Method for operating a power device, computer program and control device for carrying out such a method and power arrangement with such a control device

Abstract

Method for operating a power device (3), wherein a) system values ​​(33) and parameter values ​​(34) of the power device (3) are passed to an estimating module (27), wherein the system values ​​(33) comprise at least measured values ​​(41) on the power device (3), and wherein the parameter values ​​(34) comprise constants characterizing the power device (3), wherein b) the estimating module (27) calculates at least one residual (47) between at least one test measurement of the measured values ​​(41) and an associated estimated test measurement calculated by the estimating module (27) based on the system values ​​(33) and the parameter values ​​(34), wherein c) an error measure (49) is determined based on the at least one residual (47), and wherein d) the error measure (49) is optimized by repeating steps a) to c) while changing the parameter values ​​(34).wherein) the estimation module (27) estimates a state (35) of the power device (3) based on the system values ​​(33) and the parameter values ​​(34), wherein f) the power device (3) is operated based on the estimated state (35), wherein the estimated state (35) is passed to a model predictive control module (29), wherein the model predictive control module (29) calculates at least one manipulated variable (37) for at least one manipulated variable (39) of the power device (3), and wherein the power device (3) is operated based on the at least one manipulated variable (37).

Description

The invention relates to a method for operating a power device, a computer program for carrying out such a method, a control device for carrying out such a method and a power arrangement with such a control device. Power systems with numerous components, such as electrolysis plants for hydrogen production, are difficult to control precisely because of the multitude of individually complex components, and especially because of their intricate interactions. While model predictive control can generally achieve good results, a significant challenge lies in the need for highly precise knowledge of the specific power system to be controlled in order to create a sufficiently accurate model. Although some parameters may be available from datasheets or measurements of individual components, unknown parameters always remain, such as flow resistances influenced by or arising from pipe connections. Furthermore, it may be necessary to adjust certain known parameters to simplify the model.Furthermore, such parameters can change during the operation of a power device, for example, due to the aging of components. An initially or due to aging poorly adapted parameterization of the model underlying the predictive control has an adverse effect on the control quality and thus, in particular, on the efficiency of the controlled power device. DE 10 2019 127 906 A1 discloses a method for operating a power device in which system values ​​and parameter values ​​of the power device are passed to an estimating module, wherein the system values ​​include at least measured values ​​measured on the power device, and wherein the parameter values ​​include constants characterizing the power device, wherein the estimating module calculates at least one residual between at least one test measurement value and an associated estimated test measurement value calculated by the estimating module based on the system values ​​and the parameter values, wherein an error measure is determined based on the at least one residual, wherein the error measure is optimized by repeating the preceding steps while changing the parameter values, and wherein the estimating module estimates a state of the power device based on the system values ​​and the parameter values.and wherein the power device is operated based on the estimated condition. A method for operating a power device is also described in WO 2009 / 067 019 A1. The invention is therefore based on the objective of creating a method for operating a power device, a computer program for carrying out such a method, a control device for carrying out such a method and a power arrangement with such a control device, wherein the aforementioned disadvantages are reduced, preferably avoided. The problem is solved by providing a method according to claim 1, a computer program according to claim 7, a control device according to claim 8, and a power arrangement according to claim 10. Advantageous embodiments are described in the dependent claims and the description. The problem is solved in particular by creating a method for operating a power device, wherein a) system values ​​and parameter values ​​of the power device are passed to an estimating module, wherein the system values ​​include at least measured values ​​measured at the power device, and wherein the parameter values ​​include constants characterizing the power device, wherein b) the estimating module calculates at least one residual between at least one test measurement value of the measured values ​​and an associated estimated test measurement value calculated by the estimating module, based on the system values ​​and the parameter values, wherein c) an error measure is determined based on the at least one residual, wherein d) the error measure is optimized by repeating steps a) to c) while changing the parameter values, and wherein e) the estimating module estimates a state of the power device based on the system values ​​and the parameter values.and wherein f) the power device is operated based on the estimated state. By optimizing the error measure through changes in parameter values, the parameter values ​​themselves are simultaneously modified and thus appropriately adjusted—that is, also optimized—with a view to optimal control of the power device. In this way, mismatches in the parameter values—both initial and age-related—can be reduced, preferably avoided altogether, and a high level of control performance is achieved for the operation of the power device. A further advantage arises from the fact that the estimation module does not need to be specifically configured for a particular