Method for model-based operation of a coolant system, control unit and fuel cell system

Abstract

For the operation of the electrochemical energy converter, a control object (Σ p The model-based actuation of the control variable (y(k)) is determined and the control object (Σ p (·)) extended object model (Σ m,aug (·)) to determine the time derivative of the controlled variable y(k) (equation (I), (I), equation (II), (II),..., y(n)) up to the system dimension (n) of the controlled object (Σ p (·)) actuator (Σ a (·)) actual adjustment amount (u act (k)) is determined; and depending on the determination, the compensation and / or correction required (dr eq (k)).

Description

[Technical Field]

[0001] The present invention relates to a method for model-based activation, in particular control, of a control object, preferably in the form of a subsystem of a coolant system, preferably for the operation of an electrochemical energy converter, for example in the form of a fuel cell system, an electrolyzer, etc. Furthermore, the present invention relates to a corresponding computer program product, a corresponding controller, and a corresponding electrochemical energy converter. [Background technology]

[0002] Model-based control methods are well known, for example as so-called internal model controllers (IMC). A prior art IMC calculates / predicts a deviation / error term (i.e., model error or disturbance) as the difference between the output of the controlled object (i.e., the controlled variable) and the output of the controlled object model. The deviation / error term is often used to compensate / correct the control variable / setpoint accordingly. This correction is then made at the input of the action chain. However, if the main model error (tolerance) is located near the output of the action chain (e.g., near the adjusted variable), the action point of the model equation, located between the input and output of the action chain, is incorrect. This leads to an incorrect operating-point-dependent linearization of the model. Therefore, in such cases, poor control behavior (e.g., oscillations) must be expected. Summary of the Invention

[0003] The present invention relates to a method for model-based operation, in particular control, of a control object, preferably in the form of a subsystem of a coolant system, preferably for the operation of an electrochemical energy converter, for example in the form of a fuel cell system, an electrolyzer, etc. Furthermore, the present invention relates to a corresponding computer program product, in particular a corresponding controller, in particular in the form of an improved internal model controller (IMC), and a corresponding energy converter, in accordance with the features of the subclaims, whereby elements and details described in the context of different embodiments and / or aspects of the invention are naturally also valid in the context of other embodiments and / or aspects and vice versa, so that cross-reference is or can always be made to the disclosure relating to the individual embodiments and / or aspects.

[0004] According to a first aspect, the present invention provides a method (hereinafter also referred to as control method) for model-based operation, in particular control, of a control object, preferably in the form of a subsystem of a coolant system, preferably for the operation of an electrochemical energy converter, for example in the form of a fuel cell system, an electrolyzer, or the like.

[0005] The method has the following method steps / actions: - The controlled variable y(k) is determined (e.g., measured or observed) and the controlled object Σ p (·) target model Σ m (·) or the extended object model Σ m,aug (·) time derivative of the controlled variable up to the system dimension n

number

[0006] In this regard, the steps of the method may be performed simultaneously, sequentially, and / or with at least partial overlap.

[0007] The method proposes a novel control method or control concept, which has the advantage that it can improve the control, in particular the control quality, in the case of a nonlinear plant and / or a nonlinear plant model. The novel control method has the advantage that it can include a flatness-based controller approach, in which the setpoint trajectory y(k) for the controlled variable y(k) is set A feedforward control portion / feedforward control functionality may be included to enable tracking of (k).

[0008] This novel control method is based on the maximum uncertainty / inaccuracy of the model and / or the unknown disturbance quantity z u (k) can be directly addressed (i.e., estimated and corrected). In other words, this model-based control method uses nothing but the target model Σ m Uncertainty / inaccuracy in (·) and / or unknown disturbance quantities z u (k) is maximized, and the required compensation / correction term d req (k) provides a control intervention, or the controller acts to control the target model Σ m Such uncertainties / inaccuracies in (·) and / or unknown disturbance quantities z u (k) can be compensated for. In this way, the (implicit / intrinsic) operating point-dependent linearization achieved at the operating point of the nonlinear plant and the operating point of the nonlinear plant model can be adapted to each other, which can lead to a significant increase in the robustness of the control method.

