阅读说明：本技术 通过确定受控参量的实际值来控制燃料电池反应物的调控单元的受控参量 (Controlling a controlled variable of a control unit of a fuel cell reactant by determining an actual value of the controlled variable ) 是由 J·坎萨尔 S·亚库贝克 C·库格勒 于 2018-11-30 设计创作，主要内容包括：为了确定燃料电池反应物的受控参量的至少一个实际值,以用于以尽可能小的测量误差进行控制,规定：利用调控单元(3)的模型<Image he="74" wi="145" file="DDA0002522118230000011.GIF" imgContent="drawing" imgFormat="GIF" orientation="portrait" inline="no"></Image>来计算受控参量模型值<Image he="76" wi="222" file="DDA0002522118230000012.GIF" imgContent="drawing" imgFormat="GIF" orientation="portrait" inline="no"></Image>利用用于测量传感器(Sn)的传感器模型<Image he="79" wi="143" file="DDA0002522118230000013.GIF" imgContent="drawing" imgFormat="GIF" orientation="portrait" inline="no"></Image>来计算受控参量实际值的模型值<Image he="77" wi="262" file="DDA0002522118230000014.GIF" imgContent="drawing" imgFormat="GIF" orientation="portrait" inline="no"></Image>由利用测量传感器(Sn)测量出的受控参量实际值(RGnist)和受控参量实际值的利用传感器模型<Image he="77" wi="153" file="DDA0002522118230000015.GIF" imgContent="drawing" imgFormat="GIF" orientation="portrait" inline="no"></Image>计算出的模型值<Image he="77" wi="240" file="DDA0002522118230000016.GIF" imgContent="drawing" imgFormat="GIF" orientation="portrait" inline="no"></Image>来计算用于受控参量(RGn)的校正值(RGcorr),并且作为所述校正值(RGcorr)与利用调控单元(3)的模型<Image he="76" wi="153" file="DDA0002522118230000017.GIF" imgContent="drawing" imgFormat="GIF" orientation="portrait" inline="no"></Image>计算出的受控参量模型值<Image he="75" wi="208" file="DDA0002522118230000018.GIF" imgContent="drawing" imgFormat="GIF" orientation="portrait" inline="no"></Image>的总和来计算所述至少一个受控参量的实际值<Image he="77" wi="262" file="DDA0002522118230000019.GIF" imgContent="drawing" imgFormat="GIF" orientation="portrait" inline="no"></Image>其中,所述受控参量的实际值<Image he="77" wi="236" file="DDA00025221182300000110.GIF" imgContent="drawing" imgFormat="GIF" orientation="portrait" inline="no"></Image>也在传感器模型<Image he="77" wi="146" file="DDA00025221182300000111.GIF" imgContent="drawing" imgFormat="GIF" orientation="portrait" inline="no"></Image>中用于计算受控参量实际值的模型值<Image he="77" wi="259" file="DDA00025221182300000112.GIF" imgContent="drawing" imgFormat="GIF" orientation="portrait" inline="no"></Image>(In order to determine at least one actual value of a controlled variable of a fuel cell reactant for controlling with as little measurement error as possible, provision is made for: model using regulatory unit (3) To calculate the controlled parameter model value Using a sensor model for measuring the sensor (Sn) To calculate a model value of an actual value of the controlled parameter Sensor model of the actual value (RGnist) of the controlled variable measured by the measuring sensor (Sn) and the actual value of the controlled variable Calculated model value To calculate a correction value (RGcorr) for the controlled variable (RGn) and to use the correction value (RGcorr) as a model for the control unit (3) Calculated controlled parameter model value To calculate the actual value of said at least one controlled quantity Wherein the actual value of the controlled quantity Also in the sensor model Model value for calculating actual value of controlled parameter)

