Device and method for determining the composition and / or concentration of bad gases in the fuel of a fuel cell
Patent Information
- Application Number
- PCT/EP2026/053968
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- Priority Date
- 2025-02-20
- Filing Date
- 2026-02-13
- Publication Date
- 2026-08-27
Smart Images
Figure EP2026053968_27082026_PF_FP_ABST
Abstract
Description
[0001] R.415969
[0002] - 1 -
[0003] Description
[0004] title
[0005] Device and method for determining the composition and / or concentration of bad gases in the fuel of a fuel cell
[0006] The invention relates to a device and a method for determining the composition and / or concentration of contaminants in fuel supplied to the anode of a fuel cell of a fuel cell system. The invention also relates to a fuel cell system equipped with a device according to the invention and to a motor vehicle with such a fuel cell system.
[0007] State of the art
[0008] To reduce harmful emissions from motor vehicles, electric motors are increasingly being used in vehicles instead of combustion engines. To supply these electric motors with electrical energy, fuel cells powered by hydrogen can also be used instead of batteries.
[0009] PEM fuel cell systems comprise at least one fuel cell with a polymer electrolyte membrane (PEM) and convert hydrogen into electrical energy using oxygen, generating waste heat and water in the process.
[0010] A PEM fuel cell comprises an anode supplied with hydrogen, a cathode supplied with oxygen or air, and the polymer electrolyte membrane placed between them. Several such individual fuel cells can be stacked to increase the available electrical voltage. Within this stack, also known as a "fuel cell stack," are supply channels that provide the individual cells with hydrogen and air.
[0011] - 2 -
[0012] to remove depleted, i.e. reduced, O2 concentration compared to fresh air, humid air, and the depleted anode exhaust gas.
[0013] To improve the efficiency of the fuel cell system, an approach has become established in which the still hydrogen-rich anode exhaust gas is enriched with fresh hydrogen using gas supply units and then fed back to the anode inlet. This process is called recirculation.
[0014] During operation of the fuel cell, nitrogen from the air supplied to the cathode of the fuel cell passes through diffusion processes to the anode side of the membrane. Nitrogen is an inert gas for the electrochemical reaction taking place in the fuel cell. As such, it reduces the cell voltage and, if present in excessively high concentrations at the anode, can damage the fuel cell membrane because it is no longer adequately supplied with hydrogen.
[0015] From time to time, a portion of the recirculating gas is therefore diverted from the recirculation loop and replaced with fresh hydrogen to reduce the nitrogen concentration in the recirculation loop. This process is called purging. While frequent purging keeps the nitrogen concentration in the recirculation loop low, it simultaneously reduces the efficiency of the fuel cell because fuel in the form of hydrogen is also removed from the system and lost during purging.
[0016] Another source of unwanted gases in the recirculation loop is those contained in the hydrogen-based fuel supplied to the fuel cell's anode. Because the supplied fuel already contains unwanted gases, purging the recirculation loop also introduces new unwanted gases into the anode along with the fresh hydrogen. These unwanted gases, contained in the supplied fuel, can include nitrogen, helium, and argon.
[0017] It is an object of the invention to provide a device and a method that make it possible to determine the composition and / or concentration of contaminants in the fuel supplied to the anode of a fuel cell of a fuel cell system during the operation of a fuel cell system. R.415969
[0018] - 3 -
[0019] Disclosure of the invention
[0020] The invention comprises a training method for training a data model intended to determine the composition and / or concentration of bad gases in fuel supplied to the anode of a fuel cell of a fuel cell system, wherein the fuel cell system comprises an anode recirculation circuit with a recirculation blower.The training procedure comprises operating the fuel cell system successively at different operating points and measuring the electrical power consumption of the recirculation blower at constant speed at each of these operating points; measuring the composition and / or concentration of contaminants in the fuel supplied to the anode of the fuel cell system; and training the data model with the measured values so that, after training, the data model is able to associate a composition and / or concentration of contaminants in the fuel supplied to the anode of the fuel cell system with a profile of the electrical power consumption of the recirculation blower measured during operation of the fuel cell system at constant speed.
[0021] The training procedure may additionally include purging the anode recirculation circuit of the fuel cell system by opening a purge valve; measuring the electrical power consumption of the recirculation fan at constant speed during purging; measuring the composition and / or concentration of contaminants in the fuel supplied to the anode of the fuel cell system; and training the data model with the measured values so that, after training, the data model is able to correlate a composition and / or concentration of contaminants in the fuel supplied to the anode of the fuel cell system with a measured electrical power consumption curve of the recirculation fan at constant speed. R.415969
[0022] - 4 -
[0023] The invention also includes a training method for training a data model intended to determine the composition and / or concentration of bad gases in fuel supplied to the anode of a fuel cell of a fuel cell system, wherein the fuel cell system comprises an anode recirculation circuit with a recirculation blower.The training procedure comprises operating the fuel cell system successively at different operating points and measuring the rotational speed of the recirculation blower at each of these operating points at constant electrical power input; measuring the composition and / or concentration of contaminants in the fuel supplied to the anode of the fuel cell system and training the data model with the measured values, so that after training the data model is able to assign a composition and / or concentration of contaminants in the fuel supplied to the anode of the fuel cell system to a rotational speed of the recirculation blower measured during operation of the fuel cell system at constant electrical power input.
[0024] The training procedure may additionally include purging the anode recirculation circuit of the fuel cell system by opening a purge valve; measuring the rotational speed of the recirculation fan during purging of the anode recirculation circuit at constant electrical power input; measuring the composition and / or concentration of contaminants in the fuel supplied to the anode of the fuel cell system and training the data model with the measured values so that, after training, the data model is able to correlate a composition and / or concentration of contaminants in the fuel supplied to the anode of the fuel cell system with a rotational speed of the recirculation fan measured during purging of the anode recirculation circuit at constant electrical power input.
[0025] The invention also includes a method for determining the composition and / or concentration of bad gases in fuel supplied to the anode of a fuel cell of a fuel cell system, wherein the fuel cell system has an anode recirculation circuit with an R.415969
[0026] - 5 -
[0027] The method comprises operating the fuel cell system; measuring the electrical power consumption of the recirculation fan at constant speed; feeding the electrical power consumption profile of the recirculation fan measured at constant speed into a data model that has been trained using a training method according to the invention; and using the data model to correlate the electrical power consumption profile of the recirculation fan, measured during operation of the fuel cell system at constant speed, with the composition and / or concentration of contaminants in the fuel supplied to the anode of the fuel cell of the fuel cell system.