[0009] an extended (possibly dynamic and nonlinear) object model Σ m,aug (·), which is the (possibly dynamic and nonlinear) object model Σ m (·) is the target model Σ m By extending the expression (·) by inserting an additional deviation or error term d(k) (e.g. in the form of an additive term, a multiplicative term, etc.).

[0010] Actuator Σ a If (·) (or the regulator) can provide information about its actual adjustment amount, such information can be taken into account in a desirable way in the modeled control concept. Furthermore, the actual adjustment amount u act To determine (k), actuator Σ a An intrinsic additional model Σ that models the dynamic response behavior of (·) a,m (·) can be used.

[0011] This new control method is m This allows us to estimate / calculate the deviation / error term d(k) in a preferred way for almost any model equation in (·). The deviation / error term d(k) is the deviation / error term of the target model Σ m (·) can be expanded as an internal model quantity in any desired manner (e.g., as an additive term, multiplicative term, etc.).

[0012] Actuator Σ within the framework of the present invention a The (·) can include, for example, a pump, a valve, etc.

[0013] This new control method is applicable to various different control targets Σ p (·) and / or various target models Σ m (·) can be applied in a preferred manner.

[0014] A coolant system can be understood to mean any temperature regulation system that makes it possible to regulate the temperature of an electrochemical energy converter, in particular in the form of a fuel cell stack or an electrolyzer in a fuel cell system, by means of a cooling fluid or coolant.

[0015] Coolants can generally be referred to as temperature regulators.

[0016] In some operating modes of the fuel cell system, e.g., frozen start and / or cold start, the coolant system can serve to warm the fuel cell stack to an appropriate operating temperature. In further operating modes of the fuel cell system, e.g., normal operation, high load operation, etc., the coolant system can serve to cool the fuel cell stack to an appropriate operating temperature.

[0017] A coolant system within the framework of the present disclosure may comprise a coolant pump for circulating the coolant, and may further comprise a cooler bypass valve (e.g., in the form of a three-way valve, a so-called mixing point of the coolant system) for directing the coolant in one part through the cooler (e.g., a radiator) and / or in another part past the cooler.

[0018] Coolant system control object Σ p (·) may be, for example, a mixing point where the coolant in the cooling path mixes with the coolant in the bypass path.

[0019] Coolant system control object Σ p (·) may be, for example, a coolant path through a fuel cell stack in a fuel cell system.

[0020] Coolant system control object Σ p (·) may be a coolant path through, for example, a cooler.

[0021] Furthermore, the controlled variable y(k) is the measured controlled variable y measd(k) or the observed controlled variable y obs (k). Furthermore, the time derivative of the controlled variable y(k)

number

number

number

number

number

number

[0022] Furthermore, the actual adjustment amount u act (k) is the measured adjustment amount u measd (k) or as the actuator model Σ a,m (·) (i.e., the required adjustment u req From (k), the control object Σ p It is intended that the actual adjustment u(k) be determined using a mathematical model that describes the dynamic response behavior to the actual realistic physical adjustment u(k) acting on (·). act This has the advantage of providing two fundamentally different options for how (k) can be determined. These two fundamentally different options are described in detail below.

[0023] Actuator (i.e., regulator) Σ a Depending on (·), the measured adjustment amount u measd (k) is the actuator Σ a (·). For example, it may be the case that a coolant pump directly measures the coolant mass flow rate that is occurring. Or, an actuator (i.e., a regulator) Σ a Depending on (·), in some cases, the measured actuator information u measd,a (k) (e.g., the position of the controller, the number of rotations of the controller, etc.) is used to calculate the measured adjustment amount u measd (k) can be indirectly determined. Thus, the augmented object model in which the equations are formulated (i.e., solved) to calculate the estimated deviation / error term d(k) is

number

[0024] Actuator Σ a (·) is the measured adjustment amount u measd In cases where (k) cannot be supplied directly or indirectly, the actual adjustment amount u act To determine (k), we use the actuator model Σ a,m (·) can be taken into account. In other words, a mathematical model that models the dynamic response behavior of the actuator can be included in the proposed novel control method. Actuator model Σ a,m (·) is the required adjustment u req From (k), the control object Σ p It is possible to describe the (particularly dynamic) response behavior to the actual realistic physical adjustment u(k) (i.e., the actual effective adjustment) acting on (·). The "requested" adjustment u(k) of the preceding time step req (k-1) (k is the time step) is defined as the actuator model Σ a,m (·) can be used as input for the so-called "actual" adjustment u act (k) is the actuator model Σ a,m It can be equated with the output signal of (·).