1. Method for determining an actual value of at least one controlled variable (RGn) of a control unit (3) for a reactant of a fuel cell (2), wherein a measured value of the actual value (RGnist) of the controlled variable is measured using a measuring sensor (Sn), characterized in that a model of the control unit (3) is usedTo calculate the controlled parameter model valueUsing a sensor model for measuring the sensor (Sn)To calculate a model value of an actual value of the controlled parameterControlled parameters measured by means of a measuring sensor (Sn)Sensor model for measuring actual values (RGnist) and controlled variablesCalculated model valueTo calculate a correction value (RGcorr) for the controlled variable (RGn) and to compare the actual value of the at least one controlled variableCalculated as said correction value (RGcorr) and a model using a control unit (3)Calculated controlled parameter model valueWherein the actual value of the controlled quantityAlso in the sensor modelModel value for calculating actual value of controlled parameter

2. The method according to claim 1, characterized in that the model at the regulatory unit (3)The measured variable (m) measured in the control unit (3) is processed.

3. Method according to claim 1 or 2, characterized in that a model of the regulatory unit (3) is usedTo calculate an on-sensor model of the control unit (3)The state parameter (x) of the process.

4. Control of a controlled variable (RGn) of a control unit (3) for a reactant of a fuel cell (2), wherein a deviation between a setpoint value (RGNsoll) of the controlled variable and an actual value (RGnist) of the controlled variable is compensated for by means of the control, and for the purpose of the control the actual value (RGnist) of the controlled variable is determined according to one of claims 1 to 3.

5. System for controlling a controlled variable (RGn) of a reactant of a fuel cell (2) in a control unit (3) for the reactant, wherein a measurement sensor (Sn) is provided in order to measure a measured value of an actual value (RGnist) of the controlled variable; and a control unit (15) in which a controller (R) is implemented, which compensates for a deviation between a setpoint value (RGnsoll) of the controlled variable and an actual value (RGnist) of the controlled variable, characterized in that a model of the control unit (3) is providedThe model calculates a model value of the controlled parameterSensor model provided with a sensor (Sn) for measuringThe sensor model calculates the model value of the actual value of the controlled parameterA correction unit (omega) is provided, which uses a sensor model from the actual value (RGnist) of the controlled variable measured by the measuring sensor (Sn) and the actual value of the controlled variableCalculated model valueCalculating a correction value (RGcorr) for the controlled variable (RGn); and the controller (R) compares the correction value (RGcorr) with a model using a control unit (3)Calculated controlled parameter model valueIs used as the actual value of the controlled variableWherein the sensor modelAlso using modeled actual values of the controlled parametersFor calculating model values of actual values of the controlled variables

Technical Field

The invention relates to a method for determining an actual value of at least one controlled variable of a control unit for a fuel cell reactant, wherein a measured value of the actual value of the controlled variable is measured by a measuring sensor; and a system for controlling a controlled variable of a fuel cell reactant in a regulating unit for the reactant, wherein a measuring sensor is provided for measuring a measured value of an actual value of the controlled variable, and a control unit is provided in which a controller is implemented which compensates for a deviation between a theoretical value of the controlled variable and the actual value of the controlled variable; and a corresponding control of the controlled variable of the control unit of the fuel cell reactant.

Background

In order to operate the fuel cell as intended and with high efficiency, the regulation of the gases (reactants) supplied, in particular with regard to temperature, humidity, pressure and mass flow, plays a decisive role. Incorrect regulation of the reactants may result in power loss or, in the worst case, damage and destruction of the fuel cell or fuel cell stack. In particular, the relative humidity of the supplied reactants (for example oxygen, also in the form of supplied air) is an important variable in many fuel cell types, for example in Proton Exchange Membrane Fuel Cells (PEMFC), which must be precisely controlled.

Therefore, in order to be able to fully exploit the possibilities of fuel cells, it is necessary to have precise control of the reactant regulation means. This is associated with high costs, in particular in transient and highly dynamic operation of the fuel cell. In this context, a (highly) dynamic or transient operation is understood to mean, in particular, a rapid change in the output variables (voltage, current) of the fuel cell. This is problematic in particular when developing fuel cells on test benches, on which it is generally desirable to subject the fuel cells to dynamic test runs (in the sense of the rate of change of the output of the fuel cell, or also of the rate of change of the load) in order to check or improve the properties of the fuel cell. However, even in actual operation of a fuel cell (e.g., in an automobile), the regulation of reactants must enable transient and highly dynamic operation of the fuel cell. In particular, such rapid changes are understood to be dynamic, i.e., the dynamic system does not reach a steady state, but rather the transient behavior between the changes is reflected.

