motor vehicles. A frequently-used test plan is the so-called NEDC Cycle.
Numerous methods to establish test plans are known from the prior art. The methods used are hereby usually selected according to the purpose intended with the measurement. For example, in the measuring of dynamic characteristics of the test machine, test plans are usually used, which fundamentally differ from such test plans with which the test variables are to be measured under stationary operating conditions. Known methods for creating test plans are, for example, summarized under the headings “design of experiments” and “statistical test planning”.
In addition, it is desirable in several areas to design test plans such that, through the test plans, a real usage profile of a fielded machine is obtained. This is desirable, for example in establishing the fuel consumption of a motor vehicle, as the known standardized driving cycles or test plans for the determining of the fuel consumption, such as the NEDC are based on rather unrealistic speed profiles. In addition, no multidimensionality, such as for example the simultaneous consideration of the speed and the motor temperature, is taken into consideration. To that end, no methods are known from the prior art, which permit a simple, quickly adaptable creation of an, in particular multidimensional test plan, which achieves an actual use of the fielded machine in the test machine.
It is therefore seen as the object of the invention to provide a method for determining support points of a test plan, wherein the test plan, if possible, describes a real usage profile, but can be measured in a substantially shorter time period than in the actual usage.
This object is achieved by a method for determining support points of a test plan for measuring or simulating pre-defined test variables of a test machine on the basis of previously metrologically established operating values of operating variables of at least one fielded machine during its intended use, wherein, in an aggregation step, the detected operating values are assigned, according to a pre-defined classification rule with respect to one or multiple selected operating variables, to classes, wherein, in a specification step, prior to or after the aggregation step, specification variables are selected, wherein the specification variables form at least one subset of the operating variables, wherein, in an identification step following the aggregation step, the frequency of operating classes for each class is determined, and wherein, in a subsequent determination step, the support points of the test plan are determined on the basis of the operating class frequency, wherein the support points are determined in the identification step such that a deviation of a relative test class frequency, established on the basis of the test plan, and specification values of the specification variables assigned to classes according to the classification rule, is minimized, by a relative operating class frequency of the operating values classified according to the classification rule of the operating variables, corresponding to the specification variables, according to a pre-defined optimization criterion. In this way, it is possible, for example, to establish the fuel consumption of a motor vehicle on the basis of realistic test plans. Additionally, for example also regional differences in the usage of motor vehicles and different usage profiles of buyer groups of established motor vehicles can, in this way, be taken into account, in turn, with respect to the establishing of fuel consumption. The use of test plans which map the actual usage of the fielded machine, can, for example, however also be used for the design of components of new machines, since, in this way, an entire life cycle of the new components installed in the test machine can be established on the basis of real usage profiles.
The test machine can for example be a newly developed motor vehicle. The at least one fielded machine relates, for example, to a series-produced motor vehicle used in normal traffic and comparable with the test machine. The test variable can for example be a fuel consumption related to a 100 km drive. The operating variables are at least the operating variables of the motor vehicle significantly determining the fuel consumption, for example the vehicle speed, the vehicle acceleration, the respectively selected gears, etc. The operating values of these operating variables are advantageously continuously detected during the use of the motor vehicle, and/or are established on the basis of measurement variables metrologically detected and processed with the help of mathematical models.
The operating values, directly metrologically determined or established with the aid of mathematical models on the basis of measurement values are classified with respect to one or multiple selected operating variables. For example, the operating values of the operating variables used for measuring the fuel consumption of a motor vehicle, e.g. actually driven speeds, are assigned to individual classes, such as for example, 0-30 km/h, 31-50 km/h, 51-100 km/h, and >100 km/h, and in the establishing step, the frequency for each of the classes with which the respective speed range was previously detected is determined.
On the basis of these frequency values, the test plan is subsequently or the support points of the test plan are subsequently determined such that the relative test class frequency of specification values identified on the basis of the test plan and assigned to classes according to the classification rule, of the specification variables correspond, as much as possible, to the respective relative operating class frequency. In this way, a test plan can be established, with which speeds are driven, which fall into the class 30-51 km/h and into the class 0-30 km/h proportionally as often as the actually-established operating values. The test class frequency is advantageously determined with the aid of a mathematical model of the test machine. It is, however, also possible and provided to establish the test class frequency through the execution of tests with the test machine.