application, but can be used modularly for a multitude of different applications, especially for a wide variety of power devices.In this respect, the method can also be used for a large number of applications, in particular for a large number of power devices, without requiring complex adjustments in each individual case. In particular, the estimated test measurement value is calculated by the estimation module based on the system values ​​and the parameter values. By calculating the estimated test measurement value in this way, and by comparing it with the corresponding test measurement value, the at least one residual and from this the error measure can be determined. This allows for the advantageous determination of the quality of the estimation module, that is, in particular, of the model underlying the estimation module. Furthermore, by changing the parameter values ​​while optimizing the error measure, the parameterization of the estimation module, and in particular the model underlying the estimation module, can be advantageously adapted and, in particular, improved. The test measurement value is, in particular, a value actually measured on the power device, and the associated estimated test measurement value is a corresponding value, that is, a value estimated for the same physical quantity by the estimation module. In one embodiment, the at least one residual is calculated by calculating a difference between the estimated test measurement value and the associated test measurement value. In one embodiment, in step b), the estimation module calculates a plurality of residuals between a plurality of test measurements, in particular all measurements, and their respective associated estimated test measurements calculated by the estimation module, based on the system values ​​and the parameter values; preferably, a residual vector is calculated for the plurality of test measurements, in particular for all measurements as test measurements. In this way, a particularly complete and thus high-quality analysis and ultimately an improvement in control performance can be achieved. In particular, in step e), the estimation module estimates a current or updated state of the power device based on the system values ​​and the changed, especially continuously - cyclically - changed, parameter values. Optimizing the error measure specifically involves finding an extremum of the error measure, which, depending on the definition of the error measure, can be either a maximum or a minimum. In one particular implementation, error measure optimization is understood as minimizing the error measure. In one embodiment, the error measure is calculated by forming a norm from the residual vector, for example, a 1-norm or a 2-norm. The norm thus formed is then, in particular, the error measure, or the error measure is calculated from the norm in a suitable manner. However, in a typical power device, the residual vector comprises a plurality of different physical quantities with different ranges of values ​​and different physical units as vector elements. Therefore, it is preferable to perform a suitable normalization of the individual vector elements of the residual vector before forming the norm, such that they are each dimensionless and have a uniform range of values. In particular, the estimation module is based on the model underlying the model predictive control. In the context of the present technical teaching, this preferably means that the estimation module uses this model for calculating the at least one estimated test measurement value and the state estimation of the state of the power device, or that the estimation module is created – particularly mathematically – on the basis of or derived from this model. In the context of this technical teaching, a power device is understood to be, in particular, a device configured to provide power, especially electrical and / or mechanical power, or to convert or consume power. The power device can thus be designed, in particular, as a power provision device or as a power conversion device. Specifically, a power provision device is understood to be a device that provides power, especially electrical and / or mechanical power, using electrical, mechanical, chemical, or electrochemical energy—or another form of energy.A power conversion device is understood to be, in particular, a device that utilizes or consumes power, especially electrical or mechanical power, in particular to convert or store energy, for example, to provide chemical energy in the form of certain substances such as hydrogen or methanol, or electrochemical energy, using electrical energy. Specifically, the power device may be an internal combustion engine, an internal combustion engine-generator combination device (i.e., a genset), a fuel cell, an energy storage device (in particular a battery), or an electrolysis device (in particular an electrolyzer). However, the power device may also be a larger, complex system, for example, consisting of several of the aforementioned devices, or, in particular, a data center or a microgrid.In particular, the power device may also be a controllable or adjustable load on an electrical network. In the context of this technical teaching, a module is generally understood to be, in particular, a conceptually or physically definable or delimited functional unit that is designed to perform at least one specific function. This can be a separate computing device, a part of a computing device, a hardware structure, or a software structure, each of which is designed and intended to fulfill at least one specific function. According to a further development of the invention, historical values ​​of the power device, i.e., values ​​measured in the past, are used as the system values. The estimation module, after optimization of the error measure, is operated with the resulting optimized parameter values, and the power device is operated based on a state estimated by the optimized estimation module. Advantageously, the estimation module can thus be optimized independently of the current operation of the power device—essentially upstream of the operation—so that an optimized estimation module with high control accuracy can be