[0025] The advantage of mathematical models for dynamic response behavior of actuators can be illustrated by an example. a In the case where (·) is the coolant pump, the required adjustment amount u req (k) is the required coolant mass flow rate to be provided by the coolant pump. Then the actual adjustment amount u act(k) is the actual physically effective coolant mass flow rate currently provided by the coolant pump. Since the rotating parts of the coolant pump have a moment of inertia and the coolant must be accelerated, it takes some time for the required coolant mass flow rate to actually be provided by the coolant pump. The mathematical model of the coolant pump can reflect this in a desirable way. A simple mathematical actuator model Σ a,m As an example of (·), a PT1 filter can be used.

[0026] The method further comprises: - a filter Σ whose input is the calculated / estimated deviation / error term d(k) IMC (·) (e.g., the output of the PT1 filter) as the compensation / correction term d req (k) is determined; Filter Σ IMC The parameters of (·) (e.g. time constant, amplification) affect the control intervention and therefore become tuning parameters of the improved internal model controller (IMC), which must be properly adjusted to achieve the desired control quality.

[0027] Additionally, the method may further comprise: - at least one disturbance quantity z, in particular calculated and / or measured m (k) is determined; At least one disturbance quantity z, in particular calculated and / or measured m (k) is the compensation / correction term d req (k) will be taken into account in providing the compensation / correction item d. req (k), the calculated and / or measured disturbance quantity z m (k), the desired control amount y des (k) (the desired value of the controlled variable y(k)), and its time derivative

number

[0028] On the other hand, the control object Σ in the form of a fuel cell stack p It is conceivable that the method can be applied to actuate, in particular to control, (·), in which case the coolant outlet temperature from the fuel cell stack can preferably be used as control variable y(k), in particular to control the coolant temperature difference through the fuel cell stack.

[0029] Furthermore, the control object Σ of the shape of the mixing point of the cooling system p It is conceivable that the method can be applied to operate, in particular to control, (·), in which case the coolant inlet temperature to the fuel cell stack can preferably be used as the controlled variable y(k).

[0030] Furthermore, the control object Σ in the form of a cooler (or generally a heat transfer device) of the coolant system p It is conceivable that the method can be applied to actuate (·), in particular to control it, in which case the coolant outlet temperature from the cooler can preferably be used as the control variable y(k).

[0031] The method thus makes it possible to enable a preferred modelled control of coolant systems, in particular for the operation of electrochemical energy converters.

[0032] Furthermore, it may be advantageous to be able to apply the method to operate, in particular to control, a controlled object having a plurality of input quantities and / or a plurality of adjustment quantities and / or a plurality of control quantities.

[0033] In another aspect, the invention provides a computer program product comprising commands that, when executed by a computer, cause the computer to carry out the method, which can proceed as described above. Such a computer program product can achieve the same advantages as those described above in relation to the method according to the invention, and these advantages are also incorporated herein by full reference.

[0034] A corresponding controller provides a further aspect of the invention. A computer program in the form of code can be stored in a storage unit of the controller, which implements a method that can proceed as described above when the code is executed by a calculation unit of the controller. This controller makes it possible to achieve the same advantages as those described above in connection with the method according to the invention, and these advantages are also fully referenced herein.

[0035] A further aspect of the invention is a corresponding electrochemical energy converter with a corresponding controller, which makes it possible to realize the same advantages as those explained above in connection with the method according to the invention, to which full reference is also made herein.

[0036] The invention, its developments and its advantages will now be explained in more detail with reference to the drawings, which respectively show, diagrammatically: [Brief explanation of the drawings]

[0037] [Figure 1] A nonlinear control plant Σp(·) with flat output and a nonlinear object model Σm(·) (the equations of which are formulated / solved here as follows:

number

number

number

[0038] In the blocks in Figures 1 to 5, the "(·)" has been omitted to save space (e.g., Σ p (·) is Σp abbreviated as ).

