Method for optimizing an operating strategy for operating a fuel cell system, which operating strategy can use exhaust-gas recirculation

Abstract

The invention relates to a method for optimizing an operating strategy for operating a fuel cell system (100) having at least one or more fuel cell stacks (101), which operating strategy uses exhaust-gas recirculation (EGR) in at least one cathode system (10) of the fuel cell system (100), the method comprising: - carrying out optimization (P1) during ongoing operation (BB) of the fuel cell system (100), wherein, in the course of the optimization, operation (BB) of the fuel cell system (100) with exhaust-gas recirculation (EGR) is compared with operation (BB) of the fuel cell system (100) without exhaust-gas recirculation (EGR), and / or - carrying out a defined test (P2), wherein, in the course of the defined test (P2), operation (BB) of the fuel cell system (100) with exhaust-gas recirculation (EGR) is compared with operation (BB) of the fuel cell system (100) without exhaust-gas recirculation (EGR), and - optimizing (Opt) operating parameters (BP) and / or switchover operations (U) with exhaust-gas recirculation (EGR) or without exhaust-gas recirculation (EGR) depending on the optimization (P1) and / or the defined test (P2).

Description

[0001] Description

[0002] title

[0003] Method for optimizing a to operate a can use

[0004] The invention relates to a method for optimizing an operating strategy for operating a fuel cell system that can utilize exhaust gas recirculation. Furthermore, the invention relates to a corresponding computer program product for implementing a corresponding method. Furthermore, the invention relates to a corresponding control unit for implementing a corresponding method. Furthermore, the invention relates to a corresponding fuel cell system with a corresponding control unit.

[0005] State of the art

[0006] A key aspect of fuel cell stack operation is water management in the membrane and thus also in the cathode path. The membrane must be sufficiently moist to conduct protons. The risk of drying out is particularly high in the cathode inlet area. Therefore, various humidification concepts exist (external humidification via membrane humidifiers or water injection, internal stack humidification via anode-cathode interaction). Another humidification concept is exhaust gas recirculation (EGR), in which a portion of the moist stack exhaust gas is directed from the exhaust path into the intake air path. This can increase the humidity of the intake air. Furthermore, the oxygen content of the intake air can be reduced. This also allows a higher stack cathode gas mass flow to be provided at a comparable oxygen content. Disclosure of the Invention

[0007] The present invention provides a method for optimizing an operating strategy for operating a fuel cell system that can utilize exhaust gas recirculation, with the features of the independent method claim. Furthermore, the invention provides a corresponding computer program product, a corresponding control unit, and a corresponding fuel cell system, with the features of the independent claims. Features and details described in connection with the various embodiments and / or aspects of the invention naturally also apply in connection with the other embodiments and / or aspects, and vice versa, so that reciprocal reference is or can always be made to the individual embodiments and / or aspects with regard to the disclosure.

[0008] The present invention provides, according to the first aspect: a method for optimizing an operating strategy for operating a fuel cell system having at least one or more fuel cell stacks, which uses exhaust gas recirculation in at least one cathode system of the fuel cell system, comprising:

[0009] - Carrying out an optimization (P1), in particular by means of a cost function, during ongoing operation of the fuel cell system, wherein the operating strategy is adapted adaptively, preferably online, in particular depending on the optimization, wherein preferably an operating area is expanded in an exploratory manner, wherein when carrying out the optimization, the operation of the fuel cell system with the exhaust gas recirculation and the operation of the fuel cell system without the exhaust gas recirculation are compared, and / or [ie that P1 and P2 can be carried out individually or in combination]

[0010] - Carrying out a defined test (P2), which is in particular specifically triggered and which preferably provides for a subsequent adaptation of the operating strategy, wherein, when carrying out the defined test, the operation of the fuel cell system with the exhaust gas recirculation and the operation of the fuel cell system without the exhaust gas recirculation are compared, and [ie, depending on P1 and / or P2]

[0011] - Optimization of operating parameters and / or switching with or without exhaust gas recirculation depending on the optimization and / or the defined test

[0012] The steps of the method can be performed in the specified order or in a modified order. The steps of the method can be performed simultaneously, at least partially simultaneously, and / or sequentially.

[0013] The fuel cell system (or "system" for short) can preferably be used for mobile applications, for example, in vehicles, particularly fuel-powered vehicles, preferably in the commercial vehicle sector with high service life requirements. The fuel cell system can serve as the main energy supplier for a vehicle. Furthermore, the fuel cell system according to the invention can serve as an energy supply for a power take-off drive and / or an auxiliary drive of a vehicle, for example, a hybrid vehicle. However, the fuel cell system can also be used for stationary applications, for example, in generators.