provided for the current operation of the power device. In particular, the estimation module - that is, especially the parameter values ​​- is no longer changed during the operation of the power device; rather, steps b) to d) are only carried out initially before the current operation or commissioning of the power device to optimize the estimation module. In one embodiment, the historical system values ​​are obtained from past operation of other, similar, or identical power devices with respect to their construction and design. It is also possible for the historical system values ​​to be obtained from test bench trials. However, in one embodiment it is also possible that the estimation module of a power device is optimized during its ongoing operation at regular, predetermined time intervals, in particular periodically, or depending on demand, whereby system values ​​from the past operation can then be used for the optimization, and the optimized parameter values ​​are then used for future operation. Alternatively, steps a) to d) are to be carried out continuously during the operation of the power device, wherein the estimation module in step e) is continuously supplied with the current parameter values ​​– and preferably the current measured values ​​– and wherein the power device in step f) is continuously operated based on the state estimated from the current parameter values. Advantageously, this allows for continuous optimization of the parameter values ​​and thus of the estimation module during the actual operation of the power device, so that a continuously adapted control system with very high control performance is available. In one embodiment, steps a) to d) are performed continuously in real time during the operation of the power device, the estimation module is continuously provided with the current parameter values ​​- and preferably the current measured values ​​- in step e) in real time, and the power device is continuously operated in real time in step f) based on the state estimated on the basis of the current parameter values. In one embodiment, the calculation is performed entirely on a control unit of the power device. Alternatively, the calculation can be performed entirely on a separate, particularly remote, computing device, preferably in a cloud. Another alternative is that the calculation can be divided, with certain calculation steps being performed on the control unit of the power device and other calculation steps on the separate computing device, particularly in the cloud. According to a further development of the invention, the system values ​​additionally include control variables and environmental measurements of the power device. Advantageously, this allows for particularly accurate results and a particularly high level of control accuracy. A control variable is understood to be, in particular, a quantity that directly or indirectly determines the position or functional state of an actuator of the power device. An environmental measurement is understood to be, in particular, a value measured in the vicinity of the power device, for example, air pressure, air temperature, and the like. According to a further development of the invention, it is provided that the at least one residual is calculated in which the state of the power device is estimated by the estimation module on the basis of the system values ​​and the parameter values, wherein the at least one estimated test measurement value is calculated by the estimation module on the basis of the estimated state, wherein the at least one estimated test measurement value is compared with the associated test measurement value, wherein in particular the difference between the at least one estimated test measurement value and the associated test measurement value is formed. The estimation module is thus specifically configured to estimate the state of the power device, and is additionally configured to calculate estimated measured values ​​from the estimated state – as it were, in reverse. In particular, the estimation module is further configured to calculate an estimated state from the system values, which include measured values, and then to calculate estimated measured values ​​from the calculated estimated state – in reverse – which can ultimately be compared with the measured values ​​to calculate the at least one residual. According to a further development of the invention, a Luenberger observer or a Kalman filter, in particular a stationary Kalman filter or an extended Kalman filter, is used as the estimation module. A Luenberger observer or, in particular, a Kalman filter are especially suitable for state estimation that is both very accurate and computationally inefficient. The advantages described here are realized in a particularly significant way in connection with an extended Kalman filter as the estimation module. According to a further development of the invention, an electrolysis device, in particular an electrolyzer, is used as the power supply. The advantages already mentioned are particularly evident in this context, since electrolysis devices typically comprise a large number of components that interact in a complex manner and are difficult to control, especially when they are to be operated on a power grid with fluctuating power and / or with varying control objectives. For example, a control objective in the case of fluctuating power in the power grid, due to a high proportion of renewable energies, might be to provide the best possible support for the grid with flexible, variable load reduction, while the control objective in the case of a permanently high grid load might be to draw as much power from the grid as possible.Alternatively or additionally, a regulatory objective – particularly in both scenarios – can be to produce as much hydrogen as possible per unit of power consumed, i.e., to achieve the highest possible efficiency. In one embodiment, the parameter values ​​comprise at least one parameter selected from a group consisting of a cell stack flow resistance, a hydrogen diffusion coefficient, an oxygen diffusion coefficient, an active cell area, a cell membrane thickness, a drag factor, an exchange current reference value, a