In order to control the regulation of the reactants precisely, the actual variable of the controlled variable needs to be detected in a measurement technique and supplied to the control device. Therefore, the measurement infrastructure (measurement sensors, measured value processing devices, measured value evaluation devices, etc.) must also be able to detect actual variables in transient, highly dynamic operation, in which the measured variables can change very rapidly and also very strongly in time. Accordingly, the requirements on the measurement infrastructure are equally high. This is also made difficult by the fact that the measuring sensors used for detecting the measured values are often themselves influenced by changing physical conditions (e.g. temperature, pressure, mass flow, humidity). The measuring sensor can therefore only be calibrated to a limited extent or with great effort. Furthermore, an unavoidable dead time also occurs in the measurement value detection, i.e. the measurement value is not available immediately, but only after a certain time. All this leads to distortion of the measurement results obtained, so that the measurement results obtained, in particular in transient, highly dynamic operation, do not correspond with sufficient accuracy to the physical measurement variables actually present. This reduces the controllability of the regulation of the fuel cell reactants. It was determined here that this problem is particularly relevant for the measurement of relative humidity. Inaccurate measurement of the actual variable may also lead to a power loss or to damage or even destruction of the fuel cell, in particular in transient, highly dynamic operation of the fuel cell.

Disclosure of Invention

It is therefore an object of the present invention to provide a method and a device for determining at least one actual value of a controlled variable of a fuel cell reactant with as little measurement error as possible, in order to be able to use the actual value in the control of the controlled variable.

The object is achieved in that a model value of the controlled variable is calculated using a model of the control unit, a model value of an actual value of the controlled variable is calculated using a sensor model for the measuring sensor, a correction value of the controlled variable is calculated from the actual value of the controlled variable measured using the measuring sensor and the model value of the actual value of the controlled variable calculated using the sensor model, and the actual value of the at least one controlled variable is calculated as a sum of the correction value and the model value of the controlled variable calculated using the model of the control unit, wherein the actual value of the controlled variable is also used in the sensor model for calculating the model value of the actual value of the controlled variable.

In this way, the dependency of the measurement sensor on environmental conditions, such as pressure, humidity, mass flow, temperature and thus also sensor errors in the transient behavior, can be reflected in the determination of the actual value of the controlled variable. This makes it possible to correct sensor errors and to improve the quality of the actual values used for the control. The greatest advantage is that the actual value of the actual measured variable is estimated and therefore both steady-state and dynamic sensor errors can be corrected. This is advantageous in particular in dynamic operation (the control unit (including the measurement sensor) is not in a steady state) since it was not possible until now. In the steady state case, the correction of the sensor error is reduced to zero point adjustment and calibration.

The modeled actual value of the controlled variable determined in this way can then be used in the control of the controlled variable in a control unit of the fuel cell reactant.

Drawings

The invention is explained in more detail below with reference to fig. 1 to 3, which show exemplary, schematic and non-limiting advantageous embodiments of the invention. In the figure:

figure 1 shows a conditioning unit for fuel cell reactants,

FIG. 2 shows the control of a controlled quantity of a reactant, and

fig. 3 shows the determination of a modeled actual value of a controlled quantity of a reactant for controlling the controlled quantity.

Detailed Description

The invention will be illustrated without limiting the generality by taking as an example a test stand 1 for a Proton Exchange Membrane Fuel Cell (PEMFC)2 with reference to fig. 1. Of course, the fuel cell 2 may also be used as a power supply in a machine or equipment or may be of another type. The control and regulation of this is then effected in the machine or installation. Therefore, when reference is made below to the operation of the fuel cell 2, this is always understood to mean the operation of the fuel cell 2 on the test stand 1 and the actual operation of the fuel cell 2 in the machine or installation. In general, only one fuel cell stack is also arranged on test stand 1, which is also understood to mean fuel cell 2 in the sense of the present invention.

