METHOD FOR DETERMINING CONTROL MODEL PARAMETERS OF A CONTROL MODEL OF AN AXIAL PISTON PUMP

Description

[0001] The present invention relates to a method for determining control model parameters of a control model of an axial piston pump, as well as a computing unit and a computer program for carrying it out. Background of the invention

[0002] A hydrostatic drive system typically consists of a primary variable displacement pump and a secondary variable displacement motor. The primary and secondary units can be configured in either a closed or open circuit. During operation, the primary and secondary units are adjusted either separately or in tandem. This results in a rotational speed on the secondary side that is proportional to the flow rate. The operating pressure adjusts according to the load torque and is usually limited by a pressure relief valve.

[0003] In axial piston pumps with a swashplate design, the delivery flow rate is adjusted by changing the swashplate angle. This adjustment is achieved, for example, via a hydraulic cylinder. The cylinder's chamber pressure can be regulated by a pressure control valve. The pump exhibits load-sensitive behavior, so the swashplate is swashplate- ...

[0004] Patents DE 10 2024 222638 A1, DE 10 2017 221637 A1 and WO 01 / 36828 A1 are known from the prior art. Disclosure of the invention

[0005] According to the invention, a method for determining control model parameters of a control model of an axial piston pump, as well as a computing unit and a computer program for carrying out this method, are proposed, comprising the features of the independent claims. Advantageous embodiments are the subject of the dependent claims and the following description.

[0006] The control model models at least one control variable of the adjustment unit as a function of operating variables, where operating points of the axial piston pump are each characterized (or defined) by specific operating values ​​of the operating variables. The set of operating points defines an n-dimensional space, where it is not a priori clear at which operating points in this space the axial piston pump can be operated safely or without faults, i.e., without the operation being interrupted due to a fault (such as overpressure) and without damage occurring in extreme cases. Thus, it is not known from the outset how exactly the space is divided into safe and unsafe operating points. This is because variations in production within tolerances can lead to different operating behaviors.

[0007] The invention employs a safety model in addition to the control model. This safety model is a model for the safety of operating points, meaning it allows a statement about whether or not safe operation is (probably) possible at a given operating point. Starting with several known safe operating points, further operating points are selected such that, on the one hand, the variance of the control model is maximized, and on the other hand, the probability that the further operating point is a safe operating point exceeds a predefined safety threshold. The models (control model, safety model) are then adjusted based on the further operating point.By repeatedly performing this procedure, the area of ​​known safe operating points in the space of operating points can be gradually expanded, whereby the risk of the axial piston pump being controlled with control variables that lead to errors is very small.

[0008] Specifically, the control model parameters of an axial piston pump control model are determined. The delivery volume is changed by means of an adjustable swashplate by altering a swivel angle. An adjusting force from an adjustment unit of the axial piston pump, dependent on a control pressure, acts on the swashplate to achieve this adjustment. Likewise, a pressure force resulting from the operating pressure of the axial piston pump also acts on the swashplate. The swivel angle results from these forces as a function of a rotational speed.

[0009] Operating parameters (typically measurable quantities) can include, for example: the rotational speed of the axial piston pump, the swashplate angle, or the working pressure, i.e., the pressure differential between the pressures in the working lines connected to the axial piston pump, between which the pump delivers the hydraulic fluid. Similarly, a measurable quantity of elements connected to the axial piston pump can be used; for example, the rotational speed of a hydraulic motor connected to the working lines. Additionally, first and / or second derivatives of these aforementioned operating parameters can also be used as operating parameters.

[0010] The method comprises providing initial control model parameters of the control model and an operating state set comprising several initial first operating points; and providing initial safety model parameters of a safety model based on a safety set set comprising several initial second operating points and associated safety values, wherein the safety model models a safety quantity as a function of the operating quantities, which enables a statement as to whether fault-free operation is possible or not at the respective operating point.

[0011] In these two steps, initial models and their model parameters are provided, from which the subsequent process steps are carried out. The initial control model parameters and the initial operating state quantity form a starting point considered safe for the further process, which is determined, for example, by means of a physical simulation of the axial piston pump. Likewise, the initial safety model parameters and the initial safety set quantity allow for an assessment of safety when subsequently selecting further operating points. The quantities of the initial first operating points and the initial second operating points are independent of each other; that is, the initial second operating points do not have to be (but can be, at least partially) identical to the initial first operating points. The initial first operating points should be those at which fault-free operation is possible.This may not be the case for the initial second operating points.

