Method for training soft sensor, fuel cell system, computer program product, computer readable medium and data carrier signal
By adjusting model parameters in the fuel cell system and calibrating the soft sensor using data from multiple operating points, the inaccuracy problem of the soft sensor in the fuel cell system was solved, and higher measurement accuracy was achieved.
Patent Information
- Application Number
- CN202480056150.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Priority Date
- 2023-07-21
- Filing Date
- 2024-07-19
- Publication Date
- 2026-04-17
AI Technical Summary
Soft sensors are inaccurate in detecting and measuring values, especially in fuel cell systems, where the tolerance of system units leads to inaccurate calculation of target values.
By adjusting the model parameters stored on the control device, the soft sensor is calibrated using data from multiple operating points, especially by calibrating through the deviation between the target value sensor and the soft sensor, and the model parameters are optimized to reduce or eliminate the deviation.
This improves the accuracy of the soft sensor, enabling it to accurately calculate target values at multiple operating points and reducing measurement errors in system units.
Smart Images

Figure CN121889898A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to a method for training soft sensors, a fuel cell system, a computer program product, a computer-readable medium, and a data carrier signal. Background Technology
[0002] Soft sensors (also known as virtual sensors) calculate target values based on measurements detected by the sensor and related models. Since the sensor used to detect the corresponding measurements may have tolerances and / or at least one system unit (e.g., the fluid delivery device in a fuel cell system) may have tolerances, the target values calculated by soft sensors may contain some degree of inaccuracy. Summary of the Invention
[0003] The objective of this invention is to at least partially eliminate the aforementioned disadvantages. In particular, the objective of this invention is to provide a method for training a soft sensor or a fuel cell system having such a soft sensor, wherein a model stored on a control device (for determining a target measurement value) is advantageously coordinated, in particular calibrated, at least for the corresponding sensor used to detect the measurement value in order to determine the target measurement value through the model and / or coordinated, in particular calibrated, for the system units.
[0004] The aforementioned task is accomplished by a method having the features of claim 1, a fuel cell system having the features of claim 5, and a computer program product, computer-readable medium, and data carrier signal having the features of claim 11, 12, or 13. Other features and details of the invention are derived from the dependent claims, the specification, and the drawings. Herein, the features and details described in association with the method according to the invention are also, of course, associated with the fuel cell system and / or computer program product and / or computer-readable medium and / or data carrier signal according to the invention, and vice versa, such that disclosures regarding various aspects of the invention are always mutually referential or can be mutually referenced.
[0005] According to a first aspect, the present invention discloses a method for training a soft sensor to determine a target measurement value of a fuel cell system, wherein the fuel cell system includes a first sensor for detecting a first measurement value, a second sensor for detecting a second measurement value, and a control device configured to determine the target measurement value based on a model, particularly a mathematical model, stored on the control device, at least based on the first and second measurement values. Furthermore, the fuel cell system includes a target value sensor for determining, particularly for measuring, the target measurement value. Additionally, the method includes, as a step, operating at least one system unit of the fuel cell system that affects the target value at at least one first operating point. Furthermore, the method includes, as a step, adjusting the values of at least one model parameter of the model stored on the control device, particularly model parameters of the characteristic curve, such that, at least at the first operating point of the system unit, the deviation between the target value determined by the soft sensor and the target value determined by the target value sensor is at least reduced, particularly eliminated, or substantially eliminated.
[0006] The methods and steps described above and below can be performed individually, jointly, once, multiple times, in parallel in time, and / or in any order, as long as they are technically meaningful.
[0007] In particular, a soft sensor comprises at least a first sensor for detecting a first measured value, a second sensor for detecting a second measured value, and a control device constructed for this purpose, the control device determining a target measured value based on a model stored on the control device, at least according to the first and second measured values. It is conceivable that a soft sensor may also include other sensors for detecting other measured values, such as a third sensor for detecting a third measured value.