[0014] The fuel cell system can comprise multiple fuel cell stacks (or stacks for short), each with multiple stacked fuel cells and the associated functional systems, including media systems (air or cathode system, fuel or anode system, cooling system) and an electrical system. Preferably, the fuel cell system can comprise multiple modules in the form of individual stacks with multiple stacked fuel cells.

[0015] The invention recognizes that exhaust gas recirculation (EGR for short) can positively influence both stack performance and stack aging. The invention further recognizes that exhaust gas recirculation can adversely affect the power requirements of the cathode system and possibly other auxiliary consumers (indirectly, for example, on the cooling system through stack cooling, cooling of the air compressors, and / or cooling of the compressed air).

[0016] The invention further recognizes that the use of exhaust gas recirculation across the entire operating range is challenging in:

[0017] - different environmental conditions (e.g. pamb, Tamb, fiamb, etc.),

[0018] - different stack operating conditions (e.g. fiStckln, xO2Stckln, pStck, TStck, mAirStck, aCathout cathode leakage activity, different humidity conditions, etc.),

[0019] - a wide power range (low to high stack current densities),

[0020] - different operating modes (e.g. normal operation, start, stop, freeze start, cold start, warm-up phase, recovery, standby, etc.),

[0021] - stationary and dynamic operating conditions (negative and positive load steps, hysteresis, etc.),

[0022] - different system and component states (aging of stacks / different SOH states, air system, etc.).

[0023] In addition, there are challenges:

[0024] - Switching from operation with exhaust gas recirculation to operation without exhaust gas recirculation, which results in changes in the system behaviour,

[0025] - In multi-stack systems, additional couplings between the individual stacks or individual systems or through non-stack-individual subsystems in the exhaust gas recirculation can be added,

[0026] - Conditions required for precise control of exhaust gas recirculation, e.g., exhaust gas condition, sensors, supply air, stack, etc., which may not be known with sufficient accuracy, e.g., because no sensors are provided for them.

[0027] The process addresses these challenges and ensures that exhaust gas recirculation is used in a beneficial manner. The idea is that exhaust gas recirculation can be used as needed and in a targeted manner, specifically when the benefits outweigh the disadvantages.

[0028] Optimization goals can be the following:

[0029] 1) Optimised efficiency or performance in combination with reduced energy requirements for parasitic power, especially air compression, and / or optimised system efficiency, and / or optimised H2 consumption,

[0030] 2) Accuracy of control variables in exhaust gas recirculation,

[0031] 3) Minimizing degradation, especially of the stacks, and / or

[0032] 4) Optimization dynamics.

[0033] Furthermore, the method may comprise at least one further step:

[0034] - Performing a power balancing by at least one other (i.e., not involved in P1 and / or P2) fuel cell stack and / or an energy storage device, e.g. a battery, e.g. a high-voltage battery, so that the optimization and / or the defined test has / has no significant influence on the overall performance of the fuel cell system and the energy storage device.

[0035] Power balancing can be particularly beneficial for defined tests, as it ensures that overall performance is not affected, or only marginally. This can increase confidence in the system and ensure customer convenience.

[0036] Advantageously, multivariable control can be implemented when optimizing operating parameters and / or switching. This allows for an improved operating strategy that can pursue multiple objectives.

[0037] Furthermore, it can be advantageous if multi-objective optimization, particularly using cost functions, can be performed when optimizing operating parameters and / or switching. This allows different objectives to be balanced against each other, thus providing improved overall system operation.

[0038] Furthermore, restrictions and / or limitations in the operation of the fuel cell system can be taken into account when optimizing operating parameters and / or switching. This allows for systems, subsystems, and components to be operated in a sustainable manner.

[0039] Advantageously, when optimizing operating parameters and / or switching, an operating area can be expanded in an exploratory manner, in particular taking into account restrictions and / or limitations in the operation of the fuel cell system.

[0040] Furthermore, it is conceivable that a control engineering method and / or a method based on machine learning can be used to optimize operating parameters and / or switching.

[0041] The following procedures are conceivable, for example:

[0042] - model predictive control (MPC for short), moving horizon estimation (MHE for short),

[0043] - Exploration methods with learning under constraints,

[0044] - Methods based on reinforcement learning (RL for short),

[0045] - Procedures based on dynamic, active and safe learning (DASL for short, English for “dynamic active safe learning”),

[0046] - Combinations of a control engineering method and a machine learning-based method.

