A method for predicting attenuation of vehicle fuel cell stack
By updating and cleaning data in the fuel cell stack in real time, establishing a multi-dimensional model and using neural networks for feature screening, the problems of complex model parameters and slow update in the existing technology are solved, and high-precision and fast fuel cell stacks are achieved Life expectation, suitable for various application scenarios.
Patent Information
- Application Number
- CN202411150241.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-21
- Publication Date
- 2024-11-19
- Estimated Expiration
- 2044-08-21
AI Technical Summary
The existing fuel cell stack residual service life prediction method has complex model parameters and is slow to update, so it cannot quickly adapt to changes in real operating conditions, resulting in low prediction accuracy and large errors.
A method of attenuation prediction of vehicle fuel cell stack is adopted to obtain and clean vehicle operation data, establish a multi-dimensional model, use neural network models to screen and fit data features, update model parameters in real time, and predict with a robust local weighted smoothing algorithm. , calculate the percentage of attenuation degree to achieve fast modeling and high-precision prediction.
It improves the accuracy and speed of fuel cell residual service life prediction, is suitable for various application scenarios, can update model data in real time, adapt to changes in working conditions, reduce calculation amount, and improves the reference and accuracy of predictions.
Smart Images

Figure CN118671605B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of fuel cell stack life estimation, and in particular to a method for predicting attenuation of a vehicle-mounted fuel cell stack. Background Art
[0002] Proton exchange membrane fuel cells (PEMFC) are widely used in fuel cell electric vehicles and other scenarios due to their advantages such as no pollution, high specific energy, low noise, and high energy conversion rate. They have become the core technology in the field of clean energy and are considered to be the most promising engine. However, large-scale commercial fuel cell vehicle applications are severely limited by cost and service life. The expected life of PEMFC in actual use varies significantly from the theoretical life depending on the operating conditions and control strategies. Remaining service life prediction and health management are feasible methods to solve the cost and durability problems of proton exchange membrane fuel cells PEMFC.
[0003] However, the model parameters used in the prediction methods of the remaining service life of fuel cells that are currently mature in the industry are complex and have a slow update and improvement of the prediction model. The model data cannot be continuously updated as the actual operating conditions of the fuel cell change, resulting in many errors in the currently predicted remaining service life and low accuracy. For this reason, the present application proposes a method for predicting the attenuation of a vehicle-mounted fuel cell stack. Summary of the invention
[0004] In view of the defects existing in the prior art, the purpose of the present invention is to provide a method for predicting the attenuation of a vehicle-mounted fuel cell stack, aiming to solve the technical problems in the related art to a certain extent.
[0005] In order to achieve the above purpose, the technical solution adopted by the present invention is:
[0006] A method for predicting attenuation of a vehicle fuel cell stack comprises the following steps:
[0007] S1. The program starts, obtains the operating data of the fuel cell systems of the same model of all vehicles to be evaluated, and proceeds to the next step;
[0008] S2, data extraction operation 1: extract the vehicle startup data L1 according to the vehicle running state, and delete the missing and abnormal data in the data L1 in advance, and proceed to the next step;
[0009] S3, data extraction operation 2: extract valid data through the fuel cell stack start and stop flag, and proceed to the next step;
[0010] S4, whether the start / stop flag is the set value, if not, delete the data corresponding to the non-set value, if yes, perform data cleaning, delete blank data, and select the current with the longest running time as the reference current Irefer , proceed to the next step;
[0011] S5. Get reference current I refer The voltage V refer And the influencing parameters, including power P, temperature T, humidity H, go to the next step;
[0012] S6, respectively perform monotonic prediction of voltage V pred1 The influencing parameters are fitted with the one-variable multi-function Y1 and the one-variable linear function Y2, and the influencing parameters are respectively substituted into the fitted function model Y1 and model Y2 to obtain the monotonic prediction voltage V pred1 =(Y1+Y2) / 2, go to the next step;
[0013] S7. Calculate the comprehensive predicted voltage value V pred2 , V pred2 =(V pred1 +V pred2 +V pred3 ,…,V predn ) / n, n is a positive integer, go to the next step;
[0014] S8. Determine the extracted sample voltage V sample The voltage is less than the comprehensive predicted voltage value V for three consecutive times pred2 If not, SOH=100%, no attenuation; if yes, go to the next step;
[0015] S9, extracted sample voltage V sample It is less than the preset polarization curve voltage three times in a row. If not, SOH=100%, no attenuation. If yes, substitute the comprehensive predicted voltage value V pred2 , obtain the predicted life distribution data and proceed to the next step;
[0016] S10, calculate the attenuation percentage S = [1-(V sample / V pred2 )]x100%.
