Method for model parameter adjustment of an axial piston pump

Abstract

Method for optimizing a model (Δpx(Δp, n, α)) of an axial piston pump (2) with a swashplate (10) adjustable to change its delivery volume in a swivel angle (α), on which an adjustment force of an adjustment unit (12) of the axial piston pump (2) dependent on an adjustment pressure (Δpx) and a pressure force resulting from a working pressure (Δp) of the axial piston pump (2) act, from which the swivel angle (α) results as a function of a rotational speed (n), wherein the method comprises the following steps: a. Determining initial model parameters of a quasi-stationary operating point of the axial piston pump (2) via a control unit; b. Determining an initial actuating pressure (Δp) x ) of the adjustment unit (12), depending on the initial model parameters, via the control unit; c. Determining a control variable (I A , I B) of the adjustment unit (12, 14, 16), depending on the initial actuating pressure (Δp x ) and at least one target trajectory (z, z', z'') of the working pressure (Δp) via the control unit; d. Controlling the adjustment unit (12, 14, 16) with the control variable (I A , I B ); e. Acquiring and / or determining actual values ​​of the working pressure (Δp), the rotational speed (n) and the swivel angle (α) via a sensing unit and / or the control unit; and f. Adjusting the model (Δp x (Δp, n, α)) against the actual values, via the control unit, and depending on them; g. Adjusting the model parameters via the control unit.

Description

[0001] The invention relates to a method for model parameter adjustment and control of an axial piston pump according to claim 1.

[0002] A hydrostatic drive system consists of a primary variable displacement pump and a secondary variable displacement motor. The primary and secondary units can be configured in an 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 limited by the pressure relief valve.

[0003] In axial piston pumps of this type with a swashplate design, the delivery volume flow is adjusted by changing the swashplate angle. This adjustment is achieved via a hydraulic cylinder. The cylinder's chamber pressure is regulated by a pressure control valve. The pump exhibits load-sensitive behavior, such that the swashplate is swashplate-swashplate pivoted back by the operating pressure, or more precisely, by the differential pressure acting upon it. This relationship between operating pressure, control pressure, and swashplate angle is used to control the axial piston pump and thus the drive system. The characteristic steady-state behavior, which also depends on the rotational speed, is currently measured on the component test bench or can be calculated.

[0004] In contrast, the invention is based on the objective of creating a method with reduced effort for modeling an axial piston pump.

[0005] The problem is solved by a method having the features of claim 1.

[0006] Advantageous further developments of the method are described in claims 2 to 10.

[0007] An inventive method optimizes a model of an axial piston pump designed with an adjustable delivery volume, wherein the pump has a swashplate adjustable in its swivel angle. An adjusting force of an adjusting unit of the axial piston pump, dependent on a control pressure, and a pressure force resulting from a working pressure or working pressure differential of the axial piston pump act on this swashplate. The swivel angle is adjusted depending on these forces and the rotational speed of the axial piston pump. The method comprises the following steps: Determining initial model parameters of a quasi-stationary operating point of the axial piston pump, in particular as a function of the operating pressure, the rotational speed and the swivel angle, via a control unit; determining an initial actuating pressure of the adjustment unit, Depending on the initial model parameters, in particular from a force equilibrium on the swashplate and its mechanics and / or kinematics, via the control unit; determining a control variable of the adjustment unit, depending on the initial actuating pressure and at least one target trajectory of the working pressure, via the control unit; controlling the adjustment unit with the control variable, via the control unit; acquiring and / or determining actual values ​​of the working pressure, the rotational speed and the swivel angle, via a sensing unit and / or the control unit; and comparing the model against the actual values, via the control unit, and depending on that; adjusting the model parameters via the control unit.

[0008] In a further training course, the procedure has the following steps: Determining an adapted control pressure from the control variable and a characteristic curve of a control valve of the adjustment unit, and; Determining a deviation of the adapted control pressure from the control pressure; via the control unit, whereby the step of adapting the model parameters is carried out in such a way that the deviation is minimized.