[0047] DASL:

[0048] With DASL, the operating range can be expanded exploratively. Restrictions and / or limitations can be taken into account. A safety model (distance to operating limits or distance to damage potential) can be learned based on safety criteria (e.g., monitoring by CVM, individual cell voltages, or impedance measurements, etc.).

[0049] MPC:

[0050] MPCs fundamentally have the ability to deal with the limitations of systems, subsystems and components and to enable (almost) optimal operation in different operating states as well as an adaptive operating strategy over the lifetime.

[0051] RL:

[0052] Using RL, the operation can be expanded exploratively and optimized according to the same / similar cost functions as with MPC.

[0053] Appropriate definitions can be made for the methods mentioned above:

[0054] - restrictions and / or limitations (for RL: “environment”),

[0055] - Especially for stack and the actuators,

[0056] - Constraints can also be specified for the EGR valve.

[0057] - Control variables (for MPC: “input vector” and / or for RL: “actions”),

[0058] - In particular from the EGR valve and, if applicable, actuators related to the EGR, in particular the actuators of the air system.

[0059] - State variables (for MPC: “state vector x” and / or for RL: “States”),

[0060] - In particular, the condition of the relevant media: exhaust gas before the EGR branch path, supply air before the EGR feed, EGR path and the stack condition with regard to water balance / humidity / SOH / etc.

[0061] - Result variables (for MPC: “output vector y” and / or for RL: “reward” or “variables for reward function”),

[0062] - In particular, stack performance, auxiliary consumer performance, system efficiency, H2 consumption, etc.,

[0063] - Predicted target parameters (paramCtl).

[0064] For processes that require plant models, such as MPC and MHE or hybrid processes, corresponding transmission paths of the system can be modeled. When optimizing operating parameters and / or switching, multivariable control is implemented, in which several physical state variables (paramCtl) are controlled:

[0065] - Oxygen mass flow, oxygen content, lambda, volume flow, molar flow and / or mass flow of supply air and / or exhaust air (or exhaust gas),

[0066] - Pressure, oxygen partial pressure, pressure differences, and / or pressure ratios of supply air and / or exhaust air,

[0067] - Humidity, water content, and / or activity of exhaust air,

[0068] - temperature of supply air and / or extract air,

[0069] - Mass flow, volume flow or molar flow at an inlet point of a purge gas, and / or

[0070] - Activity of supply air and / or exhaust air.

[0071] These variables (setpoint variables or target parameters) can be specified by a control unit (e.g., for water management). The humidity at the cathode stack inlet can be influenced by EGR. The oxygen content of the gas at the cathode inlet can also be influenced by EGR. However, all of the other variables mentioned, such as mass flows and pressures, can also be influenced, which is why EGR can be used in an improved way with multi-variable control. Depending on the system topology, one or more stacks can be influenced by one EGR path. These couplings can be advantageously handled by multi-variable control during optimization. It may not always be possible to control all variables simultaneously. Therefore, it is preferable to prioritize the variables, whereby the prioritization can preferably be made dependent on the operating mode. In normal operation, for example,The oxygen mass flow in the stack represents a high-priority control variable. The reference point can be specified as the input to the cathode path, which may be relevant for the air system. Furthermore, other reference points (stack output or reference point, depending on the respective size) can also be used. When optimizing operating parameters and / or switching, usable power and / or consumption can be optimized.

[0072] When optimizing operating parameters and / or switching, optimization can also be carried out for control accuracy with regard to control variables that are influenced by exhaust gas recirculation.

[0073] When optimizing operating parameters and / or switching, an optimization of a degradation of the at least one or more fuel cell stacks can also be carried out.

[0074] When optimizing operating parameters and / or switching, an optimization of the dynamics of at least one or more fuel cell stacks can also be carried out.

[0075] Preferably, multi-objective optimization can be performed when optimizing operating parameters and / or switching.

[0076] According to a further aspect, the invention provides a computer program product comprising instructions that, when executed by a computer, such as the processing unit of the control unit, cause the computer to perform the method, which can proceed as described above. Using the computer program product, the same advantages can be achieved that were described above in connection with the method according to the invention. These advantages are incorporated herein by reference.

[0077] A corresponding control unit provides a further aspect of the invention. A computer program in the form of code can be stored in a memory unit of the control unit. When the code is executed by a computing unit of the control unit, the program performs a method that can proceed as described above. The control unit can achieve the same advantages as those described above in connection with the method according to the invention. These advantages are incorporated herein by reference in their entirety.

[0078] A corresponding fuel cell system with a corresponding control unit provides a further aspect of the invention. Using the fuel cell system, the same advantages described above in connection with the method according to the invention can be achieved. These advantages are incorporated herein by reference.