[0028] The method can further comprise: purging the anode recirculation circuit during operation of the fuel cell system by opening a purge valve; measuring the electrical power consumption of the recirculation blower at constant speed during purging of the anode recirculation circuit; feeding the electrical power consumption of the recirculation blower measured at constant speed during purging of the anode recirculation circuit into a data model that has been trained using a training method according to the invention; and using the data model to correlate the electrical power consumption of the recirculation blower measured at constant speed during purging of the anode recirculation circuit with the composition and / or concentration of contaminants in the fuel supplied to the anode of the fuel cell of the fuel cell system.
[0029] The invention also includes a method for determining the composition and / or concentration of contaminants in fuel supplied to the anode of a fuel cell of a fuel cell system, wherein the fuel cell system comprises an anode recirculation circuit with a recirculation fan. The method comprises operating the fuel cell system; measuring the rotational speed of the recirculation fan during operation of the fuel cell system at constant electrical power input; and feeding the rotational speed of the recirculation fan measured at constant electrical power input into a data model that is processed using a training method according to the invention. R.415969
[0030] - 6 -
[0031] has been trained; and to use the data model to assign the composition and / or concentration of bad gases in the fuel supplied to the anode of the fuel cell of the fuel cell system to the course of the rotational speed of the recirculation blower measured during the operation of the fuel cell system at constant electrical power consumption.
[0032] The invention can further comprise purging the anode recirculation circuit of the fuel cell system by opening a purge valve; measuring the rotational speed of the recirculation blower during purging the anode recirculation circuit at constant electrical power input; feeding the rotational speed of the recirculation blower measured at constant electrical power input during purging the anode recirculation circuit into a data model that has been trained using a training method according to the invention; and using the data model to correlate the rotational speed of the recirculation blower measured during purging the anode recirculation circuit at constant electrical power input with the composition and / or concentration of contaminants in the fuel supplied to the anode of the fuel cell of the fuel cell system.
[0033] The invention further comprises a device for determining the composition and / or concentration of bad gases in fuel supplied to the anode of a fuel cell of a fuel cell system, wherein the fuel cell system comprises an anode recirculation circuit with a recirculation blower.The device comprises an electrical power sensor configured to measure the electrical power consumption of the recirculation fan during operation of the fuel cell system; a speed sensor configured to measure the rotational speed of the recirculation fan during operation of the fuel cell system; and a data model configured and trained to determine the composition and / or concentration of contaminants in the fuel supplied to the fuel cell anode from the measured electrical power consumption profile and / or the measured rotational speed profile of the recirculation fan. The data model can be, in particular, equipped with an R.415969.
[0034] - 7 -
[0035] have been trained according to the training method of the invention, as described above.
[0036] The invention also includes a fuel cell system with at least one fuel cell and a device according to the invention for determining the composition and / or concentration of bad gases in the fuel which is supplied to the anode of a fuel cell of the fuel cell system.
[0037] The invention further comprises a motor vehicle with at least one electric motor and a fuel cell system according to the invention, which is designed and configured to provide electrical energy for driving the electric motor.
[0038] An apparatus and methods according to the invention make it possible to determine the composition and / or concentration of bad gases in fuel supplied to the anode of a fuel cell of a fuel cell system simply and reliably during operation.
[0039] By evaluating the electrical power consumption and / or the rotational speed of the recirculation blower according to the invention, the composition and / or the concentration of bad gases in the fuel supplied to the fuel cell can be determined with good accuracy without having to intervene in the fuel cell system.
[0040] The operation of the fuel cell system can be adapted to the composition and / or concentration of contaminants in the fuel supplied to the fuel cell, as determined according to the invention. For example, the electrical power drawn from the fuel cell system can be limited and / or reduced at high concentrations of contaminants. This reliably prevents inefficient operation of the fuel cell system, excessive wear of the at least one fuel cell, and damage to the at least one fuel cell that could be caused by excessively high concentrations of contaminants. R.415969
[0041] - 8 -
[0042] The purging of at least one fuel cell can be adjusted from tank to tank depending on the concentration of contaminants in the fuel being refueled. If the fuel has an increased concentration of contaminants, the purging intervals and duration can be adjusted accordingly. If compensating for the contaminants in the fuel through prolonged and frequent purging alone is not possible, the maximum electrical current drawn from the fuel cell system can also be limited, as described previously.
[0043] Adjusting the purge time and / or limiting the maximum electrical current drawn from the fuel cell system can be particularly advantageous with new tank systems. Due to manufacturing processes, new tank systems often contain a high proportion of impurities, usually nitrogen. This decreases over time with each refueling, provided the fuel is of sufficiently good quality. The invention makes it possible to adjust the valve switching times accordingly during the start-up and / or shutdown procedure. Furthermore, diagnostics of, for example,
[0044] Components of the gas delivery unit are adapted to the current concentration of contaminants in the fuel.
[0045] The user of the fuel cell system can receive feedback on the actual amount of bad gas, on the basis of which he can assess the quality of the fuel supplied.
[0046] Balances, e.g. leakage diagnoses, can be refined by the proportion of bad gas determined according to the invention.
[0047] The methods and devices according to the invention also make it possible to identify different types of unwanted gases in the fuel, provided that these gases, due to their different material properties, affect the performance of the recirculation blower differently. The unwanted gases can include, in particular, nitrogen, helium, argon, and carbon dioxide.
[0048] In one embodiment, the data model can comprise a neural network. Neural networks have proven to be well-trainable data models that are well-suited for use according to the invention. R.415969
[0049] - 9 -
[0050] In one embodiment, a training method according to the invention comprises training the data model in such a way that a rapidly increasing electrical power consumption of the recirculation blower over time when the purge valve is closed indicates a high nitrogen content in the fuel.
[0051] In one embodiment, a training method according to the invention comprises training the data model such that a rapidly decreasing rotational speed of the recirculation blower over time when the purge valve is closed indicates a high nitrogen content in the fuel.
[0052] In one embodiment, a training method according to the invention comprises training the data model such that a slowly decreasing electrical power consumption of the recirculation blower over time when the purge valve is open indicates a high nitrogen content in the fuel.
[0053] In one embodiment, a training method according to the invention comprises training the data model such that a slowly increasing rotational speed of the recirculation blower over time when the purge valve is open indicates a high nitrogen content in the fuel.
[0054] In one embodiment, a training method according to the invention comprises training the data model such that a slowly increasing electrical power consumption of the recirculation blower over time when the purge valve is closed indicates a high content of helium and / or argon in the fuel.
[0055] In one embodiment, a training method according to the invention comprises training the data model such that a slowly decreasing rotational speed of the recirculation blower over time when the purge valve is closed indicates a high content of helium and / or argon in the fuel.
[0056] In one embodiment, a training method according to the invention comprises training the data model such that a rapidly decreasing electrical power consumption of the recirculation blower over time when the purge valve is open indicates a high helium and / or argon content in the fuel. R.415969
[0057] - 10 -
[0058] In one embodiment, a training method according to the invention comprises training the data model such that a rapidly increasing rotational speed of the recirculation blower over time when the purge valve is open indicates a high content of helium and / or argon in the fuel.