In the example according to fig. 1, the PEMFC fuel cell 2 is arranged on a test stand 1 and is operated on the test stand 1. As is known, the fuel cell 2 comprises a cathode C to which a first reactant gas, for example oxygen (also in the form of air), is fed as a first reactant, and an anode a to which a second reactant gas, for example hydrogen H, is fed₂As a second reactant to the anode. The two reaction gases are separated from each other by a polymer membrane inside the fuel cell 2. A voltage U can be tapped between the cathode C and the anode a. The basic structure and function of the fuel cell 2 are sufficiently known and therefore need not be discussed in detail here.

At least one reactant, usually an oxygen-carrying reactant, in particular air, is regulated in the regulating unit 3. A plurality of (n is more than or equal to 1) are set in the regulation and control unit 3Regulated controlled parameters RGn (e.g. regulated pressure p of reactant gas, relative humidity)Temperature T and mass flow) In fig. 1, these four exemplary controlled parameters RGn are shown at the input of the cathode C. Of course, the reactant on the anode side can also be regulated in the same way. "regulating" means in this case that the value of the at least one controlled variable RGn is controlled to a predefined controlled variable setpoint value RGnsoll in such a way that: the controller R calculates a manipulated variable SGn set at the actuator An for at least one actuator An at each time step of the control with respect to the controlled variable RGn to be controlled by the actuator.

For controlling the controlled variable RGn, a corresponding actuator An is therefore provided in the control unit 3. For example, a humidifying device 4 for humidifying the reactant is provided as the actuator An to adjust the relative humidity of the reactantTemperature control means 5 for controlling the temperature of the reactants, for controlling the mass flow of the reactantsAnd a pressure control device 7 for controlling the pressure p of the reactant.

Of course, a source 8 for the at least one reactant is also provided, which is connected to the control unit 3 or is likewise provided in the control unit 3. The source 8 is, for example, an accumulator with compressed, dry reactant (e.g., air). Alternatively, the ambient air can also be treated, for example filtered, compressed, dried, etc., in the case of using air as the gas source 8.

The temperature control device 5 is, for example, an electric heating and cooling device or a heat exchanger. As temperature control device 5, a device as described in AT 516385 a1 can also be used.

In this embodiment, the humidifying device 4 comprises a water vapor generator 9, a mass flow controller 10 for water vapor and a mixing chamber 11. As mass flow controller 10 for the water vapor and also as mass flow control device 6 for the reactants, conventional, suitable, commercially available, controllable mass flow controllers can be used. In the mixing chamber 11, the water vapour is mixed with the gas from the source 8 into a conditioned reactant for the fuel cell 2.

Other embodiments of the humidifying device 4 are of course also conceivable. For example, water may be delivered (e.g., sprayed) to the gas in source 8.

In this example, a backpressure valve on the exhaust side, i.e. behind the fuel cell 2, which regulates the pressure p of the reactant via a controllable opening cross section, is used as the pressure control device 7. The back pressure valve 7 is provided in the gas conditioning unit 3 downstream of the fuel cell 2. This enables the pressure upstream of the fuel cell 2 to be controlled, and the pressure control means therefore remain unaffected by possible pressure losses in other components of the gas conditioning unit 3.

After the mixing chamber 11, the reactant is located in a reactant line 12, which is connected to the fuel cell 2, more precisely to the cathode C or the anode a of the fuel cell 2, so that the reactant has the desired controlled variable RGn, for example a specific temperature T, a specific relative humidityDetermined pressure p and/or determined mass flow

However, this configuration of the control unit 3 described with the aid of fig. 1 is merely exemplary, and other configurations of the control unit 3 and also other specific embodiments of the actuator An (here the humidifying device 4, the mass flow control device 6, the temperature control device 5 and the pressure control device 7) are of course possible and conceivable. In particular, fewer or more or other controlled variables RGn of the reactants can also be controlled in the control unit 3, whereby fewer or more or other actuators An can also be provided.