[0079] Preferred embodiments:

[0080] The invention and its further developments, as well as their advantages, are explained in more detail below with reference to the accompanying drawings. They show schematically:

[0081] Figure 1 shows an example system topology,

[0082] Figure 2 shows another example system topology,

[0083] Figure 3 shows another exemplary system topology, and

[0084] Figure 4 shows an exemplary sequence of a method according to the invention.

[0085] In the different figures, identical parts of the invention are always provided with the same reference numerals, which is why they are usually only described once.

[0086] Figs. 1 to 4, including in particular Fig. 4, serve to explain a method which is used to optimize an operating strategy for operating a fuel cell system 100 with at least one or more fuel cell stacks 101 which uses exhaust gas recirculation EGR in at least one cathode system 10 of the fuel cell system 100.

[0087] Figures 1 to 3 show various system topologies that can utilize exhaust gas recirculation (EGR). A distinction can be made between

[0088] Fig. 1 : EGR usage within a system 100,

[0089] Fig. 2: EGR use via connecting line(s) in a multistack system,

[0090] Fig. 3: EGR use in a split system 100, in the example of Fig. 3 a cathode system 10 for 2 stacks.

[0091] As shown in Fig. 4, the process comprises:

[0092] - Carrying out an optimization P1, in particular by means of a cost function, during ongoing operation BB of the fuel cell system 100, wherein the operating strategy is adapted adaptively, preferably online, in particular depending on the optimization P1, wherein preferably an operating area can be expanded in an exploratory manner, wherein when carrying out the optimization (P1), the operation BB of the fuel cell system 100 with the exhaust gas recirculation EGR and the operation BB of the fuel cell system 100 without the exhaust gas recirculation EGR are compared, and / or [ie that P1 and P2 can be carried out individually or in combination]

[0093] - Carrying out a defined test P2, which is in particular specifically triggered and which preferably provides for a subsequent adaptation of the operating strategy, wherein when carrying out the defined test P2 the operation BB of the fuel cell system 100 with the exhaust gas recirculation EGR and the operation BB of the fuel cell system 100 without the exhaust gas recirculation EGR are compared.

[0094] Furthermore [depending on P1 and / or P2] the method comprises:

[0095] - Optimize operating parameters (Opt) and / or switchovers (U) with or without exhaust gas recirculation (EGR), depending on the optimization (P1) and / or the defined test (P2). One finding is that exhaust gas recirculation (EGR) can positively influence both stack performance and stack aging.

[0096] However, the invention also recognizes that the exhaust gas recirculation EGR can adversely affect the power requirements of the cathode system 10 and possibly other auxiliary consumers (indirectly, for example, on the cooling system through the stack cooling, cooling of the air compressor Comp, the cooling of the compressed supply air L1, etc.).

[0097] The use of exhaust gas recirculation (EGR) is challenging across the entire operating range of the System 100, particularly in:

[0098] - different conditions in the environment Env (e.g. pamb, Tamb, fiamb, etc.),

[0099] - different stack operating conditions (e.g. fiStckln, xO2Stckln, pStck, TStck, mAirStck, aCathout cathode leakage activity, different humidity conditions, etc.),

[0100] - a wide power range (low to high stack current densities),

[0101] - different operating modes (e.g. normal operation, start, stop, freeze start, cold start, warm-up phase, recovery, standby, etc.),

[0102] - stationary and dynamic operating conditions (negative and positive load steps, hysteresis, etc.),

[0103] - different system and component states (aging of the stacks / different SOH states, air system or cathode system 10, etc.).

[0104] The use of exhaust gas recirculation (EGR) is additionally challenging with regard to:

[0105] - Switching from operation BB with exhaust gas recirculation EGR to operation BB without exhaust gas recirculation EGR, which results in changes in the system behavior,

[0106] - In multi-stack systems, additional couplings between the individual stacks or individual systems (see Fig. 2) or through non-stack-individual subsystems (see Fig. 3) can be added to the exhaust gas recirculation (EGR), - Conditions that are required for precise control of the exhaust gas recirculation (EGR), e.g., condition of the exhaust gas, sensors, intake air, stack, etc., which may not be known with sufficient accuracy, e.g., because no sensors are provided for them.

[0107] Especially in coupled systems 100 in multi-stack applications (e.g. Fig. 2 and 3), the application of switchable exhaust gas recirculation EGR can become complex.

[0108] The method overcomes these challenges and ensures that exhaust gas recirculation (EGR) is used in a beneficial manner.