[0059] With a high nitrogen content in the fuel, the power consumption of the recirculation blower can increase approximately three times faster with the purge valve closed, for example, than in the case of a high helium and / or argon content in the fuel.
[0060] With a high helium and / or argon content in the fuel, the power consumption of the recirculation blower with the purge valve open can drop about three times as fast as in the case of a high nitrogen content in the fuel.
[0061] In one embodiment, a training method according to the invention for training a data model further comprises setting or measuring at least one additional parameter of the fuel cell system and taking it into account when training the data model. By taking additional parameters into account, the reliability and accuracy of the results provided by the data model can be further improved.
[0062] In one embodiment, a method according to the invention for determining the composition and / or concentration of bad gases in fuel supplied to the anode of a fuel cell of a fuel cell system comprises setting or measuring at least one further parameter of the fuel cell system and making it available to the data model.
[0063] The additional parameters may include at least one of the following: an electrical voltage of the fuel cell; an electrical current drawn from the fuel cell; a pressure difference of the gas mixture in the anode of the fuel cell; a temperature of the gas mixture in the anode; a mass flow rate of the anode exhaust gases exiting the anode; a pressure of the anode exhaust gases exiting the anode; R.415969
[0064] - 11 -
[0065] a temperature of the anode exhaust gases exiting the anode; a mass flow rate of the cathode exhaust gases exiting the cathode of the fuel cell; a pressure of the cathode exhaust gases exiting the cathode; a temperature of the cathode exhaust gases exiting the cathode; and a rotational speed of the recirculation blower.
[0066] In order to determine at least one of these parameters, a device according to the invention can include at least one of the following sensors: a voltage sensor designed to measure the electrical voltage of the fuel cell; a current sensor designed to measure the electrical current drawn from the fuel cell; at least one pressure sensor designed to measure a pressure difference within the anode; at least one temperature sensor designed to measure a temperature in the anode; at least one mass flow sensor designed to measure the mass flow rate of the anode exhaust gases exiting the anode; at least one pressure sensor designed to measure the pressure of the anode exhaust gases exiting the anode; at least one temperature sensor designed to measure the temperature of the anode exhaust gases exiting the anode;at least one mass flow sensor designed to measure the mass flow rate of the exhaust gases exiting the cathode; at least one pressure sensor designed to measure the pressure of the exhaust gases exiting the cathode; at least one temperature sensor designed to measure the temperature of the exhaust gases exiting the cathode; and at least one speed sensor designed to measure the speed of the recirculation fan.
[0067] In one embodiment, a training method according to the invention for training a data model comprises purging the anode recirculation circuit of the fuel cell system with different opening states and / or with different opening times of the purge valve in order to further increase the amount of data with which the data model is trained according to the invention. This allows the quality of the results provided by the data model to be improved even further. R.415969
[0068] - 12 -
[0069] In one embodiment, a method according to the invention for determining the composition and / or concentration of bad gases in fuel supplied to the anode of a fuel cell of a fuel cell system comprises purging the anode recirculation circuit of the fuel cell system depending on the previously determined composition and / or concentration of bad gases in the fuel supplied to the anode of the fuel cell of the fuel cell system and / or limiting the electrical current that can be drawn from the fuel cell in order to avoid excessive wear and / or damage to the fuel cell.
[0070] A method according to the invention may in particular include purging the fuel cell system with different opening states and / or with different opening times of the purge valve, depending on the previously determined composition and / or concentration of bad gases in the fuel.
[0071] In this way, it can be reliably prevented that the concentration of harmful gases in the fuel cell exceeds a predetermined limit.
[0072] Brief description of the characters
[0073] Figure 1 shows a schematic view of a fuel cell system according to the invention.
[0074] Figure 2 shows a diagram illustrating the time course of the electrical power consumption of a recirculation blower motor and the state of a purge valve during operation of the fuel cell system.
[0075] Figure 3 shows a flowchart illustrating a method according to the invention for determining the composition and / or concentration of bad gases in the fuel supplied to the anode of a fuel cell.
[0076] Figure 4 shows a flowchart illustrating a training method according to the invention for training a data model. R.415969
[0077] - 13 -
[0078] Figure 5 shows a schematic view of a motor vehicle with an electric motor supplied with electrical energy by a fuel cell system according to the invention.
[0079] Character description
[0080] Figure 1 shows a schematic view of a fuel cell system 2 with at least one fuel cell 4, comprising a device 52 according to the invention, which is designed and configured to determine the composition and / or concentration of bad gases in the anode 6 of the at least one fuel cell 4 of the fuel cell system 2 using a previously trained data model 54.
[0081] Very few fuel cells 4 have an anode 6 and a cathode 8.
[0082] Between the anode 6 and the cathode 8 there is a membrane 10, in particular a polymer electrolyte membrane 10.
[0083] Although only a single fuel cell 4 is shown in Figure 1 for the sake of simplicity, the invention can also be used in combination with fuel cell systems 2 that have several fuel cells 4. In particular, several fuel cells 4 of a fuel cell system 2 can be combined to form a fuel cell stack.
[0084] Gaseous oxygen (O2), for example in the form of oxygen-rich air from the environment, is supplied to the cathode 8 of the fuel cell 4 through a cathode air inlet 81. The exhaust gases 80 generated in the cathode 8 during operation of the fuel cell 4 are discharged from the cathode 8 through a cathode outlet 82.
[0085] A mass flow sensor 84 is provided at the cathode outlet 82, which makes it possible to measure the mass flow of the cathode exhaust gases 80 exiting the cathode 8.
[0086] Furthermore, a cathode exhaust pressure sensor 86 and a cathode exhaust temperature sensor 89 are provided at the cathode outlet 82, which make it possible to measure the pressure and temperature of the cathode exhaust gases 80 exiting the cathode 8. R.415969
[0087] - 14 -
[0088] The anode 6 of the least one fuel cell 4 is supplied with gaseous fuel by a hydrogen supply system 12, which mainly contains hydrogen (H2), but also undesirable bad gases, for example nitrogen (N2), helium (He) and / or argon (Ar).
[0089] The hydrogen supplied to the anode 6 of at least one fuel cell 4 reacts with the oxygen supplied to the cathode at or within the membrane 10 to form water (H₂O). During this reaction, an electrical voltage U is generated between electrodes 6a and 8a, which are located in the anode 6 and cathode 8 of the fuel cell 4, respectively. An electric current i can thus be drawn from the at least one fuel cell 4, which makes it possible to drive electrical loads, for example, electric motors.
[0090] The fuel cell system 2 also includes a cooling system 15, which is designed and configured to cool the at least one fuel cell 4 during operation. The cooling system 15 can optionally be operated in a heating mode, in which it heats the at least one fuel cell 4 so that the at least one fuel cell 4 can always be operated at temperatures that are favorable for efficient operation.