[0012] Furthermore, the procedure includes one or more instances of determining or selecting an additional operating point not yet included in the operating state set, whereby the additional operating point is determined in such a way as to maximize a variance of the control model, provided that a probability obtained by the safety model that fault-free operation is not possible at the additional operating point is less than or equal to a predefined safety threshold; updating the operating state set and the control model parameters of the control model taking the additional operating point into account; and updating the safety set set and the safety model parameters of the safety model taking the additional operating point into account.

[0013] In other words, the additional operating point is selected from the set of all operating points not yet included in the operating state set, where the probability, as determined by the safety model, that fault-free operation is not possible at the additional operating point is less than or equal to a predefined safety threshold. The selection from this set is made in such a way as to maximize the variance of the control model. By selecting an additional operating point that maximizes the variance, operating parameter ranges can be explored where there is still a large degree of uncertainty about the behavior of the axial piston pump.

[0014] The variance of the control model can be determined from the uncertainty of the prediction model for the operating points. If Gaussian processes are used, the prediction yields a mean and a variance.

[0015] The probability that fault-free operation is possible or impossible for a given operating point is obtained by applying the safety model (conceived as a mapping) to the respective operating point and comparing the resulting estimated safety value, possibly taking into account the distribution of a fault term, with the safety threshold. The distribution of a fault term can also be determined within the model, for example, in regression models.

[0016] The (actual) safety value for each subsequent operating point can be determined depending on the operating parameters and / or other parameters, which can, in particular, be measured. For example, the working pressure Δp can be used. Similarly, a pressure p₁, p₂ in a working line can be used, where the working pressure is the difference between the pressures in the two working lines: Δp = p₁ - p₂. A (partial) safety value S can then be calculated, for example, as: 1 - P / P₁MAX; where P represents the working pressure Δp or a pressure P₁, P₂ in the working lines, and P₁MAX denotes a corresponding maximum pressure (e.g., determined by a pressure relief valve). Similarly, a (partial) safety value can be determined based on the swivel angle relative to a maximum possible swivel angle.For the axial piston pump, a maximum speed, which should not be exceeded, can be specified, even if a fixed speed limit exists. Preferably, combinations, such as suitable sums or averages, of these and possibly other partial safety values ​​are used as the (overall) safety value. Corresponding safety values ​​for the initial second operating points can be determined in the same way. Alternatively or additionally, expert knowledge can be incorporated, allowing a person skilled in the art to directly define the safety model or its safety model parameters for specific operating ranges or operating points.

[0017] Preferably, the control model and / or the safety model are regression models, wherein the operating variables are independent variables and at least one control variable is at least one dependent variable; preferably Gaussian process models are used as regression models.

[0018] The models can therefore be formulated as regression models as follows: Control model: Yj = fj(Xi) + εj; Safety model: S = g(Xi) + τ; where Yj represents the at least one control variable, Xi represents the operating variables, S represents the safety variable, and εj and τ represent the respective error terms, which are assumed to be normally distributed. Within the framework of the regression, the mappings fj and g, which are characterized by the control model parameters and the safety model parameters respectively, are determined, and the distribution of the error terms can also be determined. Corresponding regression methods are known to those skilled in the art. For example, the method of least squares can be applied, i.e., a functional is minimized that is determined by the squareDeviations between measured control values / safety values ​​and function values, obtained by applying the functions fj and g to the corresponding operating values ​​(the operating state quantity and the safety set quantity, respectively), depend on this. Similarly, a so-called maximum likelihood estimation can be used. In this case, the control model parameters and the safety model parameters are determined such that the probability that the respective model yields the control values / safety values ​​at the operating points (the operating state quantity and the safety set quantity) is maximized, taking into account the distribution of the respective error terms.

[0019] In principle, the use of other models (besides regression models) is also possible, such as the use of neural networks, provided that these models are able to determine a variance or uncertainty estimate.

[0020] Preferably, updating the operating state quantity and the control model parameters of the control model, taking into account the additional operating point, comprises determining or measuring at least one additional control value of the at least one control variable such that the additional operating point is reached; adding the additional operating point and the determined additional control value to the operating state quantity to form a supplemented operating state quantity; and re-determining the control model parameters of the control model based on the supplemented operating state quantity. The control value is determined by operating the axial piston pump (e.g., in a test bench), adjusting the control variables so that the additional operating point is reached. The re-determination of the control model parameters is performed according to the method used for the modeling (e.g., regression method).