contact activation energy, a contact activation temperature reference value, a membrane water content, a charge transfer coefficient, an electrical cell resistance, a maximum current density, a cell stack radiation constant, a cell stack heat capacity, a flow resistance of at least one component of the power device through which a fluid, for example a gas, a liquid or a gas-liquid mixture, flows, and a combination of at least two of the aforementioned parameters. In one embodiment, parameter values ​​include all flow resistances of all components of the power device through which the flow passes. Alternatively, the power device can be an internal combustion engine, an internal combustion engine-generator combination device (i.e., a genset), a fuel cell, or an energy storage device, particularly a battery. The power device can also be a larger, complex system, for example, consisting of several of the aforementioned devices, or, in particular, a data center or a microgrid. Specifically, the power device can also be a controllable or adjustable load on an electrical network. According to the invention, the estimated state is passed to a model predictive control module, which calculates at least one manipulated variable for at least one actuator of the power device based on the estimated state, and operates the power device based on this at least one manipulated variable. A particularly advantageous feature is that the model predictive control module enables high-quality control with high control accuracy while requiring minimal computational effort. The control system benefits significantly from the high accuracy of the state estimated based on the optimized parameter values. A control variable is understood to be, in particular, a specification for controlling at least one actuator to achieve a specific control variable. The control variable can, in particular, be a direct control variable for an actuator, that is, a variable that is used directly to control the actuator. Alternatively, the control variable can be a variable on which a further control variable, in particular a direct control variable for an actuator, is determined, and in particular calculated. In one embodiment, the control variable is a valve position for a control valve on a water tank serving as a gas separator. The at least one actuator can therefore be, in particular, an actuator or control element of the power device. Specifically, the actuator can be a valve or a flap. However, the actuator can also be an actuator outside the power device, for example, an actuator intended to influence an externally supplied cooling circuit, such as a valve, a pump, or the like, or an actuator of an electrical device with which the power device is operatively connected. The problem is also solved by creating a computer program that includes machine-readable instructions by virtue of which a method according to the invention or a method according to one or more of the embodiments described above is carried out when the computer program runs on a computing device, in particular a control device according to the invention for a power device or a control device according to one or more of the embodiments described below. The advantages that have already been explained in connection with the method arise particularly in connection with the computer program. The invention also includes a data carrier on which a computer program according to the invention or a computer program according to one or more of the previously described embodiments, in particular in machine-readable form, is stored. In one embodiment, the data carrier is designed as a storage device – in particular electronic – especially as a hard drive, SSD, flash memory, magnetic tape, floppy disk or optical disc. The problem is also solved by creating a control device for a power device, wherein the control device is configured to carry out a method according to the invention or a method according to one or more of the embodiments described above. In connection with the control device, the advantages are particularly those that have already been explained in connection with the method or the computer program. In one embodiment, the control device is a control device for an electrolysis device. In one embodiment, the control device includes the estimation module. In one embodiment, the control device additionally features the model predictive control module. The control device is specifically designed to operate the power device. In one embodiment, the control device is designed to operate an internal combustion engine, an internal combustion engine-generator combination device, a fuel cell, an energy storage device, in particular a battery, an electrolysis device, in particular an electrolyzer, a data center or microgrid, or another controllable or adjustable load on an electrical network. According to a further development of the invention, the control device includes an optimization module configured to change the parameter values ​​in step d). Advantageously, the optimization module is configured to optimize the error measure by changing the parameter values, or—in other words—to change the parameter values ​​in such a way as to optimize the error measure. The problem is also solved by creating a power arrangement comprising a power device and a control device according to the invention, or a control device according to one or more of the embodiments described above, which is operatively connected to the power device for its control. In connection with the power arrangement, the advantages that have already been explained in connection with the method, the computer program, or the control device become particularly apparent. According to a further development of the invention, the power device is designed as an electrolysis device, in particular as an electrolyzer. Alternatively, the power device can also be designed as an internal combustion engine, as an internal combustion engine-generator combination device, as a fuel cell, as an energy storage device, in particular a battery, as a data center or microgrid, or as another controllable or adjustable load on an electrical network. The invention is explained in more detail