[0091] The hydrogen supply system 12 comprises a fuel source 14, for example a hydrogen tank 14 or a hydrogen cylinder 14, which provides gaseous fuel 90, containing in particular hydrogen, via an optional shut-off valve 16. The fuel supplied from the fuel source 14 may also contain undesirable gases, such as nitrogen and / or helium, which are also referred to as harmful gases.
[0092] Since the hydrogen introduced into the anode 6 of a fuel cell 4 does not completely react to form water during a single flow through the anode 6, the hydrogen-containing anode exhaust gases 60 flowing out of the anode 6 are returned to the anode 6 by an anode recirculation circuit 18. This allows a large portion of the hydrogen contained in the anode exhaust gases 60 to be used for generating electrical energy, thus increasing the efficiency of the fuel cell system 2.
[0093] - 15 -
[0094] In the anode recirculation circuit 18, a water separator 20 is provided downstream of a gas outlet 62 of the anode 6 in order to separate water contained in the anode gases 60 exiting the anode 6 from the gas mixture.
[0095] The water separator 20 is equipped with a drain valve 22, which makes it possible to remove the water separated from the water separator 20 by opening the drain valve 22 from the water separator 20.
[0096] During operation of at least one fuel cell 4, nitrogen from the air supplied to the cathode 8 of the fuel cell 4 passes through the membrane 10 from the cathode 8 into the anode 6 of the fuel cell 4 and accumulates over time in the anode recirculation circuit 18. As a result, and due to the undesirable gases already contained in the supplied fuel 90, the proportion of undesirable gases in the gas mixture circulating in the anode recirculation circuit 18 increases during operation of the fuel cell 4 at the expense of the hydrogen content.
[0097] In order to enable the most efficient operation possible of the fuel cell 4, the anode recirculation circuit 18 is regularly purged with hydrogen-rich fuel 90 from the fuel source 14 in order to reduce the proportion of bad gases in the gas mixture circulating in the anode recirculation circuit 18 and to increase the hydrogen content in the gas mixture again.
[0098] A purge valve 24 is provided at a gas outlet 21 of the water separator 20 for purging the anode recirculation circuit 18. By opening the purge valve 24, a gas mixture rich in harmful gases can be released from the anode recirculation circuit 18 to reduce the nitrogen concentration in the anode recirculation circuit 18.
[0099] The fluids discharged from the anode recirculation circuit 18 through the drain valve 22 and the purge valve 24 are mixed with the cathode exhaust gases 80 exiting the cathode 8 of the at least one fuel cell 4 in order to reduce the concentration of hydrogen in the exhaust gases, and are discharged into the environment through an exhaust gas outlet 75.
[0100] - 16 -
[0101] A further gas outlet 23 of the water separator 20 is connected to a suction inlet 27 of a jet pump 26. A drive nozzle in the jet pump 26 is fluidically connected to the fuel source 14 via the shut-off valve 16 and a hydrogen metering valve 28.
[0102] Hydrogen-containing fuel 90, supplied from the fuel source 14 to the drive nozzle of the jet pump 26, drives the jet pump 26. This causes the gas mixture 60 exiting the anode 6 of the at least one fuel cell 4 to be drawn into the suction inlet 27 of the jet pump 26 after passing through the water separator 20. The drawn-in gas mixture mixes in and downstream of the jet pump 26 with the hydrogen-containing fuel 90 supplied from the fuel source 14.
[0103] The gas mixture 38 enriched in this way with hydrogen-containing fuel 90 is then supplied to the anode 6 through a gas inlet 61.
[0104] A recirculation blower 25 is provided between the gas outlet 23 of the water separator 20 and the suction inlet 27 of the jet pump 26. This blower assists the flow of the gas mixture 38 in the anode recirculation circuit 18. The recirculation blower 25 is driven by a motor 29, which is controlled by a motor controller 31. The motor controller 31 includes an electrical power sensor 33, which allows the electrical power consumption P of the motor 29 of the recirculation blower 25 to be measured.
[0105] The recirculation blower 25 is equipped with a speed sensor 48, which makes it possible to measure the speed of the recirculation blower 25.
[0106] In the anode 6 of the fuel cell 4, at least two anode pressure sensors 42a, 42b are provided, which make it possible to determine a pressure difference in the anode 6. A first anode pressure sensor 42a is provided, for example, near the gas inlet 61 of the anode 6, and a second anode pressure sensor 42b is provided, for example, near the gas outlet 62 of the anode 6. R.415969
[0107] - 17 -
[0108] The anode 6 can also be equipped with at least one temperature sensor 41, which makes it possible to measure the temperature of the gas mixture in the anode 6.
[0109] A mass flow sensor 46 is provided between the gas outlet 62 of the anode 6 and the water separator 20, which makes it possible to measure the mass flow of the anode gases 60 exiting the anode 6.
[0110] In addition, a further pressure sensor 42c and a temperature sensor 40 can be provided between the gas outlet 62 of the anode 6 and the water separator 20, which make it possible to measure the pressure and temperature of the anode gases 60 exiting the anode 6.
[0111] An electrical current sensor 45 is provided on at least one of the electrodes 6a, 8a of the at least one fuel cell 4, which makes it possible to measure the electrical current i that is taken from the at least one fuel cell 4.
[0112] Heating elements 50 can be provided at various points in the anode recirculation circuit 18, which are shown by way of example in Figure 1, to heat the flowing gas mixture 38. The positions of the heating elements 50 shown in Figure 1 are only examples and are not exhaustive.
[0113] Figure 1 also shows a control device 55, which is configured to control the components, in particular the valves 16, 26, 22, 24, the recirculation fan 25, the heating elements 50 and / or the cooling system 15 of the fuel cell system 2, in order to operate the fuel cell system 2 in a desired, most efficient operating state. For this purpose, the control device 55 is connected to the valves 16, 26, 22, 24, the recirculation fan 25, the heating elements 50 and / or the cooling system 15 via wireless or wired connections (not shown in Figure 1).
[0114] A fuel cell system 2 according to the invention also comprises a device 52 for determining the composition and / or concentration of bad gases in the fuel 90 that is supplied to the anode 6 of the fuel cell of the fuel cell system. R.415969
[0115] - 18 -
[0116] The device 52 for determining the composition and / or concentration of bad gases in the fuel 90 can be designed as part of the control device 55 or separately from the control device 55.
[0117] The device 52 comprises a data model 54, for example a neural network, which is designed and trained or trainable in such a way that it is able to determine the composition and / or the concentration in the fuel 90 supplied to the anode 6 of the fuel cell 4 from the electrical power input P measured by the electrical power sensor 33 and / or the rotational speed of the motor 29 of the recirculation blower 25 measured by the speed sensor 48.