[0039] Figure 1 shows a nonlinear control plant Σ with flat output y(k). p (·) and the nonlinear target model Σ m For (·), a block diagram of a (flatness-based) internal model controller based on the prior art is shown, and the inverted model

number

number

[0040] 2 is intended to illustrate the proposed method, which is also called control method, which is based on the control object Σ , preferably in the form of a subsystem of a coolant system 100, for the operation of an electrochemical energy converter 101, for example in the form of a fuel cell system or an electrolyzer. p It has been developed specifically for model-based operation and control of (·). An exemplary coolant system 100 for a fuel cell system having a fuel cell stack 101 is shown in FIG.

[0041] The method has the following method steps / actions: - The controlled variable y(k) is determined (e.g., measured or observed) and the controlled object Σ p (·) extended object model Σ m,aug (·) time derivative of the control variable y(k) up to the system dimension n

number

[0042] In this regard, the steps of the method may be performed simultaneously, sequentially, and / or with at least partial overlap.

[0043] By this method, a new control method or control concept 200 can be proposed. This new control method is based on the nonlinear control object Σ p (·) and / or a nonlinear target model Σ m In the case of (·), the control can be made more robust in a preferred manner. As suggested in Figure 2, this novel control method can include a flatness-based controller approach in a preferred manner, and the desired trajectory y(k) for the controlled variable y(k) can be set A feedforward control portion / feedforward control functionality may be included to enable tracking of (k).

[0044] As shown in Figure 2, this novel control method is m The maximum uncertainty / inaccuracy in (·) and / or unknown disturbance quantity z u (k) is calculated / estimated directly by the deviation / error term d(k), followed by the compensation / correction term d req (k) can be modified in a desirable way. For this purpose, the target model Σ m Uncertainty / inaccuracy in (·) and / or unknown disturbance quantities z u(k) is maximized, the controller 110 (see FIG. 2) adjusts (in particular the required compensation / correction term d req (k)) to create such an object model Σ m Uncertainty / inaccuracy in (·) and / or unknown disturbance quantities z u (k) can be compensated.

[0045] As shown in Figure 4, the nonlinear control plant Σ p At the operating point OP of (·), and the nonlinear target model Σ m At the operating point OP′ of (·), the realized (implicit / intrinsic) operating point dependent linearization can be adapted, which can lead to a significant increase in robustness in the control concept 200.

[0046] As shown in FIG. 2, the actuator (i.e., the adjuster) adjusts its actual adjustment amount u act Measurement value u for determining (k) measd,a (k), the model-based control concept 200 can measd,a (k) can be considered in a preferred manner.

[0047] Furthermore, as shown in Figure 2, the actual adjustment amount u act To determine (k), an intrinsic additional actuator model Σ a,m (·) can be used.

[0048] This new control concept 200 is applicable to the target model Σ m For almost any model equation of (·), it is possible to estimate / calculate the deviation / error term d(k) in a preferred manner. The deviation / error term d(k) can be, for example, an internal model quantity (e.g., as an additive term, multiplicative term, etc.) of the target model Σ m (·), and the target model Σ m (·) is the so-called extended object model Σ m,aug It can be extended in a preferred way to (·).

[0049] Compensation / Amendment d rek (k) (i.e., the control intervention) is a filter Σ(k) whose input is the calculated / estimated deviation / error term d(k). IMC (·) (for example, the output of the PT1 filter), and the filter Σ IMC The parameters of (·) (e.g., time constant, amplification) influence the control intervention and therefore become tuning parameters of the improved internal model controller (IMC), which must be adjusted appropriately to achieve the desired control quality.

[0050] An actuator within the scope of the present invention may comprise, for example, a coolant pump 105, a valve 106 (for example a three-way valve), etc.

[0051] This new control concept 200 is designed to handle a wide range of different control targets Σ p (·) and / or various target models Σ m (·) can be applied in a preferred manner.

[0052] This novel control concept 200 can be used with multiple input variables and / or multiple adjustment variables u(k) and / or multiple control variables y(k), for various different control objectives Σ p (·) and / or various target models Σ m It can be applied in a preferred manner together with (·).