[0008] This adjustment of at least one model parameter stored in the control device reduces the deviation between the target value determined by the soft sensor and the target value determined by the target value sensor, at least at the first operating point of the system unit. This is achieved, in particular, by eliminating the deviation or error between the target value determined by the soft sensor and the target value determined by the target value sensor through adjustment or change of the at least one model parameter, preferably by means of a regulator. When the deviation between the target value determined by the soft sensor and the target value determined by the target value sensor approaches zero, the regulator outputs a correct or adjusted model parameter corresponding to the first operating point of the system unit. The adjusted model parameter can be stored in the control device as a result, i.e., as a point on the adjusted characteristic curve. Adjusting the model parameter can also be understood as adjusting the (corresponding) value of the model parameter.
[0009] The concept of the “operating point” of the at least one system unit of the fuel cell system can also be understood as the “operating state”, such as the operating state of a valve (as a system unit).
[0010] Specifically, the target measurement value has a first target measurement value state at the first operating point of the fuel cell system unit. When the system unit has multiple operating points, the target measurement value has different target measurement value states at the corresponding operating points of the multiple operating points of the system unit.
[0011] In particular, the method may also include, as another step, switching the fuel cell system to a calibration mode in order to train the soft sensor.
[0012] By adjusting the at least one model parameter, the model stored on the control device (for the soft sensor used to determine the target measurement value) can be advantageously coordinated, in particular calibrated, at least for the corresponding sensor used to detect the measurement value in order to determine the target measurement value through the model and / or coordinated, in particular calibrated, for the system unit.
[0013] In the method according to the invention, it is advantageous to operate at least one system unit of the fuel cell system affecting the target value at multiple, particularly many, operating points, thereby adjusting the corresponding values of at least one model parameter, particularly the at least one model parameter, stored in the control device, such that the corresponding deviations between the target value determined by the soft sensor and the target value determined by the target value sensor at the corresponding operating points of the multiple, particularly many, operating points of the system unit are at least reduced, particularly eliminated or substantially eliminated. By operating the system unit at multiple, particularly many, operating points, different target measurement states of the target measurement can be generated. In other words, the fuel cell system having the at least one system unit can be operated in different states, with the target measurement having different target measurement states in the different states of the fuel cell system. Therefore, the corresponding adjusted model parameter can be stored in the control device as a corresponding point of the adjusted characteristic curve and the soft sensor can be trained. After adjusting the at least one model parameter, the new or adjusted model parameter can be continuously stored in the control device, such as the controller, or in non-volatile memory, so that the parameter can be used from the next operation.
[0014] In the method according to the invention, it is advantageous to train the soft sensor of the fuel cell system by operating the fuel cell system at least temporarily in a steady state. This training of the soft sensor specifically involves operating the fuel cell system at steady state at corresponding operating points of the plurality, particularly many, operating points of the at least one system unit, for at least a predetermined period of time. Therefore, it is particularly advantageous to train the soft sensor of the fuel cell system.
[0015] In the method according to the invention, it is advantageous that the training of the soft sensors is performed at time intervals. For the corresponding training of the soft sensors, target value sensors are set up, particularly mounted, on the fuel cell system to determine target measurement values, and after the corresponding training of the soft sensors, the target value sensors are separated from, particularly removed, from the fuel cell system. Therefore, the target value sensors can be installed on the fuel cell system only when needed. In particular, the training of the soft sensors of the fuel cell system can be performed, for example, during maintenance of a vehicle equipped with such a fuel cell system. For example, to calibrate the air mass flow rate soft sensor, the air mass flow rate sensor can also be set up, particularly mounted, on a sealable opening in the intake pipe of the fuel cell system's fluid path, or the fuel cell system's intake pipe can be replaced or replaced with a modified calibration intake pipe already equipped with an air mass flow rate sensor, such as a hot-film air mass flow meter, for example, during maintenance. After training the air mass flow rate sensor, the air mass flow rate sensor or the modified calibration intake pipe equipped with the air mass flow rate sensor can be removed again, and the fuel cell system can be restored to its original state.