[0118] With a constant electrical power input P of the electric motor 29, which drives the recirculation blower 25, the speed of the recirculation blower 25 depends on the composition of the gas mixture conveyed by the recirculation blower 25 in the anode recirculation circuit 18.
[0119] Accordingly, if the rotational speed of the recirculation blower 25 is kept constant, the electrical power consumption P of the electric motor 29, which drives the recirculation blower 25, depends on the composition of the gas mixture conveyed by the recirculation blower 25 in the anode recirculation circuit 18.
[0120] Figure 2 contains a diagram which, in the upper section, shows the time course of the electrical power consumption P of the motor 29 of the recirculation blower 25 during operation of the fuel cell system 2 when the speed of the recirculation blower 25 is kept constant. The lower section schematically depicts the state of the purge valve 24. Here, "1" symbolizes an open purge valve 24 and "0" a closed purge valve 24.
[0121] During operation of the fuel cell system 2, the concentration of undesirable gases in the gas mixture circulating in the anode recirculation circuit 18 increases. At a constant rotational speed, this increased concentration of undesirable gases in the gas mixture results in an increase in the electrical power consumption P of the electric motor 29 of the recirculation blower 25. R.415969
[0122] - 19 -
[0123] Opening the purge valve 24 at time ti purges the anode recirculation circuit 18. Purging reduces the concentration of the contaminants in the gas mixture circulating in the anode recirculation circuit 18 relative to the hydrogen content. As a result, the electrical power consumption P of the motor 29 also decreases again, at a constant speed of the recirculation blower 25.
[0124] The rise and fall of the electrical power input P of the motor 29 and the speed of the recirculation blower 25 depend not only on the concentration but also on the type of bad gases in the gas mixture.
[0125] The solid line in Figure 2 illustrates the course of the electrical power consumption P of the motor 29 when the concentration of the bad gases in the gas mixture circulating in the anode recirculation circuit 18 increases only by nitrogen (N2) diffusing through the membrane 10 from the cathode 8 into the anode 6 of the fuel cell 4.
[0126] The dashed line illustrates the course of the electrical power consumption P of the motor 29 when the fuel 90, which is supplied to the anode 6 of the fuel cell 4, contains helium, which is supplied to the anode 6 of the fuel cell 4 together with the hydrogen.
[0127] An increased helium content in the fuel 90, which is supplied to the anode 6, results in an accelerated increase in the electrical power consumption P of the motor 29 when the purge valve 24 is closed in the period to < t < ti, and a delayed decrease in the electrical power consumption P of the motor 29 when the purge valve 24 is open in the period ti < t < t2.
[0128] The dotted line in Figure 2 illustrates the course of the electrical power consumption P of the motor 29 when the fuel 90 supplied to the anode 6 of the fuel cell 4 contains nitrogen, which is supplied to the anode 6 of the fuel cell 4 together with hydrogen.
[0129] An increased nitrogen content in the fuel 90, which is supplied to the anode 6, results in an even faster increase in the electrical power consumption P of the motor 29 when the purge valve 24 is closed during the period to < t < ti and at R.415969
[0130] - 20 -
[0131] With the flushing valve 24 open during the period ti < t < t2, an even more delayed drop in the electrical power consumption P of the motor 29 results.
[0132] Based on the different temporal profiles of the electrical power consumption P of the motor 29 shown in exemplary and schematic form in Figure 2 as a function of the bad gases contained in the fuel 90, a data model 54 trained according to the invention is able to determine the composition and / or the concentration of the bad gases in the fuel 90, which is supplied to the anode 6 of the fuel cell 4, on the basis of the temporal profile of the measured electrical power consumption P of the motor 29.
[0133] The composition and / or concentration of the bad gases in the fuel 90, which is supplied to the anode 6 of the fuel cell 4, can be determined simply and reliably in this way.
[0134] Knowing the composition and / or concentration of the contaminants in the fuel 90 supplied to the anode 6 of the fuel cell 4 makes it possible to adapt the operation of the fuel cell system 2 to the composition and / or concentration of the contaminants in the fuel 90 supplied to the anode 6 of the fuel cell 4. This allows, in particular, excessive wear and / or damage to the fuel cell 4 due to operation with an excessively high concentration of contaminants in the anode 6 to be avoided.
[0135] Figure 3 shows a flowchart illustrating a method 100 according to the invention for determining the composition and / or concentration of bad gases in fuel which is supplied to the anode 6 of a fuel cell 4 of a fuel cell system 2.
[0136] A method 100 according to the invention comprises: operating the fuel cell system 2 (step 110); measuring the electrical power consumption P of the motor 29 of the recirculation blower 25 at constant speed during operation of the fuel cell system 2 (step 120); feeding the electrical power consumption P of the motor 29 of the recirculation blower 25 measured during operation of the fuel cell system 2 at constant speed to the previously trained data model (step 130); and with R.415969
[0137] - 21 -
[0138] With the help of the data model 54, the composition and / or concentration of bad gases in the fuel 90, which is supplied to the anode 6 of the fuel cell 4 of the fuel cell system 2, is to be assigned to the course of the electrical power consumption P of the motor 29 of the recirculation blower 25 measured during the operation of the fuel cell system 2 at constant speed (step 140).
[0139] Optionally, the procedure 100 can additionally include opening a purge valve 24 of the anode recirculation circuit 18 (step 150) to purge the anode recirculation circuit 18 of the fuel cell system 2; measuring the electrical power consumption P of the motor 29 of the recirculation blower 25 at constant speed during the purging of the anode recirculation circuit 18 (step 160); and feeding the electrical power consumption P of the motor 29 of the recirculation blower 25 measured during the purging of the anode recirculation circuit 18 to the previously trained data model (step 170).and, with the help of the data model 54, to assign the composition and / or the concentration of bad gases in the fuel 90, which is supplied to the anode 6 of the fuel cell 4 of the fuel cell system 2, to the course of the electrical power consumption P of the motor 29 of the recirculation blower 25 measured during the purging of the anode recirculation circuit 18 at constant speed (step 180).;
[0140] Method 100 can further comprise adjusting and / or measuring at least one additional parameter of the fuel cell system 2 during operation with the purge valve 24 closed and / or open, and making the at least one additional parameter thus determined available to the data model 54 in order to determine the composition and / or concentration of off-gases in the fuel 90 supplied to the anode 6 of the fuel cell 4 of the fuel cell system 2. The result provided by the data model can be further improved by considering additional parameters of the fuel cell system 2.