Insofar as the test plan is established on the basis of the operating values of merely one fielded machine, the fuel consumption of the test machine can, for example, advantageously be established on an individualized customer basis, and the driving behavior of one of each customer can be taken into consideration. Insofar as the goal of the measurement, however, is the establishing of the test variables for a common usage of the fielded machine, independent of individual fielded machines, it is advantageously provided that the operating values are detected in a plurality of fielded machines.
The test plan determined with the aid of the disclosed method can advantageously also be used to optimize operating strategies of a test machine with the aid of representative user profiles and to optimally parameterize the components of a test machine.
In particular insofar as the specification variables are already known before the determination of the operating variables, it is advantageously provided that the specification variables correspond to the selected operating variables. In this way, the effort necessary for the determination of the operating values can be reduced, as merely operating values must be established for operating variables corresponding to the specification variables.
It is alternatively provided that all relevant operating values are continuously determined in the fielded machines and are transmitted once or in regular intervals to a central database. In this way, the specification variables can, at a later point in time, be flexibly selected from the detected operating variables, and thus can be adapted to the selection and thus to the respective requirements.
It is also possible and provided that, in place of the transmission of the operating values, exclusively the frequency values of already classified operating values are transmitted to the central database. In this way, the data quantity to be transmitted can be significantly reduced. The aggregation step here already occurs, for example, on a control device of the various fielded machines. In this way, the private sphere of the person using the fielded machine is also protected, since exclusively aggregated data are stored and further used. It is also possible and provided that the entire method is carried out on a control device of a fielded machine.
Advantageously, it is provided that the operating values are overwritten and/or deleted directly after the aggregation step. The private sphere of the persons using the fielded machine can be particularly well protected in that the operating values are not stored.
In order to adapt the test plan determined with the disclosed method as well as possible to the actual use profile, it is provided for that, in the aggregation step, transition frequencies are determined for one or multiple classified operating variables, wherein a transition frequency is the number of the changes of an operating variable or multiple operating variables of a class into a different class, and wherein, in the determination step, the support points of the test plan are determined on the basis of the operating class frequency and the transition frequencies, wherein the support points are established in the identification step such that a deviation of a relative test class frequency of the specification values identified on the basis of the test plan and assigned to classes according to the classification rules, from a relative operating class frequency of the operating values classified according to the classification rule, of the operating variables corresponding to the specification variables, as well as a deviation of test transition frequencies for the specification variables from the transition frequencies for the operating variables corresponding to the specification variables, is minimized according to the pre-defined optimization criterion, wherein a test transition frequency is the number of changes of a specification variable or multiple specification variables from one class into another class. In this way, the influence of the frequency of the operating point changes on the test variables can be taken into account in the determination of the test plan.
Through the use of the transition frequencies, how often and between which classes the driven speed changes, can, in the test plan exemplarily described above, also be taken into account in addition to the frequency of driven speeds. In this way, how long the operating variables are usually held in the respective classes, before an operating value, new and assigned to another class is driven, is additionally also considered. In addition, it is also possible by considering multiple operating variables, to take into account transition frequencies for the changing of multiple operating variables of a class or a class combination into a different one. A class combination is the classes of the operating variables taken into account, which are assigned the operating values in the respective state.
It is also possible and provided that, in the aggregation step, transition possibilities for one or multiple classified operating variables are determined, wherein a transition possibility is a transition, the transition frequency of which is greater than zero, wherein a transition frequency is the number of changes of an operating variable or multiple operation variables from one class into another class, wherein, in the determination step, the support points of the test plan are identified on the basis of the operating class frequency and the transition possibilities, wherein the support points are established in the determining step such that a deviation of a relative test class frequency of specification values identified on the basis of the test plan and assigned to classes according to the classification rule, of the operating variables, from a relative operating class frequency of the operating values classified according to the classification rule, of the operating variables corresponding to the specification variables, is minimized according to the pre-defined optimization criterion, and in that the test plan exclusively comprises such test transitions that are covered by the transition possibilities, wherein the test transitions are all changes of a specification variable or multiple specification variables from one class into another class. In this way, a test plan can be created on the basis of the operating class frequencies, which plan comprises transitions that are exclusively present in the detected operating variables and therefore are physically possible.