[0016] According to a second aspect, the present invention discloses a fuel cell system comprising a first sensor for detecting a first measured value, a second sensor for detecting a second measured value, and a control device configured to determine a target measured value based on a model stored on the control device, at least according to the first and second measured values. Furthermore, the fuel cell system includes a target value sensor for determining the target measured value and an adjustment unit for adjusting at least one model parameter of the model stored on the control device such that at least one deviation between the target value determined by the soft sensor and the target value determined by the target value sensor at a first operating point of the system unit is at least reduced, particularly eliminated, or substantially eliminated. In particular, the fuel cell system is configured to perform the method according to the invention.
[0017] In particular, the adjustment unit can form part of the control device. In particular, the control device can be configured to adjust the at least one model parameter of the model stored on the control device in such a way that the deviation between the target value determined by the soft sensor and the target value determined by the target value sensor is reduced at least at the first operating point or multiple operating points of the system unit.
[0018] In particular, fuel cell systems can be used in vehicles, preferably motor vehicles such as passenger cars, trucks, or motorcycles. Preferably, the fuel cell system does not have a target value sensor for determining the target measurement value during normal operation. In other words, the fuel cell system or the vehicle equipped with such a fuel cell system does not have a target value sensor, such as an air mass flow sensor, during normal operation. Normal operation of a vehicle fuel cell system is, for example, the operation of the fuel cell system when the vehicle is traveling on a road.
[0019] Advantageously, the fuel cell system according to the invention includes: a fluid delivery device as a system unit for delivering cathode gas; a pressure sensor as a first sensor for detecting the pressure of the cathode gas in the fuel cell stack of the fuel cell system as a first measured value; a temperature sensor as a second sensor for detecting the temperature of the cathode gas as a second measured value; a speed sensor as a third sensor of the fuel cell system for detecting the speed of the fluid delivery device; an air mass flow sensor as a target value sensor for determining the air mass flow rate as a target measured value; and an air mass flow model as a model stored on a control device. In other words, the fuel cell system may include an air mass flow soft sensor, the at least one model parameter of which is adjusted, preferably calibrated, by an air mass flow sensor (e.g., by a hot-film air mass flow meter). The fluid delivery device, in particular a compressor, as a system unit of the fuel cell system, is used to compress the cathode gas, such as air, in the cathode gas path of the fuel cell system to supply cathode gas to the fuel cell stack. For example, to train the air mass flow soft sensor, the speed can be traversed in a stepped distribution across the speed range of the fluid delivery device, preferably the entire speed range. In each speed range and therefore at different pressure ratios in the fluid delivery device, the corresponding deviation or error between the air mass flow sensor and the target value sensor can be reduced, especially eliminated, by adjusting or changing the at least one model parameter, especially the corresponding value of the model parameter, (e.g., by means of a regulator). The sensor value of the hot-film air mass flow meter, which serves as the air mass flow sensor, can also be smoothed by a low-pass filter to filter out superimposed noise. This noise may be caused by geometrically based deviations between the hot-film air mass flow meter and the optimal installation location. Especially after a complete traversal of the stepped distribution, the characteristic curve is filled with the corresponding (if necessary) adjusted model parameters. Therefore, the effective range of the air mass flow soft sensor is particularly advantageous across the entire speed range of the fluid delivery device (even at low speeds). Furthermore, intermediate values can be determined or calculated by interpolation, especially linear interpolation, allowing for particularly rapid training of the soft sensor.
[0020] In the fuel cell system according to the invention, it is advantageous that the target value sensor is detachably, and in particular non-destructively, disposed on the fuel cell system for determining the target measurement value. Therefore, the target value sensor can be installed or removed (non-destructively) as needed.