[0141] The at least one additional parameter provided to data model 54 may, for example, include at least one of the following parameters: the electrical voltage U of the fuel cell 4; the electrical current i drawn from the fuel cell 4; a pressure difference ApR.415969
[0142] - 22 -
[0143] of the gas mixture in the anode 6 of the fuel cell 4; a temperature TAP of the gas mixture in the anode 6; a mass flow rate riiAn of the anode exhaust gases exiting the anode 6; a pressure PAN of the anode exhaust gases exiting the anode 6; a temperature TAn-aus of the anode exhaust gases exiting the anode 6; a mass flow rate riiKatder of the exhaust gases exiting the cathode 8 80; a pressure pKatder of the exhaust gases exiting the cathode 8 80; a temperature T at of the exhaust gases exiting the cathode 8 80; and a rotational speed n of the recirculation blower 25.
[0144] The purging of the fuel cell system 2 can be carried out with different opening states and / or with different opening times of the purge valve 24.
[0145] The method 100 may additionally include purging the anode recirculation circuit 18 of the fuel cell system 2, depending on the previously determined composition and / or concentration of bad gases in the fuel 90 supplied to the anode 6 of the fuel cell 4 of the fuel cell system 2, i.e., in particular, adjusting the purging times and / or the purging intervals depending on the previously determined composition and / or concentration of bad gases in the fuel 90, and / or limiting the electrical current i that can be extracted from the fuel cell 4 depending on the previously determined composition and / or concentration of bad gases in the fuel 90.
[0146] In this way, increased wear and damage to the fuel cell 4, which can be caused by operation with an excessively high concentration of bad gases in the fuel 90, can be prevented.
[0147] Figure 4 shows a flowchart illustrating a method 200 according to the invention, also referred to as training method 200, for training a data model 54, for example, a neural network. The training method 200 enables the data model 54 to determine, from the electrical power consumption P of the recirculation fan 25 of an anode recirculation circuit 18 measured at constant rotational speed, the composition and / or concentration of contaminants in the fuel supplied to the anode 6 of a fuel cell 4 of a fuel cell system 2. R.415969
[0148] - 23 -
[0149] The training procedure 200 comprises operating the fuel cell system 2 successively at different operating points (step 210) and at each of these operating points: measuring the electrical power consumption P of the motor 29 of the recirculation blower 25 at constant speed during operation of the fuel cell system 2 and determining the course of the electrical power consumption P of the motor 29 of the recirculation blower 25 during operation of the fuel cell system 2 (step 220);to measure the composition and / or concentration of bad gases in the fuel supplied to the anode 6 of the fuel cell 4 of the fuel cell system 2 (step 230) and to train the data model 54 with the measured values (step 240) so that, after completion of the training, the data model 54 is able to assign a composition and / or a concentration of bad gases in the fuel 90 supplied to the anode 6 of the fuel cell 4 of the fuel cell system 2 to a curve of the electrical power consumption P of the motor 29 of the recirculation blower 25 measured during operation of the fuel cell system 2 at constant speed.
[0150] A training method 200 according to the invention can further comprise opening a purge valve 24 of an anode recirculation circuit 18 of the fuel cell system 2 (step 250) in order to purge the anode recirculation circuit 18 of the fuel cell system 2; measuring the electrical power consumption P of the motor 29 of the recirculation blower 25 at constant speed during the purging of the anode recirculation circuit 18 and determining the course of the electrical power consumption P of the motor 29 of the recirculation blower 25 during the purging of the anode recirculation circuit 18 (step 260);to measure the composition and / or concentration of bad gases in the fuel supplied to the anode 6 of the fuel cell 4 of the fuel cell system 2 (step 270) and to train the data model 54 with the measured values (step 280) so that, after completion of the training, the data model 54 is able to assign a composition and / or a concentration of bad gases in the fuel 90 supplied to the anode 6 of the fuel cell 4 of the fuel cell system 2 to a curve of the electrical power consumption P of the motor 29 of the recirculation blower 25 measured during purging of the anode recirculation circuit 18 at constant speed. R.415969;
[0151] - 24 -
[0152] The described training procedure 200 can be carried out at different operating points of the fuel cell system 2 and / or with different fuels 90, wherein the different fuels 90 contain different compositions and / or concentrations of bad gases.
[0153] The data model 54 can be trained in particular such that a rapidly increasing electrical power consumption P of the motor 29 of the recirculation blower 25 when the purge valve 24 is closed indicates a high nitrogen content in the fuel 90, and that a slower increasing electrical power consumption P of the motor 29 of the recirculation blower 25 when the purge valve 24 is closed indicates a high helium and / or argon content in the fuel 90 which is supplied to the anode 6 of the fuel cell 4.
[0154] The data model 54 can be trained such that a slowly decreasing electrical power consumption P of the motor 29 of the recirculation blower 25 when the purge valve 24 is open indicates a high nitrogen content in the fuel 90, and that a rapidly decreasing electrical power consumption P of the motor 29 of the recirculation blower 25 when the purge valve 24 is open indicates a high helium and / or argon content in the fuel 90, which is supplied to the anode 6 of the fuel cell 4.
[0155] The training of the data model 54 with a training method 200 according to the invention can be carried out in particular under laboratory conditions under which all parameters of the fuel cell system 2, in particular the composition and / or the concentration of bad gases in the fuel 90 which is supplied to the anode 6 of the fuel cell 4, can be controlled or measured with high accuracy.
[0156] The data obtained during the training 100 of the data model 54 according to the invention can be used for a large number of structurally identical fuel cell systems 2. It is therefore sufficient to perform the complex training of the data model 54 once on an exemplary fuel cell system 2 in order to later use the data obtained to determine the composition and / or concentration of the fuel 90, which supplies the anode 6 of the fuel cell.
[0157] - 25 -
[0158] 4 is supplied, in order to be able to determine in a large number of identical fuel cell systems 2.
[0159] Optionally, the training procedure 200 can include setting or measuring at least one further parameter of the fuel cell system 2 during operation with the purge valve 24 closed and / or open, and providing the at least one further parameter thus determined to the data model 54 as training data in order to determine the composition and / or concentration of off-gases in the fuel 90 supplied to the anode 6 of the fuel cell 4 of the fuel cell system 2. The result provided by the data model can be further improved by considering additional parameters.
[0160] This at least one further parameter can include at least one of the following parameters: the electrical voltage U of the fuel cell 4; the electrical current i drawn from the fuel cell 4; a pressure difference Ap of the gas mixture in the anode 6 of the fuel cell 4; a temperature TAP of the gas mixture in the anode 6; a mass flow rate riiAn of the anode exhaust gases 60 exiting the anode 6; a pressure PAN of the anode exhaust gases 60 exiting the anode 6; a temperature TAn-aus of the anode exhaust gases 60 exiting the anode 6; a mass flow rate riiKatder of the exhaust gases 80 exiting the cathode 8; a pressure pKatder of the exhaust gases 80 exiting the cathode 8; a temperature T at of the exhaust gases 80 exiting the cathode 8; and a rotational speed n of the recirculation blower 25.