[0021] Preferably, updating the safety set quantity and the safety model parameters of the safety model, taking into account the additional operating point, comprises determining a safety value for the additional operating point; adding the additional operating point and the associated determined safety value to the safety set quantity to form a supplemented safety set quantity; and re-determining the safety model parameters of the safety model based on the supplemented safety set quantity. The determination of the safety value also takes place during operation of the axial piston pump (preferably during the operation described in the aforementioned embodiment). For this purpose, the operating parameters and / or other parameters on which the safety value is calculated are measured. The re-determination of the safety model parameters is also carried out here according to the method used for the modeling (e.g.,regression procedure).

[0022] Preferably, providing initial control model parameters and an operating state set includes determining the initial first operating points by means of a physical model of the axial piston pump, wherein the first operating points are determined in the physical model such that fault-free operation is possible; and / or providing the initial first operating points using standard operating points known to enable fault-free operation. The initial first operating points can be obtained by physical modeling and / or from standard operating points. In either case, they should be chosen such that fault-free operation is possible at the first operating points. The standard operating points can be determined, for example, using a reference axial piston pump or a prototype axial piston pump.Both in physical modeling and in the use of standard operating points, operating points can be obtained that run flawlessly for all axial piston pumps within a series of axial piston pumps, i.e., within the series variation, these operating points can be controlled flawlessly in every case.

[0023] Preferably, providing initial control model parameters and an operating state set further includes determining the control values ​​associated with the first operating points by means of physical modeling or by operating the axial piston pump and adjusting the control variables so that the respective first operating point is reached; and determining the control model parameters based on the determined first operating points and the determined associated control values. Alternatively, the use of standard control model parameters, determined, for example, with a reference axial piston pump or a prototype axial piston pump, would also be possible.

[0024] Preferably, when providing initial safety model parameters, the initial second operating points of a safety model include both safe operating points, where error-free operation is possible, and unsafe operating points, where error-free operation is not possible; preferably, the safe operating points include at least some of the initial first operating points. Unsafe operating points, where error-free operation is not possible, can be operating points where it is clear that errors will occur. For example, when a pressure P used as an operating variable reaches or exceeds a maximum possible pressure PMAX (e.g., determined by a pressure relief valve). This configuration is advantageous if the safety model is a classification model. If it is a regression model, this is not strictly necessary.

[0025] Preferably, the number of initial first operating points and the number of initial second operating points are independent of each other and range from 10 to 100. Equally preferably, the safety threshold is less than or equal to 5%; where the safety threshold is preferably 2%, more preferably 1%, even more preferably 0.5%, and most preferably 0.1%. These numbers for the first / second operating points and these safety thresholds enable the (relatively) reliable determination of further operating points. In general, a different number of initial first operating points and / or a different number of initial second operating points can also be chosen. This is particularly model-dependent. A smaller safety threshold means higher safety; the choice accordingly depends on the specific risk tolerance.The greater the risk, the faster the exploration of the area of ​​operating points takes place, but this is associated with a higher risk of damage.

[0026] Preferably, determining an additional operating point, updating the set of operating states and the control model parameters, and updating the set of safety data and the safety model parameters are performed multiple times until a termination condition is met. Preferably, the termination condition includes one or more of the following: the variance of the control model is below a predetermined threshold, the change in the control model parameters during the update is below a predetermined threshold, the change in the safety model parameters during the update is below a predetermined threshold, or the distance of the additional operating point to the set of other operating points in the set of operating states is below a predetermined threshold.The termination conditions are therefore chosen in principle such that, if they are met, no significant improvement in the control model is to be expected. A suitable metric can be used on the space of operating variables to determine the distance between operating points. Since the different operating variables have different units and different value ranges, scaling can be performed first. For example, the operating values ​​can be scaled so that values ​​lying between minimum and maximum operating values ​​(or other suitably defined lower / upper operating values) are mapped to the interval between -1 and +1 (for operating variables that can take on positive and negative values, e.g., swivel angle) or to the interval between 0 and +1 (for operating variables that can only take on positive values, e.g., operating pressure). After scaling, for example...The Euclidean metric is used. Scaling can be linear or non-linear; the latter allows certain operational parameter ranges to be weighted more or less heavily.

[0027] Preferably, the at least one control variable comprises at least one control current for at least one pressure regulating valve of the adjustment unit, which is particularly electromagnetically actuated.

[0028] A computing unit according to the invention, e.g. a control unit of a test bench, is, in particular in terms of programming, equipped to carry out a method according to the invention.