[0109] The method ensures in particular that the exhaust gas recirculation EGR can be used as needed and in a targeted manner, namely when the advantages of operation BB with the exhaust gas recirculation EGR outweigh the disadvantages of operation BB of the fuel cell system 100 without the exhaust gas recirculation EGR.

[0110] Optimization goals for optimizing Opt of operating parameters BP and / or switching U with exhaust gas recirculation EGR or without exhaust gas recirculation EGR can be the following:

[0111] 1) Optimised efficiency or performance in combination with reduced energy requirements for parasitic power, especially air compression, and / or optimised system efficiency, and / or optimised H2 consumption,

[0112] 2) Accuracy of control variables in exhaust gas recirculation EGR,

[0113] 3) Minimizing degradation, especially of the stacks 101 , and / or

[0114] 4) Optimization of dynamics during operation of the BB system 100.

[0115] Furthermore, Fig. 4 indicates that the method may comprise at least one further step:

[0116] - Performing a power equalization LA by at least one uninvolved fuel cell stack 101 and / or an energy storage device, e.g. a battery, e.g. a high-voltage battery, so that the optimization P1 and / or the defined test P2 have / has no significant influence on the overall performance of the fuel cell system 100 and the energy storage device

[0117] The power compensation LA can be particularly advantageous for defined tests P2, so that the overall performance of the system 100 is not or not noticeably affected.

[0118] Fig. 4 shows an overview of the processes P1 and P2, which can optimize the operation BB with / without exhaust gas recirculation EGR and the switching U by learning and exploring on the left side (P1) and defined tests for gaining knowledge D (labeled data) on the right side (P2).

[0119] The optimization Opt can run based on cost functions KF, which can preferably be adaptable and / or switchable, e.g. depending on certain operating modes and / or on the respective process P1 and / or P2 and / or on the system states (e.g. aging) and / or the specific use case.

[0120] For example, in the case of stacks 101 that are already significantly aged, the optimization focus can be placed more on preventing further aging, whereby a slightly increased H2 consumption can be accepted.

[0121] In both process steps P1 and P2, the power split or power balancing LA between existing stacks 101 and / or energy storage devices can be divided in such a way that the respective required power can be provided as desired and the learning or exploratory processes can still be carried out.

[0122] In both process steps P1 and P2, the BB operation with EGR and the BB operation without EGR can be compared, and both operating variants can be adaptively optimized if necessary. Based on this, the switching between U mil and without EGR can also be optimized.

[0123] Both process steps P1 and P2 can also be combined.

[0124] Both process steps P1 and P2 can access cost functions KF for optimization Opt. Furthermore, these cost functions KF can be defined differently.

[0125] The use of control engineering methods and / or AI / ML methods for optimising Opt can be advantageous, especially due to strong and / or versatile couplings, e.g.

[0126] - in one stack 101 or in several stacks 101 below each other,

[0127] - between supply air L1 and exhaust air L2 through the exhaust gas recirculation EGR, but also through other components (e.g. energy recuperation with turbine T, (gas-gas) heat exchanger HE, etc.),

[0128] - other subsystems, e.g. a cooling system.

[0129] The following procedures are conceivable, for example:

[0130] - model predictive control (MPC for short), moving horizon estimation (MHE for short),

[0131] - Exploration methods with learning under constraints,

[0132] - Methods based on reinforcement learning (RL for short),

[0133] - Procedures based on dynamic, active and safe learning (DASL for short, English for “dynamic active safe learning”),

[0134] - Combinations of a control engineering method and a machine learning-based method.

[0135] DASL:

[0136] With DASL, the operating range can be expanded exploratively. Restrictions and / or limitations can be taken into account. Based on safety criteria such as monitoring by CVM, individual cell voltages, or impedance measurements, a model of the safety distance to operating limits or the distance to damage potential can be learned.

[0137] MPC:

[0138] MPCs fundamentally have the ability to deal with the limitations of systems, subsystems and components and to enable near-optimal operation in different operating states as well as an adaptive operating strategy over the lifetime.

[0139] RL:

[0140] Using RL, the BB operation can be expanded exploratively and optimized according to the same / similar cost functions as in MPC.

[0141] Appropriate definitions can be made for the methods mentioned above:

[0142] - Restrictions and / or limitations for RL: “environment”,

[0143] - Especially for stack and the actuators,

[0144] - Constraints can also be specified for EGR valves.

[0145] - Control variables for MPC: “input vector” and / or for RL: “actions”,

[0146] - In particular EGR valves and possibly actuators related to the EGR.