[0021] In the fuel cell system according to the invention, it is advantageous that, in order to operate the target value sensor, the target value sensor is technically connected to an energy source at least independently of the fuel cell system. For example, the target value sensor can have its own energy source to operate. Thus, for example, an air mass flow sensor can be installed particularly simply or at very low cost in a sealable opening in the inlet pipe of the cathode gas path of the fuel cell system for setting up the air mass flow sensor, in order to train the soft sensor.
[0022] In the fuel cell system according to the invention, it is advantageous that the target value sensor is configured to transmit data about the target measurement value determined by the target value sensor via radio technology, for example, to the adjustment unit or control device of the fuel cell system. Therefore, the training of the soft sensor can be particularly simple and at a particularly low cost.
[0023] Therefore, the fuel cell system according to the second aspect of the invention has the same advantages as those already described for the method according to the first aspect of the invention.
[0024] According to other aspects, the present invention discloses a computer program product, a computer-readable medium, and a data carrier signal, wherein the computer program product, computer-readable medium, and data carrier signal have the same advantages as those already described for methods according to a first aspect of the invention, fuel cell systems according to a second aspect of the invention, and computer program products, computer-readable media, or data carrier signals according to other aspects of the invention.
[0025] The computer program product includes commands, especially instructions, that cause the fuel cell system according to the invention to perform the method according to the invention. The computer program product can be implemented as computer-readable instruction code in any suitable programming language such as JAVA, C++, etc., and in particular, the instruction code of the computer program product can be used to program a computer or other programmable device, such as the control device of the fuel cell system according to the invention, to perform the desired function.
[0026] The computer program product can be implemented through instruction code, i.e., software (computer program), or through one or more dedicated electronic circuits, i.e., hardware, or in any hybrid form, i.e., through software components and hardware components.
[0027] Computer program products may be stored on computer-readable storage media, such as volatile or non-volatile memory and / or the processor of a control device.
[0028] Furthermore, computer program products can be provided on networks such as the Internet, and users can download them from said networks as needed. In particular, when downloading computer program products, data carrier signals are transmitted according to the computer program products of the present invention.
[0029] Further improvements to the invention are derived from the following description of some embodiments of the invention, which are schematically illustrated in the accompanying drawings. All features and / or advantages derived from the claims, specification, or drawings, including structural details, spatial arrangements, and method steps, can be essential to the invention not only individually but also in various combinations. It should be noted that the drawings are descriptive only and are not intended to limit the invention in any way. Attached Figure Description
[0030] The attached image is as follows:
[0031] Figure 1 This demonstrates one method; and
[0032] Figure 2 A fuel cell system is shown. Detailed Implementation
[0033] In the following figures, the same reference numerals are used for the same technical features in different embodiments.
[0034] Figure 1 A method for training a soft sensor to determine target measurements of a fuel cell system (100) is schematically disclosed, the system being, for example, referenced to... Figure 2 The method is described in detail. To train the soft sensors of a fuel cell system (100), specifically to enable the fuel cell system (100) to operate at least temporarily in a steady state, the fuel cell system (100) includes a first sensor (10) for detecting a first measurement value, a second sensor (20) for detecting a second measurement value, and a control device (50) configured to determine a target measurement value based on a model stored on the control device (50), at least according to the first and second measurement values. Furthermore, the fuel cell system (100) includes a target value sensor (40) for (directly) determining or (directly) measuring the target measurement value. Additionally, the method, as a first step, includes: operating at least one system unit (110) of the fuel cell system (100) that influences the target value at at least one first operating point (320). Furthermore, the method includes, as a step, adjusting (340) at least one model parameter or the value of a model parameter of the characteristic curve of the model stored on the control device (50) such that, at least at the first operating point of the system unit (110), the deviation between the target value determined by the soft sensor and the target value determined by the target value sensor (40) is at least reduced, preferably eliminated.