below with reference to the drawings. Figure 1 shows a schematic representation of an embodiment of a power arrangement; Figure 2 shows a schematic representation of an embodiment of a control device for the power device according to Figure 1; and Figure 3 shows a schematic representation of an embodiment of a method for operating the power device according to Figure 1. Fig. 1 shows a schematic representation of an embodiment of a power arrangement 1 with a power device 3 designed as an electrolysis device 2 - in particular as an electrolyzer - and a control device 5. The control device 5 is specifically designed to operate the power device 3. It is schematically indicated that the control device 5 is configured to apply or specify a voltage across a cell stack 7 of the power device 3. For the sake of clarity, other functional connections between the control device 5 and other components are not explicitly shown here. Process water 9 is supplied to the cell stack 7 for electrolysis, where it absorbs heat from a product stream 13 of the power device 3 in a first heat exchanger 11. The product stream 13 comprises, in particular, hydrogen and residual water. It is possible that the process water 9 is purified upstream of the first heat exchanger 11 for electrolysis and collected in a storage tank (not shown). The product stream 13 is preferably passed downstream of the first heat exchanger 11 via at least one gas separator (not shown) to separate the hydrogen from the residual water, and optionally via a further heat exchanger. The residual water separated in the at least one gas separator can be at least partially returned to the process water stream, preferably at a feed point between the first heat exchanger 11 and a water tank 15. The process water 9 is passed downstream of the first heat exchanger 11 through the water tank 15, which also serves as a gas separator; downstream of the water tank 15, the process water is introduced into the cell stack 7. In the cell stack 7, the process water 9 is electrochemically split into hydrogen and oxygen 16 in a manner known per se, with the hydrogen being carried away with the product stream 13, and the oxygen 16 being carried away with a waste stream 17 containing residual water. The waste stream 17 is directed via a second heat exchanger 19 into the water tank 15, where, in a first stage, residual water is separated from the oxygen 16, and the separated residual water can then be fed back into the cell stack 7 as process water 9. In an embodiment not shown here, the heat extracted from the mass flow 17 in the second heat exchanger 19 can be supplied to the process water 9, in particular via a bypass heat exchanger arranged in a bypass path (not shown), which is connected to the second heat exchanger 19 via a heat transfer medium flow of a coolant 20. The bypass path can branch off from the process water flow between the water tank 15 and the cell stack 7 and rejoin the process water flow between the first heat exchanger 11 and the water tank 15. It is possible that the process water 9 flowing through the bypass path is purified again downstream of the second heat exchanger; alternatively or additionally, it is possible that the bypass heat exchanger arranged in the bypass path is cooled by means of a recooling device. This recooling then also has at least an indirect effect on the second heat exchanger 19.Alternatively or additionally, the second heat exchanger 19 can be directly recooled. In this respect, a recooling device is schematically depicted here as a third heat exchanger 21, which is fluidically connected to the second heat exchanger 19 by the heat exchanger fluid flow of the coolant 20, and which can optionally also recool the bypass heat exchanger. In this respect, the bypass heat exchanger can also be referred to as a "third heat exchanger". The water-depleted waste stream 17 flows from the water tank 15 to a fourth heat exchanger 23, which is cooled, in particular recooled, by a coolant (not shown). From the fourth heat exchanger 23, the waste stream 17 flows to a gas separator 25 for the separation of the oxygen 16 from the residual water in a second stage. The oxygen 16 is preferably discharged into the environment, while the separated residual water is fed back into the process water stream, which is schematically represented here by a flow-related connection between the gas separator 25 and the first heat exchanger 11. Preferably, the separated residual water is discharged from the gas separator 25 into the storage tank (not shown). To generate and / or maintain the various fluid flows, conveying devices, preferably pumps, are provided at suitable locations in a manner known in themselves; these are not shown here for the sake of clarity. Fig. 2 shows a schematic representation of an embodiment of the control device 5 for the power device 3 according to Fig. 1 . Identical and functionally equivalent elements are provided with the same reference symbols in all figures, so that reference is made to the preceding description in each case. The control device 5 is configured to carry out an embodiment of a method for operating the power device 3, which is described in more detail below. In particular, the control device 5 shown here comprises an estimation module 27, a model predictive control module 29, and an automation module 31. The estimation module 27 is preferably an extended Kalman filter, but it can also be a Luenberger observer or another Kalman filter, in particular a stationary Kalman filter. Within the framework of the procedure for operating the power device 3, the estimating module 27 receives, in particular from the automation module 31, system values ​​33 of the power device 3 and parameter values ​​34 characterizing the power device 3. The estimating module 27 uses the system values ​​33 and the parameter values ​​34 to estimate a state 35 of the power device 3. The power device 3 is then operated based on the estimated state 35, in particular by passing the estimated state 35 to the model predictive control module 29. The model predictive control module 29 then calculates actuator values ​​37 based on the estimated state 35 and passes them to the automation module 31 