[0053] A coolant system 100 can in principle be understood as any temperature regulation system. Figure 6 shows an exemplary coolant system 100 that can allow for temperature regulation of a fuel cell stack 101 of a fuel cell system by means of a cooling fluid or coolant.

[0054] 6 shows, the coolant system 100 may have a coolant pump 105 for circulating the coolant. Furthermore, the coolant system 100 may have a cooler bypass valve 106, e.g. in the form of a three-way valve, and a so-called mixing point 102, for directing the coolant in one part through a cooler 103 (e.g. a radiator) and / or in another part past the cooler 103.

[0055] Control object Σ of coolant system 100 p (·) may be, for example, the mixing point 102 where the coolant in the cooler path 103a mixes with the coolant in the bypass path 106a. In this case, the coolant inlet temperature to the fuel cell stack (101) can be used as the controlled variable y(k).

[0056] Control object Σ of coolant system 100 p (·) may be, for example, the coolant path through the fuel cell stack 101 of the fuel cell system. As a controlled variable y(k), the coolant outlet temperature Tc,out from the fuel cell stack 101 can preferably be used, in particular to control the coolant temperature difference through the fuel cell stack 101.

[0057] Control object Σ of coolant system 100 p (·) may be, for example, the coolant path through the cooler 103. In this case, the coolant outflow temperature from the cooler 103 can be used as the controlled variable y(k).

[0058] Figure 1 shows a nonlinear control plant Σ with flat output y(k). p (·) and the nonlinear target model Σ m (·) shows a block diagram of a well-known (flatness-based) internal model controller (IMC). Here, the deviation / error term d(k) is the product of the output (i.e., the controlled variable) of the controlled object y(k) and the object model y m (k) is calculated as the difference between the output of m(k)). As already mentioned, the basic idea of ​​the well-known IMC is to use the deviation / error term d(k) (i.e., model error or unknown disturbance (n)z u (k)) and accordingly calculate the control amount / target value y set (k). As is clear from Figure 1, such compensation involves the so-called required correction term d req (k) is the target value y set (k) is subtracted from the required correction term d. req (k) is a filtered version of the estimated deviation / error term d(k).

[0059] Block Σ des (·) is the trajectory generator. Block Σ IMC (·) is a filter whose parameters (e.g. time constant, amplification) define the control intervention.

[0060] block

number

number

number

[0061] It should be noted that the control method of the internal model controller based on the prior art differs from the present invention proposed here in that the target model Σ m (·) is the target model Σ m The expanded object model Σ (·) is expanded by a quantity (e.g., in the form of an additive term, multiplicative term, etc.) inserted into one of the equations. m,aug It does not become (·).

[0062] Block Σ p (·) is a nonlinear control object.

[0063] If the maximum model error (tolerance) is located near the output of the action chain (for example, if the maximum model error is incorporated into the equation f3(·)), the operating points of the model equations located between the input and output of the action chain (i.e., f1(·), f2(·) and f3(·)) will be incorrect, which will lead to an incorrect linearization of the model depending on the operating point, as shown in Figure 3 (but for a dynamic nonlinear controlled object Σ p (·) or dynamic nonlinear object model Σ m (·), we use a static nonlinear control object Σ p (·) or a static nonlinear object model Σ m (·) has been suggested for the simpler case.

[0064] Figure 3 shows the target model Σ m (·) is presented. For simplicity, we focus on the case of a static (rather than dynamic) nonlinear model. The fact that Figures 3 and 4 focus on a static example instead of a dynamic one means that the functions u(k) and y(k) or y m (k) and (k) can be easily illustrated in two dimensions as a graph. req (k) is introduced at the input of the chain (i.e., the first variable). The "required" correction term d req (k) is the target value y set (k) (see also Figure 1), the corrected target value e(k)=y set,corr (k) is determined. As is clear from Figure 3, the control object Σ p (·) operating point OP(target value y set (k) is the actual nonlinear control object Σ p The slope of the tangent at (the point where it intersects with the curve of ) is the slope of the tangent at the point where it intersects with the curve of the target model Σ m (·) operating point OP' (corrected target value e(k) = yset,corr (k) is a nonlinear target model Σ m The slope of the tangent at OP' is clearly different from the slope of the tangent at OP'. m Since the operating point dependent linearization of (·) is clearly incorrect, we are forced to expect poor control behavior (e.g., oscillations) in Figure 3.