In order to be able to control the at least one controlled variable RGn, as shown in fig. 1, the associated actuators An, for example the humidifying device 4, the mass flow control device 6, the temperature control device 5 and the pressure control device 7, can be controlled by corresponding actuating variables SGn. The manipulated variable SGn is calculated by a control unit 15, in which a controller R is implemented, in such a way that the controlled variable actual value RGnist follows a predetermined setpoint value RGnsoll. In the exemplary embodiment shown in fig. 1, the manipulated variable u is used by a mass flow controller 10 for water vapor_STo control the humidifying device 4 by using the operation parameter u_GTo control the mass flow control device 6, using the manipulated variableTo control the temperature control device 5 and to use the operating variable u_NTo control the pressure control means 7. The manipulated variable SGn is used to actuate the respective actuator An and to set the actuator An in order to cause a desired change in the controlled variable RGn.

For controlling the controlled variable RGn, the controlled variable actual value RGnist is also required in order to compensate for a deviation between the controlled variable actual value RGnist and the controlled variable setpoint value RGnsoll by the controller R. As shown in fig. 2, for example, the difference between the setpoint controlled variable value RGnsoll and the actual controlled variable value RGnist is supplied to the controller R, which thus calculates An manipulated variable SGn according to the implemented control law, which is set in the control unit 3 by the associated actuator An. The actual value RGnist of the controlled variable is measured here by a measuring sensor Sn, which is of course arranged at a suitable point of the control unit 3, for example in the reactant gas line 12. In this case, the measurement sensor Sn does not necessarily have to measure the manipulated variable RGn directly, but may also measure a measured value representing the manipulated variable RGn. This measurement is often subject to the limitations mentioned at the outset. In order to improve the quality of the measurement of the actual value RGnist of the controlled variable, the measured value detected by the measurement sensor Sn is therefore not used directly for the control according to the invention, but rather a corrected measured value is used, as explained below with reference to fig. 3.

From suitable models of regulatory units 3(supply of manipulated variables SGn to the model), calculation of new controlled variables in response to the manipulated variables SGnHere, if further actuators An are contained in the control unit 3 and the model is describedIf the actuator is required, the model can also be givenOther operating variables of the control unit 3 are supplied. Likewise, models may also be givenThe desired measured value m of the regulating unit 3 is delivered. Suitable sensor modelModel values for the measured values of the controlled variable RGn are calculated from the controlled variable RGnFor this purpose, the sensor model can also be provided if requiredProviding a state variable x of the control unit 3, which can likewise be modeledAnd (4) calculating. The actual value RGnist of the controlled variable detected by the measuring sensor Sn and the sensor model are usedCalculated model valueIs supplied to a correction unit Ω, in which a correction value RGcorr of the controlled variable RG is calculated. The correct sign value RGcorr is combined with the model in the control unit 3Calculated controlled quantityIs used as the actual value of the modeled controlled variable for controlActual value of the modeled controlled parameterAlso fed to the sensor modelTo thereby calculate a model value of the controlled parameter RGn

Model of regulatory Unit 3Sensor modelAnd the correction unit Ω may be implemented, for example, as suitable software in the control unit 15, but may of course also be implemented individually or jointly as suitable hardware and/or software, respectively.

This way of detecting the actual value RGnist of the controlled variable for the control is particularly suitable for the relative humidity of the controlled reactantHowever, other controlled variables RGn, such as pressure p, temperature T, or mass flow rate, can also be used

The control unit 3 according to fig. 1 can be produced, for example, by means of a mathematical physical model described belowFor modeling, wherein of course also other models, also trained models, can be used.

Exemplary models for the regulatory unit 3Given as follows:

from the mass balance in the mixing chamber 11

Containing the mass m of the gas_GMass flow of gas into the mixing chamber 11Mass flow of gas out of the mixing chamber 11Mass flow of water vapor into the mixing chamber 11And the mass flow of water vapor out of the mixing chamber 11The mass flow of gas and water vapor out of the mixing chamber 11 is passed throughGiven, the total mass m and the mass m of the gas contained in the regulatory unit 3_GAnd mass m of water vapor_SAnd mass flow of reactantsIt is needless to say that m ═ m must be applied here_G+m_S。