To establish the support points of the test plan, it is advantageously provided that the support points of the test plan are determined, in the determination step, with the aid of a Markov chain process, on the basis of the operating class frequencies and the transition frequencies. Classes can be selected, with the aid of Markov chain processes, on the basis of the determined frequencies, in which classes the specification variables of the next support point of the test plan lie, wherein the test plan, resulting with the aid of the Markov chain process, comprises test class frequencies and test transition frequencies, which, in a sufficiently-long test plan or in a test plan with a sufficient number of support points, can, very strongly, approximate the operating class frequencies or in particular the transition frequencies, or can correspond to these.
In order to be able to simply determine test plans, for example for different test requirements, it is provided that support points of an initial test plan are determined, in the determination step, with the aid of a Markov chain process, on the basis of the operating class frequencies and the transition frequencies. The initial test plan is determined in the same manner and fashion, with the aid of the Markov chain process, as the previously-described test plan. However, the initial test plan comprises especially many support points, and serves as the basis for the following production of one or multiple test plans. Advantageously, the initial test plan comprises 3 to 4 times as many support points as the test plan to be determined on the basis of initial test plans, and particularly advantageously, 10 times as many support places.
To establish different test plans from the initial test plan, it is provided that, in the identification step subsequent to the determination of the initial test plan, the initial test plan is subdivided into initial test plan segments with the aid of a segmentation process. The initial test plan segments can then, with the aid of a suitable method, advantageously a random process, be combined to one or multiple new test plans.
Advantageously, it is provided for that for the segmentation, for at least one initial specification variable of the initial test plan, respectively at least one status value, is specified, and that the initial test plan segments are generated such that each first and last initial specification value of each initial specification variable has the respective status value, for which at least one status value was pre-defined. In this way, it can, inter alia, be achieved that the test plans constituted from the initial test plan segments have a continuous signal curve and no undesired transitions and in particular no transitions which are not included in the transition possibilities, between successive support points, on the places, at which different initial test plan segments were combined or composed with one another.
Advantageously, it is provided for that, respectively at least one status value is pre-defined for the segmentation, for at least two initial specification variables of the initial test plan, and that the initial test plan segments are generated such that each first and last initial specification value comprises the respective status value of each initial specification variable, for which at least one status value was pre-defined. In this way and manner, multidimensional test plans can be composed with the disclosed method in a simple manner. Through such a segmentation, multidimensional initial test plan segments can be generated, wherein a continuous transition between the initial test plan segments, in the test plans generated on the basis of these initial test plan segments, is made possible. For example, the driven speed and the acceleration of the fielded machines are used, as operating variables, for the determination of a test plan for the measurement of a motor vehicle, and the respective operating values are assigned to classes in the aggregation step, as well as the transition frequencies are determined. Subsequently, an initial test plan is created with the aid of the Markov chain process. This initial test plan is then segmented with the aid of the segmentation process, wherein the initial test plan segments are generated such that each first and last initial test value of each initial specification variable, for which at least one status value was pre-defined, comprises the respective status value. For example, for the segmentation, status values are pre-defined for the acceleration as well as also for the speed. A segmentation can here result, for example, if, in the initial test plan, the speed, as well as the acceleration comprises one of the specified combinations of the status values for the speed and the acceleration. For example, the initial test plan is segmented at the places at which the speed and the acceleration is zero and is segmented at the places at which the speed amounts to 10 km/h and the acceleration is zero.
The combinations of the status values for the segmentation are advantageously pre-defined taking into account the operating class frequencies, wherein the combinations of the status values are advantageously selected from the classes, which comprise comparatively large operating class frequencies. In this way, numerous initial test plan segments can be composed.
Advantageously, it is provided for that multidimensional test plans are created with the aid of the disclosed method. The segmentation with the segmentation process therefore occurs likewise in multiple dimensions or regarding multiple specification variables.
In order to determine one or multiple test plans on the basis of the determined initial test plan segments, it is provided for that, subsequently, the initial test plan segments are to be composed together, using a random process, to the test plan such that the pre-defined optimization criterion is minimized.