[0065] Figure 4 shows that instead of introducing compensation or correction at the input of the action chain (i.e., the first variable), as in Figures 1 and 3, in Figure 4 the compensation or correction is made at the output of the action chain (i.e., the last variable), i.e., at the adjustment u(k). Because in Figure 4 the compensation or correction is made at the output of the action chain (i.e., the last variable), the inverted model

number

[0066] The output of the controlled object y(k) (i.e., the controlled variable) and the object model y m Instead of calculating the deviation / error term d(k) as the difference between the outputs of (k), the present invention proposes to calculate it as an internal model quantity.

[0067] In the case of a nonlinear controlled plant and a nonlinear object model, this means that the object model Σ m (·) is a nonlinear control plant Σ with high model uncertainty / model inaccuracy p The part inside (·) provides the freedom to estimate / calculate the deviation / error term d(k).

[0068] In this way, improved operating point dependent linearization can be achieved, which leads to a robust (Internal Model Controller or IMC) control concept 200.

[0069] Figure 2 shows the basic structure of the proposed novel nonlinear flatness-based Internal Model Controller (IMC) control concept 200. The proposed IMC control concept 200 can also be applied to multiple-input multiple-output (MIMO) systems (German: Mehrgroessensysteme) (replace "oe" with "oe" and "ss" with "eszett"), but for simplicity, Figure 2 shows only the single-input single-output (SISO) (German: Eingroessensysteme) case (i.e., the relationship between the adjusted variable u(k) and the controlled variable y(k) or its measured variable y(k)). measd (k) is a scalar signal, not a vector signal as in the MIMO case).

[0070] In Figure 2, blocks 101, 102, and 103 represent the entire controlled system or the real physical system. In this description, the entire controlled system is defined as the actuator (i.e., the regulator) Σ a (·), the control object to be controlled Σ p (·), and the measured control variable y measd Sensor Σ that supplies (k) s Used whenever (·) is considered to be a unit.

[0071] Depending on the type of actuator (i.e., regulator) or sensor (i.e., their response behavior) and the desired control quality, models of the actuators and / or sensors may or may not have to be taken into account in the control concept.

[0072] In reality, the disturbance z(k) is the control object Σ p (·) and / or the entire controlled object. Such a disturbance z(k) always acts on the unknown disturbance z(k) in principle. u (k), but sometimes includes a calculable / measurable (and therefore known) disturbance quantity z m (k) may also be included. A calculable / measurable (and therefore known) disturbance quantity z m(k) can be systematically taken into account in this new control method because it is used as an input to the block

number

[0073] In the following, Σ p (·) represents the (possibly dynamic, nonlinear) controlled object (i.e., a real physical system, but with actuators Σ a (·) and sensor Σ s (·) is not included. Σ m (·) represents the (dynamic, nonlinear) object model (i.e., a mathematical model of a real physical system, but with actuators Σ a (·) and sensor Σ s (·)

[0074] Σ m,aug (·) is the (possibly dynamic, nonlinear) target model Σ m (·) is the objective model Σ m represents the extended (possibly dynamic nonlinear) target model, which is achieved by extending it with one of the equations in (·).

[0075] In addition, the control object Σ p (·), target model Σ m (·), and the extended target model Σ m,aug (·) can, in the general case, be represented / described by dynamic nonlinear functions. The symbol "(·)" is used to denote that these nonlinear functions depend on a number of input arguments, which for readability are listed only if necessary for understanding.

[0076] In the following, the symbol Σ m,aug(·) represents the mathematical formulation of the (dynamic nonlinear) extended plant model, which is the modeled plant output y m (k) (i.e., the value of the controlled variable y(k) calculated by the model) in terms of the adjusted variable u(k), the calculated and / or measured (and therefore known) disturbance variable z m (k), and the estimated deviation / error term d(k), i.e.: y m =Σ m,aug (u,z m ,d).

[0077] symbol

number

number

number

[0078] symbol

number

number

number

[0079] Reference numeral 201 denotes a software portion specific to the actuator.