[0161] A training method 200 according to the invention for training the data model 54 can also include flushing the anode recirculation circuit 18 of the fuel cell system 2 with different opening states and / or with different opening times of the purge valve 24 and making the parameters measured thereby available to the data model 54 as training data.
[0162] In the previously described embodiments, a change in the electrical power consumption P of the motor 29 of the recirculation blower 25 at constant speed is evaluated and provided to the data model as an input parameter. In alternative embodiments R.415969
[0163] - 26 -
[0164] The variable speed of the recirculation blower 25 can also be measured and evaluated while maintaining a constant electrical power input P.
[0165] Figure 5 shows a schematic view of a motor vehicle 1 with an electric motor 5, which is supplied with electrical energy by a fuel cell system 2 according to the invention.
[0166] The motor vehicle 1 has four wheels 3 and at least one electric motor 5, which is designed and intended to drive at least two of the four wheels 3 of the motor vehicle 1. The electric motor 5 can also be intended to drive all four wheels 3 of the motor vehicle 1.
[0167] In an alternative embodiment, which is not explicitly shown in the figures, an electric motor 5 can be provided on at least one of the wheels 3, in particular on each of the wheels 3, of the motor vehicle 1 for driving the respective wheel 3.
[0168] The electric motor 5 is supplied with electrical energy via a motor controller 7, which is provided by the fuel cell system 2. The electric motor 5 can also be supplied with electrical energy from a high-voltage battery 13, which is electrically charged by the fuel cell system 2.
[0169] A fuel cell system 2 according to the invention can also be used in motor vehicles 1 that have more or less than four wheels 3.
Claims
R.415969 - 27 - Patent claims 1. Training method (200) for training a data model (54), in particular a neural network, for determining the composition and / or concentration of bad gases in fuel (90) which is supplied to the anode (6) of a fuel cell (4) of a fuel cell system (2), wherein the fuel cell system (2) comprises an anode recirculation circuit (18) with a recirculation blower (25), and wherein the training method (200) comprises: to operate the fuel cell system (2) at different operating points and at each of these operating points: to measure the electrical power consumption of the recirculation fan (25) at constant speed; to measure the composition and / or concentration of bad gases in the fuel (90) supplied to the anode (6) of the fuel cell (4) of the fuel cell system (2); and to train the data model (54) with the measured values so that, after training, the data model (54) is able to assign a composition and / or a concentration of bad gases in the fuel (90) supplied to the anode (6) of the fuel cell (4) of the fuel cell system (2) to a profile of the electrical power consumption of the recirculation blower (25) measured during the operation of the fuel cell system (2).
2. Training method (200) according to claim 1, wherein the training method (200) further comprises, R.415969 - 28 - to purge the anode recirculation circuit (18) of the fuel cell system (2) by opening a purge valve (24); to measure the electrical power consumption of the recirculation blower (25) at constant speed during the purging of the anode recirculation circuit (18); to measure the composition and / or concentration of bad gases in the fuel (90) supplied to the anode (6) of the fuel cell (4) of the fuel cell system (2); and to train the data model (54) with the measured values so that, after training, the data model (54) is able to assign a composition and / or a concentration of bad gases in the fuel (90) supplied to the anode (6) of the fuel cell (4) of the fuel cell system (2) to a profile of the electrical power consumption of the recirculation blower (25) measured during the purging of the anode recirculation circuit (18).
3. Training method (200) for training a data model (54), in particular a neural network, for determining the composition and / or concentration of bad gases in fuel (90) which is supplied to the anode (6) of a fuel cell (4) of a fuel cell system (2), wherein the fuel cell system (2) comprises an anode recirculation circuit (18) with a recirculation blower (25), and wherein the training method (200) comprises: to operate the fuel cell system (2) at different operating points and at each of these operating points: to measure the speed profile of the recirculation fan (25) at constant electrical power consumption; to measure the composition and / or concentration of bad gases in the fuel (90) supplied to the anode (6) of the fuel cell (4) of the fuel cell system (2); and R.415969 - 29 - to train the data model (54) with the measured values so that, after training, the data model (54) is able to assign a composition and / or a concentration of bad gases in the fuel (90) supplied to the anode (6) of the fuel cell (4) of the fuel cell system (2) to a profile of the rotational speed of the recirculation blower (25) measured during the operation of the fuel cell system (2) at constant electrical power consumption.
4. Training method (200) according to claim 3, wherein the training method (200) further comprises, to purge the anode recirculation circuit (18) of the fuel cell system (2) by opening a purge valve (24); to measure the rotational speed of the recirculation blower (25) at constant electrical power consumption during the purging of the anode recirculation circuit (18); to measure the composition and / or concentration of bad gases in the fuel (90) supplied to the anode (6) of the fuel cell (4) of the fuel cell system (2); and to train the data model (54) with the measured values so that, after training, the data model (54) is able to assign a composition and / or a concentration of bad gases in the fuel (90) supplied to the anode (6) of the fuel cell (4) of the fuel cell system (2) to a course of rotational speed of the recirculation blower (25) measured during the purging of the anode recirculation circuit (18) at constant electrical power consumption.
5. Training method (200) according to claim 2 or 4, wherein the training method (200) comprises purging the anode recirculation circuit (18) of the fuel cell system (2) with different opening states and / or with different opening times of the purge valve (24). R.415969 - 30 - 6. Training method (200) according to one of the preceding claims, wherein the training method (200) further comprises adjusting or measuring at least one further parameter of the fuel cell system (2) and taking it into account when training the data model (54); where at least one further parameter includes, for example, at least one of the following parameters: an electrical voltage of the fuel cell (4); an electric current which is taken from the fuel cell (4); a pressure difference of the gas mixture in the anode (6); a temperature of the gas mixture in the anode (6); a mass flow of the anode gases (60) exiting the anode (6); a pressure of the anode gases (60) exiting the anode (6); a temperature of the anode gases (60) exiting the anode (6); a mass flow of the cathode gases (80) exiting from the cathode (8); a pressure of the cathode gases (80) exiting from the cathode (8); and a temperature of the cathode gases (80) exiting the cathode (8) a speed of the recirculation fan (25).
7. Method (100) for determining the composition and / or concentration of bad gases in fuel (90) supplied to the anode (6) of a fuel cell (4) of a fuel cell system (2), R.415969 - 31 - wherein the fuel cell system (2) comprises an anode recirculation circuit (18) with a recirculation blower (25), and wherein the method (100) comprises: to operate the fuel cell system (2); to measure the electrical power consumption of the recirculation blower (25) at constant speed during operation of the fuel cell system (2); to feed the measured course of the electrical power consumption of the recirculation blower (25) during the operation of the fuel cell system (2) into a data model, in particular a neural network; which has been trained with a training method (200) according to one of the preceding claims; and to use the data model (54) to assign the composition and / or concentration of bad gases in the fuel (90) supplied to the anode (6) of the fuel cell (4) of the fuel cell system (2) to the electrical power consumption profile of the recirculation blower (25) measured during the operation of the fuel cell system (2).