In a particularly advantageous configuration of the disclosed method, it is provided that numerous candidate test plans are subsequently composed from the initial test plan segments with at least one random process and that subsequently, the candidate test plans are evaluated with respect to the optimization criterion and the candidate test plan which minimizes the optimization criterion is selected as test plan. In this way, suitable candidate test plans can be created on the basis of the present initial test plan segments with respect to different test plan requirements and the best, that is the test plan(s) possibly minimizing the optimization criterion, is/are selected out of these candidate test plans.
To determine the support points of the test plan, it is advantageously provided that, in an old test plan evaluation step, old test operating values of old test operating variables are determined on the basis of already present old test plans, wherein the old test operating values form the operating values and, according to the classification rule, are assigned to classes, and wherein the support points are determined from portions of the old test plans in the determination step, wherein the portions are selected such that a deviation of a relative test class frequency of the test values determined on the basis of the test plan and assigned to classes according to the classification rule, from a relative operating class frequency of the old test operating values classified according to the classification rule, is minimized according to the specified optimization criterion. In this way, the new test plan can be composed from portions of one or multiple old test plans.
Advantageously, it is provided for that the determination of the portions of the old test plans occur with an as described-above segmentation process.
Advantageously, it is provided for that the old test operating values are determined at an old test machine comparable to the test machine. In this way, it can be achieved that the machine parameter values assigned to the support points can actually be achieved and driven by the test machine.
Advantageously, it is provided that the old test operating values are determined metrologically. Advantageously, recordings of old test operating variables detected in actual measurements are used for this purpose.
It is, however, also possible and provided for that the old test operating values are determined through a mathematical model of the test machine or the old test machine. In this way, the effort to determine the old test operating values can be significantly reduced and, also, old test plans can be used to establish the support points of the test plan, for which no actually-determined operating variables are given.
Advantageously, it is provided for that the test values are determined through a mathematical model of the test machine. Advantageously, operating parameters of the test machine, metrologically-detected, in the measurement, with the test plan, are processed in the mathematical model.
Further advantageous configurations of the method are explained in greater detail by means of exemplary embodiments illustrated in the drawings.
Subsequently, the operating class frequency is determined in a determination step 6 for each class 4, into which the operating values 3 were previously categorized. On the basis of the thus-determined operating class frequencies 7, a test plan 9 is established in an identification step 8, with the aid of a suitable optimization process.
Subsequently, the detected operating values 3 are assigned, according to a pre-defined classification rule, to classes 4, with respect to speed 12 and acceleration 13. The assignment, occurring in each case separately for the fielded machines 11 is subsequently brought together, and the thus-determined operation class frequencies 7 are stored in a database 14. How long the fielded machines 11 have respectively driven the classified speed and acceleration combinations can be taken from the operating class frequencies 7.
It is alternatively also possible to identify the support points, as well as the number of the support points, using a suitable optimization process, starting from zero.
In an aggregation step, operating values of operating variables 10 of multiple field machines were initially metrologically detected. The operating variables 10 relate to the speed v and acceleration a of multiple motor vehicles.
Subsequently, the detected operating values were assigned to classes 4 according to a predefined classification rule with respect to the speed v and the acceleration a. In the illustration, individual classes 4 are denoted by a reference character.
In addition, multiple transition frequencies 24 were identified for each class 4. The drawing, by way of example, illustrates transition frequencies T of a class H31,5. Depending on a change in acceleration Δa and a change in speed Δv, the transition frequencies T indicate how often the operating values, based upon class H31,5, change in accordance with the changes in acceleration Δa, and the change in speed Δv.
Initial test plan segments 27 are to be identified on the basis of the initial test plan 25. Status values 28 of the initial specification variables 26 were predefined for the segmentation. The segmentation is in each case to be effected at an acceleration a=0 m/s2 and a speed of either v=0 km/h or v=8 km/h. At these places (a=0 m/s2 and v=0 km/h; a=0 m/s2 and v=8 km/h), the initial test plan 25 is respectively sub-divided. Support points 16 of the initial test plan 25, which are between two subdivision points 29, together form an initial test plan segment 27.
| Number | Date | Country | Kind |
|---|---|---|---|
| 10 2016 120 052.2 | Oct 2016 | DE | national |
| Filing Document | Filing Date | Country | Kind |
|---|---|---|---|
| PCT/EP2017/076562 | 10/18/2017 | WO | 00 |