[0009] In further training, a model approach function is formed from a multidimensional, in particular 3-dimensional, characteristic field, from polynomial approaches, from a function approximation, in particular a Gaussian process model, or from a neural network.

[0010] The adjustment unit preferably has a hydraulic cylinder to which the swashplate is articulated. In particular, this cylinder is double-acting and each of its two actuating pressure chambers is assigned a control valve, wherein the control variable for each actuating pressure chamber is a control current of the assigned control valve, with which it can be actuated by the control unit.

[0011] In a further training course, one step is to determine at least one target trajectory from a requested work pressure.

[0012] In a further training course, the following steps are provided: recording the working pressure via the recording unit; determining a deviation of the recorded working pressure from its target trajectory via a first comparison element; comparing the latter deviation with a second time derivative of the target trajectory via a second comparison element; and determining the control variable of the adjustment unit as a function of the latter comparison via the control unit.

[0013] In a further training course, a step is provided to determine the control variable of the adjustment unit depending on the first and / or second time derivative of the target trajectory.

[0014] In a training course, the steps are carried out recursively and / or for several operational points or areas.

[0015] An embodiment of a method according to the invention is explained in more detail below with reference to the drawings. The drawings show: Fig. 1 a schematic representation of a hydrostatic drive system with an axial piston pump for which the method according to the invention is provided; Fig. 2 a hydraulic circuit diagram of the axial piston pump of the drive system according to Fig. 1, and Fig. 3 the method according to an exemplary embodiment.

[0016] A method is described for the step-by-step and automatic identification of the steady-state input and output behavior of an axial piston pump of a hydrostatic drive system during operation, based on available sensor information or sensor substitutes, for example, quantities calculated via models.

[0017] This can then be used for controlling and regulating the axial piston pump, as well as for diagnostic purposes.

[0018] According to Fig. In the system 1, a drive unit 1 has a primary axial piston 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. 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 by the pressure relief valve. limited.

[0019] According to Fig. The axial piston pump 2 is designed in a swashplate configuration, 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 a double-acting hydraulic cylinder 12 of an adjustment unit. Both actuating chambers of the hydraulic cylinder 12 can be individually pressurized with actuating fluid. The respective actuating pressure in the actuating chambers is set via pressure regulating valves 14, 16.

[0020] 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.

[0021] According to Fig. Figure 3, which shows an embodiment of the method according to the invention, is based on the input variables working pressure Δp, rotational speed n and swivel angle α of the axial piston pump, as well as control currents I A , I B .

[0022] The working pressure Δρ is in particular a differential pressure across the working lines according to Fig. 2.

[0023] Initial model parameters are corrected at respective quasi-stationary operating points to achieve the best possible agreement between measurement and internal model. To this end, quasi-stationary ranges are first identified in which parameter identification is valid. The advantage of this approach is that the actual parameters of the axial piston pump do not need to be known exactly beforehand, but can be learned specifically for each pump, and changes over its lifetime can also be learned and taken into account. This means that tolerances no longer need to be maintained as precisely during the manufacturing process, and testing measurements can be significantly reduced, thus saving costs.

[0024] The steady-state behavior of the axial piston pump is described by the adjusting force of the adjustment unit acting on the rotary disk, which is proportional to the control pressure Δpx. This force can be calculated with a certain degree of accuracy from the pump geometry as a function of the operating pressure Δp, the rotational speed n, and the rotary angle α. These model parameters are used to initialize an adaptation function.

[0025] In this context, Δpx denotes the actuating pressure or differential pressure of the actuating chambers acting on the adjusting piston, which is necessary to keep the axial piston pump at a specific operating point. This is proportional to the actuating force in relation to the piston areas. For the control of the control valves according to... Fig. 2 Within a process section “Flatness-based Feedforward Control”, the control pressure Δpx is combined with a planned target trajectory z and its two time derivatives z', z'' into the control streams I A ; I Bconverted.