[0080] The control signal / target value signal y set (k) is the trajectory generator Σ des (·) (for example, the target model Σ m For the case where the system order n of (·) is n = 1, the PT1 filter can be used as the trajectory generator; m For the case where the system order n of (·) is n=0, the trajectory generator is not needed and can therefore be omitted.) The trajectory generator Σ des The purpose of (·) is to obtain the desired control variable y des (k) (i.e., the desired value of the controlled variable y(k)) and its time derivative

number

[0081] And the desired control amount y des (k) and its time derivative

number

number

number

number

number

[0082] Compensation / amendment requested req (k) is d req (k)=Σ IMC (d(k)). Therefore, the so-called required adjustment u req (k) is calculated as follows:

number

[0083] and the required adjustment amount u req (k) is controlled by the actuator-specific software part 201 to control each actuator Σ a The control signal u required by (·) req,a (k) (e.g., a pulse-width modulated signal).

[0084] Block Σ IMC (·) is a filter (e.g., a PT1 filter) whose parameters (e.g., time constant, amplification) define the control intervention and thus can be tuning parameters of the IMC, which can be adjusted appropriately to achieve the desired control quality. Block Σ IMC The input of (·), the estimated deviation / error term d(k), is

number

[0085] for that

number

number

number

number

[0086] Observer Σ obs (·) is the target model Σ m (·), and possibly a model of the actuator (i.e., the actuator Σ a (·)) and / or a model of the sensor (i.e., the sensor Σ s (·) mathematical model).

[0087] Time derivative

number

number

[0088] moreover

number

[0089] Actuator (i.e., regulator) Σ a Depending on (·), in some cases u measd (k) can be measured directly (e.g., a coolant pump may directly measure its coolant mass flow rate), or measured actuator information u measd,a (k) (e.g., regulator position, regulator rotation speed, etc.) can be calculated by the actuator-specific software part. For example, if a coolant pump is at its current rotation speed (i.e., u in this case), measd,a (k) is the current RPM), which can be used to calculate the coolant mass flow rate currently provided by the coolant pump. It should be noted here that this document does not go into the actuator specific software portion in depth. However, the actuator (i.e., the regulator) Σ a (·) is u measd (k) or u measd,a For cases where (k) cannot be provided (the switch in FIG. 2 is in the upper position as shown in this case), the proposed novel nonlinear flatness-based IMC control concept 200 provides a solution.

[0090] In the following, the specific actuator (i.e., regulator) Σ a (·) is u measd (k) or u measd,a (k) Explain what steps can be taken if it is not possible to provide it.

[0091] A mathematical model Σ that models the dynamic response behavior of the actuator a,m (·) is incorporated into the proposed novel nonlinear flatness-based IMC control concept 200 (see Figure 2). This model Σ a,m (·) is the required adjustment u req From (k), the control object Σ p The (dynamic) response behavior to the actual real physical adjustment u(k) (i.e., the actual effective adjustment) acting on (·) is described. In order to avoid algebraic loops within the software that can implement the novel control method or control concept 200, the required adjustment u(k) of the previous time step (k-1) is req (k-1) -1 ) and the required adjustment amount u req (k) is provided in a timely manner. The quantity k represents a time step. And, as can be seen from Figure 2, the required adjustment quantity u of the preceding time step req (k-1) is the block Σ that models the dynamic response behavior of the actuator. a,m (·) as input, and the so-called actual adjustment u act (k) is block Σ a,m is identified with the output signal of (·), i.e.: u act (k)=Σ a,m (u req (k-1)).

[0092] In the following, we use the mathematical model Σ, which models the dynamic response behavior of the actuator. a,m The advantage of incorporating (·) is explained by way of example. In the case where the actuator is a coolant pump 105, u reqwhere (k) is the desired coolant mass flow rate to be provided by the coolant pump 105, and u(k) is the actual physically effective coolant mass flow rate currently being provided by the coolant pump. Because the rotating parts of the coolant pump 105 have moments of inertia and the coolant must be accelerated, the "desired" coolant mass flow rate u req It will take some time before (k) can actually be provided by the coolant pump 105. This fact is expressed in the mathematical model Σ a,m (·). A simple model Σ that models the dynamic response behavior of the actuator a,m In this case, a PT1 filter can be used as (·).