[0029] Implementing a method according to the invention in the form of a computer program or computer program product with program code for carrying out all method steps is also advantageous, as this incurs particularly low costs, especially if an executing control unit is already available for other tasks. Suitable data carriers for providing the computer program are, in particular, magnetic, optical, and electrical storage media, such as hard drives, flash memory, EEPROMs, DVDs, etc. Downloading a program via computer networks (Internet, intranet, etc.) is also possible.

[0030] Further advantages and embodiments of the invention will become apparent from the description and the accompanying drawing.

[0031] It is understood that the features mentioned above and those to be explained below can be used not only in the combinations specified, but also in other combinations or on their own, without leaving the scope of the present invention.

[0032] The invention is schematically illustrated in the drawing using exemplary embodiments and is described in detail below with reference to the drawing. Character description

[0033] Figure 1 Figure 1 shows a schematic representation of a hydrostatic drive system with an axial piston pump, for which the method according to the invention is provided. Figure 2 shows a hydraulic circuit diagram of the axial piston pump of the drive system according to Figure 1 . Figure 3 shows a flowchart according to a preferred embodiment of the invention. Figure 4 illustrates safe and unsafe operating size ranges. Detailed description of the drawing

[0034] According to Figure 1A drive unit 1 has a primary variable displacement pump 2 and a secondary variable displacement motor 4 connected in series. The former is driven, for example, by an internal combustion engine (ICE). The primary unit 2 converts mechanical energy into hydraulic energy, while the secondary unit 4 converts hydraulic energy into mechanical energy on the output side. The process can also be reversed, so that the secondary unit 4 provides braking on the output side 2. The connection between the primary unit 2 and the secondary unit 4 can be either an open circuit, meaning the low-pressure sides of the primary unit 2 and the secondary unit 4 are connected to a pressure-balanced tank, or a closed circuit, meaning the low-pressure sides of the primary unit 2 and the secondary unit 4 are directly connected to each other. Both configurations are protected against excessive pressure by pressure relief valves.To increase the efficiency of the drive train, a power split can be used, in which a mechanical power path is installed in parallel to the hydrostatic section 2, 4. For operation, the primary unit 2 and secondary unit 4 are adjusted either separately or in conjunction. This results in a rotational speed on the secondary side that is proportional to the flow rate. The pressure adjusts according to the load torque and is limited at the upper end by the pressure relief valve.

[0035] According to Figure 2The axial piston pump 2 is designed as a swashplate pump, with its delivery volume flow in the working lines 6, 8 being adjusted by changing the swivel angle of its swashplate 10. This adjustment is achieved via a mechanical coupling of the swashplate 10 to, for example, a double-acting hydraulic cylinder 12 of an adjustment unit 13. Both actuating chambers of the hydraulic cylinder 12 can be individually pressurized with actuating fluid (e.g., hydraulic oil). The respective actuating pressure in the actuating chambers is set via pressure regulating valves 14, 16 of the adjustment unit 13. The pressure regulating valves 14, 16 are, for example, designed as electromagnetic valves and can be controlled by electrical currents, namely control currents IA, IB.

[0036] The axial piston pump 2 exhibits load-sensitive behavior, meaning that it pivots back when a high applied operating pressure p or Δp ​​is applied. If the pump needs to be kept extended despite high pressures, the pressure in the adjustment mechanism must be increased. This characteristic steady-state behavior, which depends not only on the differential pressure but also on the rotational speed and the swivel angle itself, is conventionally calculated in advance for targeted adjustment or can be measured on a component test bench.

[0037] Figure 3Figure 110 shows a flowchart according to a preferred embodiment of the invention. In step 110, initial control model parameters and an operating state set comprising several initial first operating points are provided. These initial control model parameters and initial first operating points can be determined, for example, by means of physical modeling (i.e., simulation) of the axial piston pump. It is also conceivable to use standard parameters applicable to all axial piston pumps (chosen conservatively with regard to safety).

[0038] In step 120, which can also be performed before step 110 or at least partially concurrently with it, initial safety model parameters are provided based on a safety set comprising several initial second operating points and associated safety values. As already explained, the safety model models a safety parameter as a function of the operating parameters, which allows a statement to be made as to whether fault-free operation is possible at the respective operating point or not.

[0039] In step 130, another operating point, not yet included in the set of operating states, is determined. This is done by maximizing the variance of the control model, provided that the probability, obtained from the safety model, that fault-free operation is not possible at the additional operating point is less than or equal to a predefined safety threshold.