[0017] On the basis of the above technical solution, the following data extraction steps are also included:
[0018] S101, reading the vehicle operation data D1 of multiple fuel cell vehicles of the same model;
[0019] S102, delete the row containing blank data to obtain data D2;
[0020] S103, extracting data D3 in the vehicle start state according to the two running states of vehicle start and vehicle stop in data D2;
[0021] S104, extracting data D4 in which the fuel cell start / stop position is started according to data D3;
[0022] S105 . Delete data with a running time of less than 200 seconds according to data D4 to obtain data D5 .
[0023] On the basis of the above technical solution, the following influencing parameter extraction steps are also included:
[0024] S201, extracting data distribution according to data D5;
[0025] S202, using a neural network model to perform data feature screening, extracting current distribution data in data D5 and using the most distributed current value as the reference current I with the longest running time refer ;
[0026] S203, further extracting data D5, the current value (I refer All data within the ampere range of ±5) are extracted and used as data D6.
[0027] Based on the above technical solution, data D6 includes voltage, current, power, temperature, humidity and time.
[0028] On the basis of the above technical solution, the following model building steps are also included:
[0029] S301, establishing a multidimensional model according to data D6;
[0030] S302, fitting the model, dividing the extracted data into a training set and a test set;
[0031] S303, extracting voltage, power, temperature, and humidity to build a voltage-power model, a voltage-temperature model, and a voltage-humidity model, respectively, and calculating the voltage-power model voltage V1, the voltage-temperature model voltage V2, and the voltage-humidity model voltage V3;
[0032] S304, calculate the monotonic prediction voltage Vpred1, set the univariate multi-function Y1=a 0 +a 1 x+a 2 x 2 +a 3 x 3 , where a 0 、a 1 、a 2 、a 3 is a fitting parameter and takes values in the real number range, x is the power or temperature or humidity value; and a one-dimensional linear model Y2=aX+b is set, where a and b are fitting parameters and take values in the real number range, x is the power or temperature or humidity; corresponding monotonic prediction voltage Vpred1=(Y1+Y2) / 2;
[0033] S305, calculate the comprehensive predicted voltage value Vpred2, and average the voltage-power model voltage V1, the voltage-temperature model voltage V2, and the voltage-humidity model voltage V3 to obtain the comprehensive predicted voltage value Vpred2=(V1+V2+V3) / 3.
[0034] On the basis of the above technical solution, the following prediction result output steps are also included:
[0035] S401. According to the fuel cell stack life test curve, establish a voltage-time model Y, Y=AX+B, where Y is voltage, X is time, and A and B are real constants;
[0036] S402, extracting sample voltage V sample , to predict the fuel cell stack;
[0037] S403: extract the sample voltage V sample , the comprehensive predicted voltage value Vpred2, and the preset polarization curve voltage are compared. When the extracted sample voltage V sample is less than the comprehensive predicted voltage value Vpred2 for three consecutive times, and the extracted sample voltage V sample If the voltage is less than the preset polarization curve voltage three times in a row, it means that the fuel cell stack has decayed and the decay degree can be predicted;
[0038] S404, calculating the attenuation percentage S=[1-(Vsample / Vpred2)]x100%, and calculating the time X=(YB)A according to the voltage-time model Y, where X is the time value of the consumed design life.
[0039] On the basis of the above technical solution, the influencing parameters also include operating time T, cumulative air flow Q, cumulative mileage k, current / voltage ratio, and power generation E.
[0040] Based on the above technical solution, a vehicle-mounted fuel cell stack attenuation prediction device includes a data preprocessing module, a model building module, a prediction voltage module, and an attenuation calculation module which are connected in sequence, and the data preprocessing module, the model building module, the prediction voltage module, and the attenuation calculation module are integrated in the fuel cell controller and are electrically connected to the processor of the fuel cell controller respectively.
[0041] On the basis of the above technical solution, the fuel cell vehicle is provided with an on-board fuel cell stack attenuation prediction device, and uses an on-board fuel cell stack attenuation prediction method to monitor the service life of the fuel cell stack.
[0042] Compared with the prior art, the advantages of the present invention are:
[0043] (1) Compared with the prior art, the attenuation prediction method for a vehicle-mounted fuel cell stack in the present invention is widely applicable to various application scenarios of fuel cells. The model data is continuously updated as the actual operating conditions change, so that the predicted remaining life has good accuracy and reference value.