[0093] Figure 5 shows the extended (dynamic nonlinear) target model Σ m,aug (·) (on the other hand, the mathematical formulation

number

number

number

number

[0094] In Figure 5, the quantities in the circles (e.g.

number

number

number

number

number

number

number

number

number

number

[0095] A corresponding computer program, in particular a corresponding controller 110 in the form of an improved Internal Model Controller (IMC), and a corresponding energy converter (not shown in its entirety for simplicity's sake) are also inventive aspects.

[0096] The above-described embodiments are merely illustrative of the present invention, and the individual features of the embodiments can of course be freely combined with one another, provided that this is technically meaningful, without departing from the scope of the present invention. [Explanation of symbols]

[0097] 100 Coolant System 101 Fuel Cell Stack 102 Mixing point 103 Cooler 110 Controller Σ p (·) Control target y(k) Control amount

number

number

number

Claims

1. For the operation of an energy converter, preferably of the electrochemical type, a control object (Σ ｐ In a method for model-based operation, in particular control, of (·), - The controlled variable (y(k)) is determined and the controlled object (Σ ｐ (・)) ｍ，ａｕｇ (·)) to the system dimension (n) of the control variable (y(k)) [Equation 1] is determined, - the controlled object (Σ ｐ (・)) actuator (Σ ａ The actual adjustment amount (u ａｃｔ (k)) is determined; - the compensation and / or correction terms required depending on said determination (d ｒｅｑ (k)) is provided; A method comprising:

2. The controlled variable (y(k)) is the measured controlled variable (y ｍｅａｓｄ (k)) or the observed control quantity (y ｏｂｓ (k)) and / or the time derivative of the control amount (y(k)) [Equation 2] At least one of the control variables (y(k)) is the measured time derivative of the control variable (y(k)). [Equation 3] or the observed time derivative of the controlled variable (y(k)) [Equation 4] The method of claim 1 , wherein the determination is made as follows:

3. Actual adjustment amount (u ａｃｔ (k)) is the measured actuator information (u ｍｅａｓｄ，ａ (k)) or directly measured, ｍｅａｓｄ (k)), or the actuator (Σ ａ A model (Σ) that models the dynamic response behavior of the ａ，ｍ 3. The method according to claim 1, wherein the determination is made by (•).

4. The expanded object model (Σ ｍ，ａｕｇ (·)) is the control object (Σ ｐ (・)) target model (Σ ｍ 2. The method of claim 1, wherein (·) is obtained by expanding with a model equation that models the physical power balance.

5. The method comprises: - at least one disturbance quantity (z ｍ (k)) is determined; At least one calculated and / or measured disturbance quantity (z ｍ (k)) is the compensation and / or correction required (d ｒｅｑ 5. The method according to claim 1, wherein the provision of (k) is taken into account.

6. The method is based on a controlled object (Σ ｐ (·)), in particular adapted to operate and control, and / or the coolant outlet temperature from the fuel cell stack (101) is used as the control variable (y(k)), in particular for controlling the coolant temperature difference through the fuel cell stack (101).

7. The method involves the control of a coolant system (100) in the form of a mixing point (102) (Σ ｐ (·)), in particular adapted to operate and control, and / or the control variable (y(k)) is the inlet temperature of the coolant to the fuel cell stack (101).

8. The method comprises controlling a control object (Σ ) in the form of a cooler (103) of a coolant system (100). ｐ (·)), in particular adapted to operate and control, and / or the coolant outlet temperature from the cooler (103) is used as the control variable (y(k)).

9. The method is a control object (Σ ) having a plurality of input quantities and / or a plurality of adjustment quantities and / or a plurality of output quantities and / or a plurality of control quantities. ｐ 9. The method according to any one of claims 1 to 8, characterized in that it is applied to actuate, in particular to control (.)).

10. A computer program product comprising instructions that, when executed by a computer, instruct the computer to perform the method of any one of claims 1 to 9.

11. A controller (110) having a storage unit in which code is stored and a calculation unit, the controller performing the method of any one of claims 1 to 8 when the code is executed by the calculation unit.

12. 12. An electrochemical energy converter, in particular in the form of a fuel cell stack or electrolyzer of a fuel cell system, comprising a controller (110) according to claim 11.

Images