[0040] Subsequently, the operating state quantity and the control model parameters of the control model are updated (140), and the safety record quantity and the safety model parameters of the safety model are updated (145). This is done in each case taking the next operating point into account.

[0041] Specifically, updating the operating state quantity and the control model parameters can include sub-steps 150, 160, and 170.

[0042] In step 150, at least one further control value of at least one control variable is determined so that the next operating point is reached. This is achieved by operating the axial piston pump (e.g., in a test bench) and controlling it by adjusting the control variables until the next operating point is reached. For example, one can start from an adjacent operating point and change the at least one control variable until the next operating point is reached. This is relatively easy because the (physical) effect of changing a control variable on the operating variables is known in principle, and therefore essentially only a kind of fine-tuning is required.

[0043] In step 160, the additional operating point and the determined additional control value are added to the operating state set to form a supplemented operating state set.

[0044] In step 170, the control model parameters are recalculated based on the added set of operating states. That is, the parameters (control model parameters) of the control model are determined in such a way that the operating points are modeled as accurately as possible when applied. A regression method is preferably used for this purpose.

[0045] Similarly, updating the security set quantity and security model parameters (145) can include sub-steps 155, 165, and 175.

[0046] In step 155, a safety value for the next operating point is determined. For this purpose, parameters, particularly operating parameters, can be determined and / or measured during operation of the axial piston pump (e.g., in a test rig; see step 150) that can be used to determine the safety value. For example, a determination based on the relative distance of these parameters to corresponding extreme values ​​(e.g., maximum pressure) could be used.

[0047] In step 165, the additional operating point and the associated safety value determined in step 155 are added to the safety set quantity to form a supplemented safety set quantity.

[0048] In step 175, the safety model parameters are recalculated based on the added safety set. A regression method is preferably used, whereby the safety values ​​included in the safety set are again to be modeled as accurately as possible by applying the model to the corresponding operating points.

[0049] In step 180, a decision is made as to whether the procedure of steps 130, 140, and 145 should be carried out for additional operating points. If so, the process returns to step 130 (determining an additional operating point). Thus, a loop is created in which the set of safe operating points is incrementally expanded by one operating point with each iteration.

[0050] If, on the other hand, steps 130, 140, and 145 are not to be performed for additional operating points, for example, because a termination condition is met, the procedure can be terminated in step 190, and the control model can be used in the actual operation of the axial piston pump (i.e., in the operation for which the axial piston pump is intended). Further optimization of the control model is then possible, possibly taking the safety model into account.

[0051] Possible termination conditions include, for example, that a predetermined number of loop iterations have occurred, that the variance of the control model is below a predetermined threshold, that the change in the control model parameters (in step 170) is below a predetermined threshold, that the change in the safety model parameters (in step 175) is below a predetermined threshold, or that the distance of the next operating point to the set of other operating points in the operating state set is below a predetermined threshold.

[0052] In Figure 4Operating parameters and safe / unsafe operating parameter regions are illustrated. Here, x1 and x2 represent operating values ​​of two operating parameters, expressed in arbitrary units. An unsafe operating parameter region 44 with unsafe operating values ​​(i.e., fault-free operation not possible) for the operating parameters and a safe operating parameter region 46 with safe operating values ​​(i.e., fault-free operation possible) for the operating parameters are shown. These two regions are separated by a (safety) boundary line 42. The exact position of the boundary line 42 is not initially known and may differ for various axial piston pumps of the same type due to manufacturing tolerances.

[0053] Furthermore, a line 48 is shown, illustrating the progress of the method according to the invention, wherein the corner points of line 48 correspond to the (further) operating points that are selected and controlled in such a way that they are safe in terms of the safety model and the variance of the control model is maximized in order to determine corresponding control values. Line 48 thus corresponds to the progress of the loop in Figure 3 The operating parameter range 46, considered safe according to the safety model and taking these additional operating points into account, is also shown. The safe operating parameter range 46 expands as the process progresses until the boundary line 42 is at least approximately reached. In each step, a further operating point within the respective safe operating parameter range 46 is selected. The method according to the invention thus makes it possible, so to speak, to "explore" the safe area.