[0026] The "Parameter Adaptation" process step now compares the model of the actuating force Δpx = f(Δp, n, α) with measurement data during operation. For this purpose, the control currents I A ; I B Internally, the control valve characteristic curves are used to calculate an adapted control pressure Δp'x, and the adaptation corrects the parameters of the underlying basis function f(Δp, n, α) such that the calculated control pressure Δp'x matches the model value Δpx as closely as possible.

[0027] To do this, a model error e = Δp'x - Δpx is minimized.

[0028] Various functions can be chosen as the basis function for f(Δp, n, α). For example, a 3-dimensional characteristic map, polynomial approaches, or other methods for function approximation, such as Gaussian process models, neural networks, or the like. Furthermore, existing model knowledge can be explicitly used.

[0029] For example, the dependence of the swivel angle on the approach function or the model can be factored out, thus reducing the actual estimation problem by one dimension. This means, for instance, that only a 2-dimensional characteristic map needs to be learned, not a 3-dimensional one.

[0030] Fig. Section 4 further describes that in this procedure it is possible to determine the desired working pressure Δp. des to take the pump into account so that the pump generates the desired pressure.

[0031] In one embodiment of the present invention, the actual value Δp is used for this purpose. act of the working pressure Δp, recorded via the acquisition unit and a deviation of the actual value Δp act The desired trajectory z is determined via a first comparison element.

[0032] This deviation is added to the second time derivative z'' of the target trajectory z, and the control variables I A , I BThe values ​​of the adjustment units 12, 14, 16 are determined via the control unit as a function of the thus corrected second time derivative z''. This special solution is not necessary; the control intervention can also be applied directly to the two control variables I. A , I B This will be done. Nevertheless, this solution is particularly advantageous because it allows this deviation to be taken into account in the "flatness-based feedforward controller" so that the deviation is linearized.

[0033] Online identification of the steady-state behavior of the axial piston pump can be demonstrated in combination with a control strategy by measuring the control currents and actuator pressures. These will adapt to the pump's behavior over time and change, for example, when the same operating point is repeatedly approached. When running the same pattern, this behavior can also be distinguished from that of an additional controller, as the latter exhibits the same transient response, while the steady-state behavior identification corrects and improves this behavior over time. Artificial inputs can also be generated directly at the control unit where this function is implemented, instead of actual sensor signals, to verify this behavior.

[0034] A method for optimizing a model of an axial piston pump has been disclosed, in which an adaptation function learns the specific steady-state behavior of the axial piston pump during a calibration cycle. This cycle can be performed entirely within a vehicle in which the axial piston pump is installed as the primary unit. This allows for the identification of manufacturing-related parameter uncertainties of the axial piston pump. Furthermore, the adaptation function can also be used during operation to identify and compensate for wear, i.e., changes in the steady-state behavior of the axial piston pump. The learned characteristic steady-state behavior is then used in the control function of the axial piston pump. This allows the system to switch to a purely controlled mode even in the event of a sensor failure, enabling the vehicle to continue operating, at least temporarily. Reference symbol list 1 Hydrostatic drive 2 Axial piston pump (primary unit) 4 Adjustment motor (secondary unit) 6 Work management 8 Work management 10 Slanted disc 12 hydraulic cylinders of the adjustment unit 14 Pressure regulating valve (control valve of the adjustment unit) 16 Pressure regulating valve (control valve of the adjustment unit) α Swivel angle of the swashplate (10) n speed of the axial piston pump (2) Δp working pressure of the axial piston pump (2) Δpx actuation pressure Δp'x adjusted control pressure Δp des Target value Δp act Actual value I A , I B Control parameter (control current for pressure control valve 14, 16) z target trajectory z' First temporal derivative of the target trajectory z'' Second temporal derivative of the desired trajectory