[0147] - State variables for MPC: “state vector x” and / or for RL: “States”,

[0148] - In particular, the condition of the relevant media: exhaust gas before the EGR branch path, supply air before the EGR feed, EGR path and the stack condition with regard to water balance / humidity / SOH / etc.

[0149] - Result variables for MPC: “output vector y” and / or for RL: “reward” or “variables for reward function”

[0150] - In particular, performance stack, auxiliary consumers, system efficiency, H2 consumption, etc.,

[0151] - Predicted target parameters paramCtl.

[0152] For processes that require plant models, such as MPC and MHE or hybrid processes, corresponding transmission paths of the system can be modeled. When optimizing Opt of operating parameters BP and / or switching U, a multivariable control is implemented, in which several physical state variables paramCtl are controlled:

[0153] - Oxygen mass flow, oxygen content, lambda, volume flow, molar flow and / or mass flow of a supply air L1 and / or an exhaust air L2,

[0154] - Pressure, oxygen partial pressure, pressure differences, and / or pressure ratios of supply air L1 and / or exhaust air L2,

[0155] - Humidity, water content, and / or activity of an exhaust air L2,

[0156] - Temperature of supply air L1 and / or extract air L2,

[0157] - Mass flow, volume flow or molar flow at an inlet point of a purge gas, and / or

[0158] - Activity of supply air L1 and / or extract air L2.

[0159] These variables (setpoint variables or target parameters) can be specified by an ECU control unit (e.g., for water management). The humidity at the cathode stack inlet can be influenced by the EGR. The oxygen content of the gas at the cathode inlet can also be influenced by the EGR. However, all other variables mentioned, such as mass flows and pressures, can also be influenced, which is why the EGR can be used in an improved manner with multi-variable control. Depending on the system topology, one or more stacks 101 can be influenced by an EGR path. These couplings can be advantageously handled by multi-variable control during optimization (Opt). It may not always be possible to adjust all variables simultaneously.Therefore, it is preferable to prioritize the variables, with the prioritization preferably being dependent on the operating mode. For example, during normal operation, the oxygen mass flow in the stack can be a high-priority controlled variable. The reference point can be specified as the input to the cathode path, which may be relevant for the air system. Furthermore, other reference points (stack output or reference point, depending on the respective variable) can also be used. Optimization for energy minimization and thus also consumption minimization:

[0160] When optimizing Opt of operating parameters BP and / or switching U, an optimization of usable power and / or consumption can be carried out.

[0161] The following can be suggested as an optimization function or quality function J for the period tbeg to tend:

[0162] A cost function C1 represents a measure of consumption or the inverse of system efficiency. The smaller the cost function C1, the better.

[0163] The use of exhaust gas recirculation (EGR) can improve stack performance / efficiency, but may also result in increased auxiliary power, particularly in air compression (Comp).

[0164] The cost function C1 can address the trade-off between stack performance / efficiency and auxiliary power, especially the effects with / without exhaust gas recirculation (EGR).

[0165] C1 = C1chem / (C1 a - C1 b) where the power of the stack 101 or the stacks 101 (1 , ..., n) to be taken into account in the respective system network mil / without exhaust gas recirculation EGR: and wherein the secondary consumer services (1 , ..., M), in particular in

[0166] Cathode system 10 are:

[0167] The usable power (net) is the difference between C1 a and C1 b.

[0168] The chemical energy contained in the fuel (here hydrogen) and the resulting maximum power potential (with ideal conversion) C1 can be determined from the fuel properties (upper and lower calorific value) and the converted fuel mass flow.

[0169] In this example, C1 corresponds to the inverse of the system efficiency, which is then minimized in the optimization (or thus the system efficiency is maximized).

[0170] In case of multi-stage compression (as in Fig. 3) the power C1 b can take into account several air compressors Comp.

[0171] C1 b can also be extended, e.g. with the cooling system(s), where in particular the cooling fan(s) and the coolant pumps can contribute to the auxiliary power 1 , ..., K:

[0172] If necessary, additional secondary consumers can also be taken into account.

[0173] C1 and C2 can also be combined and represent the total H2 consumption of all stacks 101 involved in the system with the exhaust gas recirculation EGR:

[0174] The optimization Opt on this cost function C1 can be particularly advantageous if the operating range is to be expanded and optimized in an exploratory manner.

[0175] As an alternative to continuous use, this cost function C1 can also be used at periodic longer intervals in order to adaptively bring aged systems 100 to energy-optimal operation.

[0176] Optimization for control accuracy of the variables that can be influenced by the exhaust gas recirculation (EGR):

[0177] When optimizing Opt of operating parameters BP and / or switching U, an optimization of control accuracy can be carried out with regard to control variables that can be influenced by exhaust gas recirculation EGR.