for controlling the power device 3. The automation module 31 calculates actuator values ​​39 from the actuator values ​​37 and uses them to control the power device 3. The parameter values ​​34 can also be initially predefined – having been optimized beforehand – and cannot be changed during operation of the power device 3. They can then be permanently implemented in the automation module 31 or in the estimation module 27, in which case they do not need to be passed from the automation module 31 to the estimation module 27. Alternatively, the parameter values ​​34 can also be optimized during the runtime of the power device 3, in particular by the estimation module 27 itself or by an optimization module 28, which can include the optimization module 28. In this case, the parameter values ​​34 can be initially passed by the automation module 31, or initial parameter values ​​34 can be implemented in the estimation module 27 and are only adjusted internally within it. The optimized parameter values ​​34 are preferably passed to the model predictive control module 29 and / or the automation module 31. The system values ​​33 comprise measured values ​​41 at the power device 3. In particular, these are measured values ​​measured at the power device 3 with real physical sensors, for example, pressures, temperatures, mass flows, electrical voltages and currents, and the like. Preferably, the system values ​​33 also include the manipulated variables 39 and / or environmental measured values ​​43 of the power device 3. The model predictive control module 29 preferably receives additionally from the automation module 31 the manipulated variables 39 and constraints 45 or limits of the actuators of the power device 3, as well as optionally status information 51 about an operating state or mode of the power device 3. The control device 5 also includes the optimization module 28, which is configured to modify the parameter values ​​34 in order to optimize an error measure, explained below, by changing the parameter values ​​34. As already stated, the optimization module 28 can be integrated into or be part of the estimation module 27. Fig. 3 shows a schematic representation of an embodiment of a method for operating the power device 3 according to Fig. 1 . In the first step of the process, the estimation module 27 receives the system values ​​33, in particular the measured values ​​41, and the parameter values ​​34 of the power device 3. The estimation module 27 also preferably receives the manipulated variables 39 and the environmental measured values ​​43. Based on the system values ​​33 and the parameter values ​​34, and especially also the manipulated variables 39 and environmental measured values ​​43, the estimation module 27 calculates residual values ​​47 between test measured values ​​of the measured values ​​41 and corresponding estimated test measured values ​​calculated by the estimation module 27. Based on the residual values ​​47, an error measure 49 is determined in a second step S2, and the parameter values ​​34 are modified in a third step S3 with optimization, in particular minimizing the error measure 49. Specifically, the error measure is optimized iteratively by repeating steps S1 to S3 while changing the parameter values ​​34. The optimization can be performed, in particular, using an evolutionary algorithm or simulated annealing. Using an evolutionary algorithm, the parameter values ​​34 converge to globally optimal parameters. Furthermore, it advantageously allows for embedding in a real-time environment. In a fourth step, the estimation module 27 estimates the state 35 of the power device 3, in particular based on the system values ​​33 and the changed parameter values ​​34, and the power device 3 is operated based on the estimated state 35. The system values ​​33 can be historical values ​​of the power device 3, that is, in particular values ​​measured in the past, wherein the estimating module 27 is then operated as an optimized estimating module 27 after optimization of the error measure 49 with the obtained, optimized parameter values ​​34, and wherein the power device 3 is operated on the basis of the state 35 estimated by the estimating module 27 in this way. Alternatively, the estimation module 27 is optimized during the ongoing operation of the power device 3 at regular, predetermined time intervals, in particular periodically, or as required, whereby system values ​​33 from the past operation can then be used for the optimization, and wherein the optimized parameter values ​​34 are then used for future operation. Alternatively, the method described here can be carried out continuously during the operation of the power device 3, wherein the current, optimized parameter values ​​34 - and preferably the current measured values ​​41 - are continuously transmitted to the estimating module 27, and wherein the power device 3 is continuously operated on the basis of the state 35 estimated on the basis of the current parameter values ​​34, wherein the method is carried out in particular in real time, preferably - as shown - on the control unit 5, or alternatively also on a separate, in particular remote, computing device, preferably in the cloud, or split partly on the control unit 5 and partly on the separate computing device, in particular in the cloud. The parameter values ​​34 preferably comprise at least one parameter selected from a group consisting of a cell stack flow resistance of the cell stack 7, a hydrogen diffusion coefficient, an oxygen diffusion coefficient, an active cell area of ​​the cell stack 7, a membrane thickness of a cell membrane of the cell stack 7, a drag factor, an exchange current reference value, a contact activation energy, a contact activation temperature reference value, a membrane water content, a charge transfer coefficient, an electrical cell resistance, a maximum current density, a cell stack radiation constant of the cell stack 7, a cell stack heat capacity of the cell stack 7, and a flow resistance of at least one component of the power device 3 through which a fluid, for example a gas, a liquid, or a gas-liquid mixture, flows.and a combination of at least two of the aforementioned parameters.