Claims

[1] Method for optimizing a model (Δpx(Δp, n, α)) of an axial piston pump (2) with a swashplate (10) adjustable to change its delivery volume in a swivel angle (α), on which an adjustment force of an adjustment unit (12) of the axial piston pump (2) dependent on an adjustment pressure (Δpx) and a pressure force resulting from a working pressure (Δp) of the axial piston pump (2) act, from which the swivel angle (α) results as a function of a rotational speed (n), wherein the method comprises the following steps: a. Determining initial model parameters of a quasi-stationary operating point of the axial piston pump (2) via a control unit; b. Determining an initial actuating pressure (Δp) x ) of the adjustment unit (12), depending on the initial model parameters, via the control unit; c. Determining a control variable (I A , I B) of the adjustment unit (12, 14, 16), depending on the initial actuating pressure (Δp x ) and at least one target trajectory (z, z', z'') of the working pressure (Δp) via the control unit; d. Controlling the adjustment unit (12, 14, 16) with the control variable (I A , I B ); e. Acquiring and / or determining actual values ​​of the working pressure (Δp), the rotational speed (n) and the swivel angle (α) via a sensing unit and / or the control unit; and f. Adjusting the model (Δp x (Δp, n, α)) against the actual values, via the control unit, and depending on them; g. Adjusting the model parameters via the control unit. [2] The method of claim 1, wherein step f comprises: - Calculating a back-calculated control pressure (Δp') x ) from the determined control variable (I A , I B) and preferably a characteristic curve of a control valve (14, 16) of the adjustment unit (12, 14, 16), and; - Determining a deviation of the back-calculated control pressure (Δp') x ) from the actuating pressure (Δp x ) via the control unit; wherein step g. is carried out such that the deviation between the back-calculated control pressure (Δp') x ) and the actuating pressure (Δp x ) is minimized. [3] Method according to one of claims 1 or 2, wherein in step b. the initial actuating pressure (Δp x ) depending on the current actual values ​​of the working pressure (Δp), the rotational speed (n) and the swivel angle (α), is determined via the control unit. [4] Method according to one of claims 1 to 3, wherein a basis function (f) of the model (Δpx(Δp, n, α)) is formed from a multidimensional, in particular 3-dimensional, characteristic field, from polynomial approaches, from a function approximation, in particular a Gaussian process model, or from a neural network. [5] Method according to any one of claims 1 to 4, wherein the adjusting unit (12, 14, 16) has a hydraulic cylinder (12) to which the swashplate (10) is articulated. [6] Method according to claims 2 to 5, wherein the hydraulic cylinder (12) is double-acting and each of its two actuating pressure chambers is assigned a control valve (14, 16), wherein the control variable per actuating pressure chamber is a control current (IA, I) B ) of the associated control valve (14, 16). [7] Method according to any one of claims 1 to 6, wherein the method further comprises the following step: - Determining at least one target trajectory (z, z', z'') from a requested working pressure (Δp) des ) the axial piston pump (2). [8] The method of claim 7, wherein the method further comprises the following steps - Recording an actual value (Δp act ) of the working pressure (Δp), via the acquisition unit; - Determining a deviation of the actual value (Δp) act ) working pressure (Δp) from its target trajectory (z), via a first comparison element; - Comparing the latter deviation with a second time derivative (z'') of the target trajectory (z), via a second comparison term, and; - Determining the control variable (I A , I B ) of the adjustment unit (12, 14, 16) depending on the latter comparison, via the control unit. [9] Method according to any of the preceding claims, wherein the said model (Δpx(Δp, n, α)) is a function of the actuating pressure (Δpx ) as a function of the rotational speed (n), the swivel angle (α) and the working pressure of the axial piston pump (2). [10] Method according to any one of the preceding claims comprising one step - Determining the control variable (I A , I B ) of the adjustment unit (12, 14, 16) as a function of the first and / or second time derivative (z', z'') of the desired trajectory (z).

Images