[0178] If it is known in which operating area and how the exhaust gas recirculation (EGR) is specifically used, or if this has been determined beforehand through targeted P2 tests, or if this is important for diagnostics, e.g., stack SOH diagnostics (SOH for short, English for "state of health"), then optimization can be targeted towards the accuracy of the desired parameters to be controlled and not towards the previously mentioned energy targets.

[0179] The following can be suggested as an optimization function or quality function J for the period tbeg to tend:

[0180] The smaller the cost function C2, the better. The function can account for all participating stacks 101. For example, with inlet humidity fi, the cost function C2 can be defined as follows: C2=C2a:

[0181] Zf ilstS tack[n] — fiSollStack[n\ fiSollStack[n] n=l

[0182] For oxygen content xO2 or oxygen mass flow mO2, the cost function C2 can be defined as follows, C2=C2b:

[0183] When using a lambda sensor in the exhaust gas path, the cost function C2 can be defined as follows, C2=C2c:

[0184] It is also possible to weight the control accuracy individually for each stack. For example, if one stack 101 has aged significantly more than the other, the accuracy requirements can be different based on the weighting. Example for the inlet humidity with a stack-specific weighting parameter beta[n]:

[0185] If several parameters are to be optimized for accuracy, C2 can also contain several target parameters, e.g.: C2 = C2a + C2b + C2c.

[0186] The individual parameters can also be given different weightings, which in turn can be adapted for each operating mode and / or operating state, e.g.: “ ß2a^2a + ß2b ^2b + / 2c c

[0187] Minimizing degradation, especially in the stacks:

[0188] When optimizing operating parameters and / or switching, an optimization of a degradation of the at least one or more fuel cell stacks 101 can also be carried out.

[0189] The exhaust gas recirculation EGR can also be used for the goal of “degradation reduction” of the Stacks 101, which can be very important, especially in the CV (commercial vehicle) sector.

[0190] The cost function C3 can represent a measure of degradation. If the degradation is small, C3 is small; if the degradation is large, C3 is large.

[0191] For example, C3 can be defined as follows (for each stack 101 and for each operating point op, the current state is compared with the state BoL = Begin of Life): If the voltage at the respective operating point is still the same as at the beginning, then C3 is zero.

[0192] The cost function C3 can also consider other values ​​instead of voltages, such as SOH or entire load spectra (including temperatures, dynamic jumps, etc.). Events such as start-stop cycles can also be considered. The various aging values ​​can also be weighted against each other.

[0193] When optimizing operating parameters and / or switching, an optimization of the dynamics of the at least one or more fuel cell stacks 101 can also be carried out.

[0194] Multi-objective optimization:

[0195] Preferably, multi-objective optimization can be performed when optimizing operating parameters and / or switching.

[0196] The following can be suggested as optimization function or total cost function J for the period tO to t1:

[0197] The multi-objective optimization to the minimum of J results in the Pareto front.

[0198] The overall cost function is composed of cost functions for the individual optimization objectives:

[0199] - C1 Energy demand or H2 consumption (description see above)

[0200] - C2 Accuracy of the control for the variables to be controlled (description see above)

[0201] - C3 Minimizing degradation, especially of the stacks

[0202] - C4 if necessary further optimization goals e.g. dynamics

[0203] The individual cost functions for C[n] can, in turn, be composed of several functions or terms. The functions and the signs for the respective cost functions are a matter of definition and are given here only as examples. Furthermore, normalization for all cost terms, e.g., between 0 and 1, may be useful for calculation in the cost function.

[0204] The optimization objectives can be weighted by the weighting factors for the individual cost functions: ß' Weighting factor for optimization objective 1 with cost function C1 ß2 Weighting factor for optimization objective 2 with cost function C2 ß Weighting factor for optimization objective 3 with cost function C3 ß4 Weighting factor for further optimization objectives

[0205] If a cost criterion is not required, the weighting factor can be set to zero. The weighting factors can also be set depending on operating modes and / or system states.

[0206] A corresponding computer program product, a corresponding control unit ecu and a corresponding fuel cell system 100 with a corresponding control unit ecu represent further aspects of the invention.

[0207] The above explanation of the embodiments describes the present invention exclusively by way of examples.

[0208] Of course, individual features of the embodiments can be freely combined with one another, provided that this is technically reasonable, without departing from the scope of the present invention.