8. Method (100) according to claim 7, wherein the method (100) further comprises to purge the anode recirculation circuit (18) of the fuel cell system (2) by opening a purge valve (24); to measure the electrical power consumption of the recirculation blower (25) at constant speed during the purging of the anode recirculation circuit (18); the measured course of the electrical power consumption of the recirculation blower (25) during the purging of the anode recirculation R.415969 - 32 - to feed the circle (18) into a data model; which has been trained using a training method (200) according to one of the preceding claims; and to assign the composition and / or the concentration of bad gases in the fuel (90) supplied to the anode (6) of the fuel cell (4) of the fuel cell system (2) to the electrical power consumption of the recirculation blower (25) measured during the purging of the anode recirculation circuit (18) using the data model (54).
9. Method (100) for determining the composition and / or concentration of bad gases in fuel (90) supplied to the anode (6) of a fuel cell (4) of a fuel cell system (2), wherein the fuel cell system (2) comprises an anode recirculation circuit (18) with a recirculation blower (25), and wherein the method (100) comprises: to operate the fuel cell system (2); to measure the speed of the recirculation fan (25) at constant electrical power consumption during operation of the fuel cell system (2); to feed the measured course of the rotational speed of the recirculation fan (25) into a data model, in particular a neural network; which has been trained with a training method (200) according to one of claims 3 to 6; and Using the data model (54), the composition and / or concentration of bad gases in the fuel (90) supplied to the anode (6) of the fuel cell (4) of the fuel cell system (2) can be assigned to the rotational speed of the recirculation blower (25) measured during operation of the fuel cell system (2). R.415969 - 33 - 10. Method (100) according to claim 9, wherein the method (100) further comprises, to purge the anode recirculation circuit (18) of the fuel cell system (2) by opening a purge valve (24); to measure the rotational speed of the recirculation blower (25) at constant electrical power consumption during the purging of the anode recirculation circuit (18); to feed the measured course of the rotational speed of the recirculation blower (25) during the purging of the anode recirculation circuit (18) into a data model; which has been trained using a training method (200) according to one of claims 3 to 6; and to use the data model (54) to assign the composition and / or concentration of bad gases in the fuel (90) supplied to the anode (6) of the fuel cell (4) of the fuel cell system (2) to the course of the rotational speed of the recirculation blower (25) measured during the purging of the anode recirculation circuit (18).
11. Method (100) according to any one of claims 7 to 10, wherein the method (100) further comprises adjusting or measuring at least one further parameter of the fuel cell system (2) and making it available to the data model (54); where at least one further parameter includes, for example, at least one of the following parameters: an electrical voltage of the fuel cell (4); an electric current which is taken from the fuel cell (4); a pressure difference of the gas mixture in the anode (6); a temperature of the gas mixture in the anode (6);R.415969 - 34 - a mass flow of the anode gases (60) exiting the anode (6); a pressure of the anode gases (60) exiting the anode (6); a temperature of the anode gases (60) exiting the anode (6); a mass flow of the cathode gases (80) exiting from the cathode (8); a pressure of the cathode gases (80) exiting from the cathode (8); and a temperature of the cathode gases (80) exiting the cathode (8); and a speed of the recirculation fan (25).
12. Method (100) according to any one of claims 7 to 11, wherein the method (100) further comprises opening the fuel cell system (2) to different states depending on the previously determined composition and / or concentration of bad gases in the fuel (90). and / or to flush with different opening times of the flushing valve (24) and / or to limit the electrical current that can be drawn from the fuel cell (4).
13. Device (52) for determining the composition and / or concentration of bad gases in fuel (90) supplied to the anode (6) of a fuel cell (4) of a fuel cell system (2), wherein the fuel cell system (2) comprises an anode recirculation circuit (18) with a recirculation blower (25), and wherein the device (52) comprises: a speed sensor (48) designed to measure the speed of the recirculation blower (25) during the operation of the fuel cell system (2); R.415969 - 35 - an electrical power sensor (33) designed to measure the electrical power consumption of the recirculation fan (25) during the operation of the fuel cell system (2); and a data model (54), in particular a neural network, which is designed and trained to be able to determine the composition and / or concentration in the fuel (90) supplied to the anode (6) of the fuel cell (4) from an electrical power input of the recirculation blower (25) measured by the power sensor (33) at constant speed and / or a speed of the recirculation blower (25) measured by the speed sensor (48) at constant electrical power input; wherein the data model (54) has been trained in particular using a training method (200) according to one of claims 1 to 5.
14. Device according to claim 13, wherein the device in particular comprises at least one of the following sensors: a voltage sensor for measuring the electrical voltage of the fuel cell (4); a current sensor (45) for measuring the electric current taken from the fuel cell (4); at least one pressure sensor (42a, 42b) for measuring a pressure difference within the anode (6); at least one temperature sensor (41) for measuring a temperature in the anode (6); at least one mass flow sensor (46) for measuring the mass flow of the anode gases (60) exiting the anode (6); R.415969 - 36 - at least one pressure sensor (42c) for measuring the pressure of the anode gases (60) exiting the anode (6); at least one temperature sensor (40) for measuring the temperature of the anode gases (60) exiting the anode (6); at least one mass flow sensor (84) for measuring the mass flow of the cathode gases (80) exiting the cathode (8); at least one cathode exhaust pressure sensor (86) for measuring the pressure of the cathode exhaust gases (80) exiting the cathode (8); and at least one cathode exhaust temperature sensor (89) for measuring the temperature of the cathode exhaust gases (80) exiting the cathode (8); and wherein the data model (54) is in particular designed to take into account at least one further parameter when determining the composition and / or concentration of bad gases in the anode (6) of the fuel cell (4); where at least one further parameter includes, for example, at least one of the following parameters: - an electrical voltage of the fuel cell (4); an electric current which is taken from the fuel cell (4); a pressure difference of the gas mixture in the anode (6); a temperature of the gas mixture in the anode (6); a mass flow of the anode gases (60) exiting the anode (6); a pressure of the anode gases (60) exiting the anode (6); R.415969 - 37 - a temperature of the anode gases (60) exiting the anode (6); a mass flow of the cathode gases (80) exiting from the cathode (8); a pressure of the cathode gases (80) exiting from the cathode (8); a temperature of the cathode gases (80) exiting the cathode (8).
15. Fuel cell system (2) comprising at least one fuel cell (4) having an anode (6) and a cathode (8), wherein the fuel cell system (2) comprises a device (52) according to claim 13 or 14 for determining the composition and / or concentration of bad gases in the fuel (90) supplied to the anode (6) of a fuel cell (4) of a fuel cell system (2).