Claims

Method for operating a power device (3), wherein a) system values ​​(33) and parameter values ​​(34) of the power device (3) are passed to an estimating module (27), wherein the system values ​​(33) comprise at least measured values ​​(41) on the power device (3), and wherein the parameter values ​​(34) comprise constants characterizing the power device (3), wherein b) the estimating module (27) calculates at least one residual (47) between at least one test measurement of the measured values ​​(41) and an associated estimated test measurement calculated by the estimating module (27) based on the system values ​​(33) and the parameter values ​​(34), wherein c) an error measure (49) is determined based on the at least one residual (47), and wherein d) the error measure (49) is optimized by repeating steps a) to c) while changing the parameter values ​​(34).wherein) the estimation module (27) estimates a state (35) of the power device (3) based on the system values ​​(33) and the parameter values ​​(34), wherein f) the power device (3) is operated based on the estimated state (35), wherein the estimated state (35) is passed to a model predictive control module (29), wherein the model predictive control module (29) calculates at least one manipulated variable (37) for at least one manipulated variable (39) of the power device (3), and wherein the power device (3) is operated based on the at least one manipulated variable (37). The method of claim 1, wherein the system values ​​(33) are historical values ​​of the power device (3), wherein the estimating module (27) is operated as an optimized estimating module (27) after optimization of the error measure (49) with the obtained, optimized parameter values ​​(34), and wherein the power device (3) is operated on the basis of a state (35) estimated by the optimized estimating module (27), or wherein steps a) to d) are carried out continuously during the operation of the power device (3), and wherein the estimating module (27) is continuously given the current parameter values ​​(34) in step e), and wherein the power device (3) is continuously operated on the basis of the state (35) estimated on the basis of the current parameter values ​​(34) in step f). Method according to one of the preceding claims, wherein the system values ​​(33) additionally include control variables (39) and environmental measurement values ​​(43) of the power device (3). Method according to one of the preceding claims, wherein the at least one residual (47) is calculated in which the state (35) of the power device (3) is estimated by the estimation module (27) on the basis of the system values ​​(33) and the parameter values ​​(34), wherein the at least one estimated test measurement value is calculated by the estimation module (27) on the basis of the estimated state (35), wherein the at least one estimated test measurement value is compared with the associated test measurement value. Method according to one of the preceding claims, wherein the estimation module (27) is a Luenberger observer or a Kalman filter. Method according to one of the preceding claims, wherein the power device (3) is an electrolysis device (2). Computer program comprising instructions by virtue of which a method according to any one of claims 1 to 6 is carried out when the computer program is running on a computing device. Control device (5) for a power device (3), configured to carry out a method according to one of claims 1 to 6. Control device (5) according to claim 8, comprising an optimization module (28) configured to change the parameter values ​​(34) in step d). Power arrangement (1) comprising a power device (3) and a control device (5) according to one of claims 8 or 9. Power arrangement (1) according to claim 10, wherein the power device (3) is designed as an electrolysis device (2).

Images