[0035] exist Figure 1 In the method shown, it is advantageous to operate (321) at least one system unit (110) of the fuel cell system (100) that affects the target value at multiple operating points, thereby adjusting (341) the corresponding values of at least one model parameter of the model stored on the control device (50), in particular the corresponding values of the at least one model parameter, so that the corresponding deviations between the target value determined by the soft sensor and the target value determined by the target value sensor (40) at the corresponding operating points of the multiple operating points of the system unit (110) are at least reduced.
[0036] exist Figure 1 The method shown may also have the advantage that the training of the soft sensors is performed at time intervals, and in order to train the soft sensors accordingly, a target value sensor (40) is set (300) on the fuel cell system (100) to determine the target measurement value, and in time after the training of the soft sensors, the target value sensor (40) is separated from the fuel cell system (100), in particular removed (360).
[0037] Figure 2 A fuel cell system (100) having a fuel cell stack (101) is schematically disclosed, wherein the fuel cell system (100) is configured to perform or implement, for example, refer to Figure 1 The method described above. Furthermore... Figure 2The cathode gas path for supplying cathode gas to the fuel cell stack (101) is schematically shown in at least a portion. The fuel cell system (100) includes a fluid delivery device disposed as a system unit (110) in the cathode gas path for delivering cathode gas. Furthermore, the fuel cell system (100) includes a pressure sensor (10) as a first sensor for detecting the pressure of the cathode gas in the fuel cell stack (101) of the fuel cell system (100) as a first measurement. Additionally, the fuel cell system (100) includes a temperature sensor (20) as a second sensor for detecting the temperature of the cathode gas as a second measurement. Furthermore, the fuel cell system (100) includes a speed sensor (30) as a third sensor for detecting the speed of the fluid delivery device as a third measurement. Finally, the fuel cell system (100) includes an air mass flow sensor (40) as a target value sensor for determining the air mass flow rate as a target measurement. Furthermore, the fuel cell system (100) includes a control device (50) configured to determine, at least based on a first measurement, a second measurement, and a third measurement, an air mass flow rate model stored on the control device (50), as a target measurement value, and the control device (50) is configured to adjust at least one model parameter of the air mass flow rate model stored on the control device (50) such that at least one deviation between the target value determined by the soft sensor and the target value determined by the target value sensor (40) at a first operating point of the system unit (110) is at least reduced.
[0038] exist Figure 2 In the fuel cell system (100) shown, it is advantageous that, in order to determine the target measurement value, the target value sensor (40) is detachably, and in particular non-destructively, disposed on the fuel cell system (100). Figure 2 In the fuel cell system (100) shown, it is also advantageous that, in order to operate the target value sensor (40), the target value sensor (40) is technically connected to an energy source at least independently of the fuel cell system (100). Figure 2 In the fuel cell system (100) shown, it is also advantageous that the target value sensor (40) is configured to transmit data about the target measurement value determined by the target value sensor (40) via radio technology, for example, to the control device (50).
[0039] List of reference numerals
[0040] 10 First Sensor
[0041] 20 Second Sensor
[0042] 30 Third Sensor
[0043] 40 Target Value Sensor
[0044] 50 Control devices
[0045] 60 Adjustment Units
[0046] 100 Fuel Cell System
[0047] 101 Fuel Cell Stack
[0048] 110 System Unit
[0049] 300. Install the target value sensor on the fuel cell system.
[0050] 320, 321 Cause at least one system unit affecting the target value to operate at the first operating point or multiple operating points.
[0051] 340, 341 Adjust at least one model parameter of the model stored on the control device.