Claims

Claims 1 . A method for optimizing an operating strategy for operating a fuel cell system (100) having at least one or more fuel cell stacks (101) that uses exhaust gas recirculation (EGR) in at least one cathode system (10) of the fuel cell system (100), comprising: Carrying out an optimization (P1) during ongoing operation (BB) of the fuel cell system (100), wherein the operating strategy is adapted adaptively, preferably online, in particular depending on the optimization, wherein, when carrying out the optimization, the operation (BB) of the fuel cell system (100) with the exhaust gas recirculation (EGR) and the operation (BB) of the fuel cell system (100) without the exhaust gas recirculation (EGR) are compared, and / or Carrying out a defined test (P2), which is in particular specifically triggered and which preferably provides for a subsequent adaptation of the operating strategy, wherein, when carrying out the defined test (P2), the operation (BB) of the fuel cell system (100) with the exhaust gas recirculation (EGR) and the operation (BB) of the fuel cell system (100) without the exhaust gas recirculation (EGR) are compared, and Optimization (Opt) of operating parameters (BP) and / or switching (U) with exhaust gas recirculation (EGR) or without exhaust gas recirculation (EGR) depending on the optimization (P1) and / or the defined test (P2).

2. The method of claim 1, further comprising: Carrying out a power equalization (LA) by at least one other fuel cell stack (101) and / or an energy storage device, e.g. a battery, e.g. a high-voltage battery, such that the optimization (P1) and / or the defined test (P2) have / has no significant influence on an overall performance of the fuel cell system (100) and the energy storage device.

3. Method according to one of the preceding claims, wherein a multi-variable control is carried out during the optimization (Opt) of operating parameters (BP) and / or switchovers (U), and / or wherein a multi-objective optimization, in particular by means of cost functions, is carried out during the optimization (Opt) of operating parameters (BP) and / or switchovers (U), and / or wherein restrictions and / or limitations in the operation (BB) of the fuel cell system (100) are taken into account during the optimization (Opt) of operating parameters (BP) and / or switchovers (U), and / or wherein an operating area is exploratively expanded, in particular taking into account restrictions and / or limitations in the operation (BB) of the fuel cell system (100).

4. Method according to one of the preceding claims, wherein a control engineering method and / or a method based on machine learning is used for optimizing (Opt) operating parameters (BP) and / or switching (U), for example: model predictive control (MPC), moving horizon estimation method (MHE), exploration method with learning under constraints, method based on reinforcement learning (RL), Procedures based on dynamic, active and safe learning (DASL), Combinations of a control engineering method and a machine learning-based method.

5. Method according to one of the preceding claims, wherein during the optimization (Opt) of operating parameters (BP) and / or switching (U) a multi-variable control is carried out in which several physical state variables are controlled: Oxygen mass flow, oxygen content, lambda, volume flow, molar flow and / or mass flow of a supply air (L1) and / or an exhaust air (L2), Pressure, oxygen partial pressure, pressure differences, and / or pressure ratios of a supply air (L1) and / or an exhaust air (L2), humidity, water content, and / or activity of an exhaust air (L2), temperature of a supply air (L1) and / or an exhaust air (L2), mass flow, volume flow or molar flow at an inlet point of a purge gas, and / or Activity of a supply air (L1) and / or an extract air (L2).

6. Method according to one of the preceding claims, wherein during the optimization (Opt) of operating parameters (BP) and / or switching (U) an optimization of a usable power and / or consumption is carried out.

7. Method according to one of the preceding claims, wherein during the optimization (Opt) of operating parameters (BP) and / or switching (U) an optimization is carried out for control accuracy with respect to controlled variables which are influenced by exhaust gas recirculation (EGR).

8. Method according to one of the preceding claims, wherein during the optimization (Opt) of operating parameters (BP) and / or switching (U), an optimization of a degradation of the at least one or more fuel cell stacks (101) is carried out.

9. Method according to one of the preceding claims, wherein during the optimization (Opt) of operating parameters (BP) and / or switching (U) an optimization of a dynamics of the at least one or more fuel cell stacks (101) is carried out.

10. Method according to one of the preceding claims, wherein a multi-objective optimization, in particular according to one of the preceding claims 6 to 9, is carried out during the optimization (Opt) of operating parameters (BP) and / or switching (U).

11. A computer program product comprising instructions which, when executed by a computer, cause the computer to perform a method according to any one of the preceding claims.

12. Control unit (ecu), comprising a computing unit and a memory unit in which a code is stored which, when at least partially executed by the computing unit, carries out a method according to one of the preceding claims 1 to 10.

13. Fuel cell system (100) comprising a control unit (ecu) according to the preceding claim.

Images