Derived from the energy balance of the control unit 3

Here, U denotes the internal energy, and h denotes the specific enthalpy of the gas (here and in the following indicated by the subscript G), of the water vapor (here and in the following indicated by the subscript S), and of the reactants after the mixing chamber 11 (here and in the following without subscript), and U denotes the specific enthalpy of the gas (here and in the following indicated by the subscript S), and_iindicating the specific internal energy of gas and water vapor. It is known that the specific enthalpy h of a gas is the specific heat capacity c at a constant pressure_pThe product of the temperature T of the gas. In the case of water vapor, latent heat r is also added₀. Internal energy u of gas_iIs the specific heat capacity c at constant volume_vThe product of the temperature T of the gas. In the case of water vapor, latent heat r is also added₀. If all cases are put into the energy balance and the mass balance is taken into account, the following system equation is obtained which describes the temperature dynamics of the control unit 3.

Further derived from thermodynamic equations of state for ideal gases

pV＝(m_GR_G+m_SR_S)T

Including the pressure p and the temperature T at the input of the fuel cell 2. R denotes in a known manner the gas constant of a gas (index G), the gas constant of water vapor (index S) or the gas constant of a reactant (no index). The volume V preferably represents not only the volume of the mixing chamber 11 but also the volume of the line in the control unit 3. Pressure p and mass flow of reactantsAlso significantly affected by the back pressure valve 7, which can be modeled as follows.

Wherein A represents the opening cross section of the back pressure valve 7 and p₀Representing ambient pressure.

Relative humidityBy modeling the average of the average values, that is,

wherein p is_W(T) denotes the saturated partial pressure, for example byIt is given. Parameter p_m、C₁、C₂Parameters of "atmospheric convection", such as may be provided by Plant r.s.et alTransformation (mapping of Atmospheric Convection) "(volume I, Imperial university Press, 2015).

Additionally, it is also possible to use the manipulated variable u_S、u_GAnd u_NTo have a time constant τ₁,τ₂,τ₃,τ₄The form of the 1 st order delay element of (a) models the dynamics of the actuator An:

wherein, T_G,0And A₀Is a predetermined offset parameter.

From the above system equations, there are nonlinear multi-parametric systems (MIMO, multiple input multiple output) of the following form

RG＝h(x)

The system functions f (x), g (x), h (x), the following state vector x, manipulated variable vector SG with manipulated variable SGn, and manipulated variable vector RG with manipulated variable RGn, which are obtained from the modeling:

for better illustration, fig. 1 shows where these variables respectively occur in the control unit 3.

The correction unit Ω uses a sensor model from the actual value RGnist of the controlled variable measured by the measuring sensor Sn and the measured value for the controlled variable RGnCalculated model valueThe deviation between them to calculate the correction value RGcorr. In the simplest case, the difference of the two values can simply be used as the correction value RGcorr. It is also possible to implement a controller, for example a PI controller, in the correction unit Ω, which controls (i.e. controls towards zero) the difference. However, more complex (e.g. model-based) control methods may also be applied in the correction unit Ω.

Simple dynamic sensor modelFor example, a known, simple 1 st order delay element (PT1 element) which delays the reactant under a controlled variable RGn, for example, humidityThe change takes into account, for example, sensor inertia (known from data of the measurement sensor Sn or by measurements with the measurement sensor Sn). In addition, sensor models can be consideredE.g., to the system pressure p. This can again be achieved not only by means of a dynamic model, but also in simple embodiments as a static correction factor or in the form of a comprehensive characteristic curve.

By using dynamic sensor modelsModeling of the measurement sensor Sn may reflect the measurement sensor Sn in relation to environmental conditions, such as pressure, humidityMass flow, temperature and therefore also reflect sensor errors in transient behavior. For this purpose, it can be advantageous to model the sensorThe state variable x of the control unit 3 is taken into account. The sensor error is corrected by a correction unit Ω. In particular, therefore, it is also not necessary to calibrate the measurement sensor Sn in the control unit 3 for all environmental conditions, since the sensor error is in the calculated actual value of the controlled variable RGn for controlling the controlled variable RGnIs compensated for.

Furthermore, an unambiguous, time-dependent modeling of the dynamic sensor model is possible. Thus, the influence of changes in the sensor properties (e.g. dynamics) over time, i.e. for example aging, can be taken into account together.