Claims

1. Method for determining control model parameters of a control model of an axial piston pump (2) with a swash plate (10) which can be adjusted for adjusting its delivery volume in a swash angle (a), on which an adjustment force of an adjusting unit (13) of the axial piston pump (2), dependent on an actuating pressure, acts for adjusting purposes, wherein the control model models at least one control variable (IA, IB) of the adjusting unit (13) as a function of operating variables, wherein operating points of the axial piston pump are each characterized by specific operating values of the operating variables; wherein the method comprises the following steps: providing (110) initial control model parameters of the control model and an operating state set comprising a plurality of initial first operating points; providing (120) initial safety model parameters of a safety model based on a safety set comprising a plurality of initial second operating points and associated safety values, wherein the safety model models a safety variable as a function of the operating variables, which allows a statement of whether fault-free operation is possible or not at the respective operating point; performing the following steps once or multiple times: determining (130) another operating point that is not yet included in the operating state set, wherein the further operating point is determined in such a way that a variance of the control model is maximized, provided that a probability obtained by the safety model that fault-free operation is not possible at the further operating point is less than or equal to a predetermined safety threshold; updating (140) the operating state set and the control model parameters of the control model taking into account the further operating point; and updating (145) the safety set and the safety model parameters of the safety model, taking into account the further operating point.

2. Method according to Claim 1, wherein updating (140) the operating state set and the control model parameters of the control model, taking into account the further operating point, comprises: determining (150) or measuring a further at least one control value of the at least one control variable, such that the further operating point is reached; adding (160) the further operating point and the determined further at least one control value to the operating state set in order to form a supplemented operating state set; redetermining (170) the control model parameters of the control model based on the supplemented operating state set.

3. Method according to either of the preceding claims, wherein updating (145) the safety set and the safety model parameters of the safety model, taking into account the further operating point, comprises: determining (155) a safety value for the further operating point; adding (165) the further operating point and the associated specific safety value to the safety set in order to form a supplemented safety set; redetermining (175) the safety model parameters of the safety model based on the supplemented safety set.

4. Method according to any one of the preceding claims, wherein providing (110) initial control model parameters of the control model and an operating state comprises: determining the initial first operating points by means of physical modelling of the axial piston pump (2), wherein, in the physical modelling, the first operating points are determined in such a way that fault-free operation is possible; and / or providing the initial first operating points using standard operating points, of which it is known that fault-free operation is possible.

5. Method according to Claim 4, wherein providing (110) initial control model parameters of the control model and an operating state set further comprises: determining the control values associated with the first operating points by means of the physical modelling or by means of operating the axial piston pump (2) and adjusting the control values, such that the respective first operating point is reached; determining the control model parameters based on the determined first operating points and the determined associated control values.

6. Method according to any one of the preceding claims, wherein, when providing (120) initial safety model parameters of a safety model, the initial second operating points comprise both safe operating points, at which fault-free operation is possible, and unsafe operating points, at which fault-free operation is not possible; wherein the safe operating points preferably comprise the initial first operating points at least partially.

7. Method according to any one of the preceding claims, wherein a number of the initial first operating points and a number of the initial second operating points lie independently of each other in the range from 10 to 100.

8. Method according to any one of the preceding claims, wherein the safety threshold is less than or equal to 5%; wherein the safety threshold is preferably 2%, more preferably 1%, even more preferably 0.5%, most preferably 0.1%.

9. Method according to any one of the preceding claims, wherein determining (130) a further operating point, updating (140) the operating state set and the control model parameters, and updating (145) the safety set and the safety model parameters are carried out multiple times until a stop condition is fulfilled; wherein, preferably, the stop condition comprises one or more of: the variance of the control model lies below a predetermined threshold, a change in the control model parameters when updating the control model parameters lies below a predetermined threshold, a change in the safety model parameters when updating the safety model parameters lies below a predetermined threshold, or a distance of the further operating point from the set of the other operating points in the operating mode set lies below a predetermined threshold.

10. Method according to any one of the preceding claims, wherein the operating variables comprise one or more of an operating pressure of the axial piston pump (2), a speed of the axial piston pump (2) and a swash angle (α) of the swash plate (10).

11. Method according to any one of the preceding claims, wherein the at least one control variable comprises at least one control current (IA, IB) for at least one pressure control valve (14, 16) of the adjusting unit (13) which is, in particular, electromagnetically actuable.

12. Method according to any one of the preceding claims, wherein the control model and / or the safety model are / is regression models, wherein the operating variables are independent variables and the at least one control variable is at least one dependent variable; wherein preferably Gaussian process models are used as regression models.

13. Computing unit, in particular control unit of a test stand, which is configured to carry out a method according to any one of the preceding claims.

14. Computer program which causes a computing unit to carry out a method according to any one of Claims 1 to 12 when it is run on the computing unit.

15. Machine-readable storage medium with a computer program according to Claim 14 stored thereon.