[0052] 360° Separation of the target value sensor from the fuel cell system
Claims
1. A method for training a soft sensor to determine target measurements of a fuel cell system (100), wherein, The fuel cell system (100) includes: - A first sensor (10) for detecting the first measured value. - A second sensor (20) for detecting a second measurement value. - A control device (50), the control device (50) being configured to determine a target measurement value at least based on a first measurement value and a second measurement value using a model stored on the control device (50). - Target value sensor (40) used to determine the target measurement value. And the method includes: - To enable at least one system unit (110) of the fuel cell system (100) that affects the target value to operate (320) at at least one first operating point. - Adjust (340) at least one model parameter of the model stored on the control device (50) such that the deviation between the target value determined by the soft sensor and the target value determined by the target value sensor (40) is reduced at least at the first operating point of the system unit (110).
2. The method according to claim 1, characterized in that, At least one system unit (110) of the fuel cell system (100) affecting the target value is operated (321) at multiple operating points, and at least one model parameter of the model stored on the control device (50) is adjusted (341) such that the corresponding deviation between the target value determined by the soft sensor and the target value determined by the target value sensor (40) at the corresponding operating points of the multiple operating points of the system unit (110) is at least reduced.
3. The method according to any one of the preceding claims, characterized in that, In order to train the soft sensors of the fuel cell system (100) so that the fuel cell system (100) can operate at least temporarily in a steady state.
4. The method according to any one of the preceding claims, characterized in that, The training of the soft sensors is performed at time intervals. In order to train the soft sensors accordingly, a target value sensor (40) is set (300) on the fuel cell system (100) to determine the target measurement value. After the training of the soft sensors, the target value sensor (40) is separated from the fuel cell system (100), and in particular, removed (360).
5. Fuel cell system (100), wherein, The fuel cell system (100) includes: - A first sensor (10) for detecting the first measured value. - A second sensor (20) for detecting a second measurement value. - Control device (50), the control device is configured to determine the target measurement value at least based on the first measurement value and the second measurement value using a model stored on the control device (50). - Target value sensor (40) used to determine the target measurement value. - Adjustment unit (60), the adjustment unit is used to adjust at least one model parameter of the model stored on the control device (50) such that at least one deviation between the target value determined by the soft sensor and the target value determined by the target value sensor (40) is reduced at the first operating point of the system unit (110).
6. The fuel cell system (100) according to claim 5, characterized in that, The fuel cell system (100) includes: -A fluid transport device, serving as a system unit (110), for transporting cathode gas, - As the first sensor (10), the pressure sensor is used to detect the pressure of the cathode gas in the fuel cell stack (101) of the fuel cell system (100) as the first measured value. - As a second sensor (20), the temperature sensor is used to detect the temperature of the cathode gas as a second measured value. - As the third sensor (30) of the fuel cell system (100), the rotational speed sensor is used to detect the rotational speed of the fluid delivery device as a third measurement value. - An air mass flow sensor (40) is used as a target value sensor to determine the air mass flow rate as a target measurement value. - An air mass flow model stored on the control device (50).
7. The fuel cell system (100) according to claim 5 or 6, characterized in that, In order to determine the target measurement value, the target value sensor (40) can be disposed on the fuel cell system (100) in a separable, and in particular, non-destructive, manner.
8. The fuel cell system (100) according to any one of claims 5 to 7, characterized in that, In order to operate the target value sensor (40), the target value sensor (40) is connected in energy technology to an energy source that is at least independent of the fuel cell system (100).
9. The fuel cell system (100) according to any one of claims 5 to 8, characterized in that, The target value sensor (40) is configured to transmit data about the target measurement value determined by the target value sensor (40) via radio technology.
10. The fuel cell system (100) according to any one of claims 5 to 9, characterized in that, The fuel cell system (100) is configured to perform the method according to any one of claims 1 to 4.
11. Computer program products, of which, The computer program product includes commands that cause the fuel cell system (100) according to any one of claims 5 to 10 to perform the method according to any one of claims 1 to 4.
12. A computer-readable medium, wherein, The computer program product according to claim 11 is stored on the computer-readable medium.
13. Data carrier signal, wherein, The data carrier signal transmission is described in the computer program product according to claim 11.