[0044] (2) The attenuation prediction method of a vehicle-mounted fuel cell stack in the present invention has the characteristics of real-time updating, rapid modeling and strong adaptability, and uses fewer influencing parameters. After multiple function model fittings, suitable parameters can be selected as model parameters to obtain a reliable performance prediction model. The prediction model is used to predict the remaining service life of the fuel cell, which effectively improves the prediction speed and prediction accuracy of the remaining service life of the fuel cell. At the same time, before the collected output voltage data of the fuel cell to be tested is input into the prediction model, the output voltage data needs to be reconstructed using a robust local weighted smoothing algorithm. This process effectively reduces the amount of calculation and can well filter out interference, thereby further improving the prediction speed and prediction accuracy of the remaining service life of the fuel cell. BRIEF DESCRIPTION OF THE DRAWINGS
[0045] Figure 1 The present invention is a logic block diagram of a method for predicting attenuation of a vehicle-mounted fuel cell stack in an embodiment of the present invention. DETAILED DESCRIPTION
[0046] The embodiments of the present invention are further described in detail below with reference to the accompanying drawings.
[0047] Exemplary embodiments will be described in detail herein, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, the same numbers in different drawings represent the same or similar elements unless otherwise indicated. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present disclosure. Instead, they are merely examples of devices and methods consistent with some aspects of the present disclosure as detailed in the appended claims.
[0048] In the description of the present application, it should also be noted that, unless otherwise clearly specified and limited, the terms "set", "install", "connect", and "connect" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be a connection between the two elements. For ordinary technicians in this field, the specific meanings of the above terms in this application can be understood according to the specific circumstances. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article, or device including a series of elements includes not only those elements, but also other elements that are not explicitly listed, or also includes elements inherent to such a process, method, article, or device. In the absence of further restrictions, the elements defined by the sentence "including one..." do not exclude the existence of other identical elements in the process, method, article, or device including the elements.
[0049] See also Figure 1 The logical block diagram of a method for predicting attenuation of a vehicle-mounted fuel cell stack in an embodiment of the present invention is shown. On the basis of the above technical solution, it also includes the following data extraction steps: first, all vehicle operation data requiring multiple fuel cells of the same model are centrally extracted. The more data, the more accurate the prediction model established. In this embodiment, the vehicle operation data D1 of multiple fuel cells of the same model is read as the basis for establishing the prediction data model; the large and complex data are preliminarily cleaned, and the rows containing blank data are deleted in advance to obtain data D2; then, according to the two operating states of vehicle start and stop in the data D2, the data D3 in the vehicle start state is extracted; then, according to the data D3, the data D4 with the fuel cell start and stop position as start is extracted; then, according to the data D3, the data with the running time less than 200 seconds is deleted to obtain data D5. The data less than 200 seconds is regarded as invalid data, and the data that may be generated due to startup failure or failure to work normally is further stripped.
[0050] In this embodiment, the influencing parameter extraction step is as follows: according to the operation in the data extraction step, the data D5 is subjected to data distribution extraction, the data feature screening is performed using a neural network model, the current distribution data in the data D5 is extracted, and the current value with the most distribution is used as the reference current I with the longest running time. refer And further extract data D5, the current value is (I refer ±5) All data within the ampere range are extracted and used as data D6. Data D6 includes voltage, current, power, temperature, humidity, and time.
[0051] In this embodiment, the model building steps are as follows: according to the influencing parameter extraction step operation, further establish a multidimensional model based on the data D6; first divide the extracted data into a training set and a test set; fit the data in the training set to the model, that is, process the training set data into a function model, use the function model to calculate the predicted value, and the test set is used to verify the function model and extract the predicted data. Then extract the voltage, power, temperature, and humidity to build a voltage-power model, a voltage-temperature model, and a voltage-humidity model respectively, and calculate the voltage-power model voltage V1, the voltage-temperature model voltage V2, and the voltage-humidity model voltage V3; then immediately perform the monotonic prediction voltage Vpred1 calculation, and in this embodiment, set the univariate multi-function Y1=a 0 +a 1 x+a 2 x 2 +a 3 x 3 , where a 0 、a 1 、a 2 、a 3 is a fitting parameter and takes values within the real number range, x is the power or temperature or humidity value; and in this embodiment, a linear model Y2=aX+b is set, wherein a and b are fitting parameters and take values within the real number range, x is the power or temperature or humidity; the corresponding monotonic predicted voltage Vpred1=(Y1+Y2) / 2.
[0052] Secondly, the comprehensive predicted voltage value Vpred2 is calculated, and the voltage-power model voltage V1, the voltage-temperature model voltage V2, and the voltage-humidity model voltage V3 are averaged to obtain the comprehensive predicted voltage value Vpred2=(V1+V2+V3) / 3.
[0053] In this embodiment, after the model building step is completed, the prediction result output step is further completed: according to the fuel cell stack life test curve, a voltage-time model Y is established in this embodiment, where Y=AX+B, where Y is voltage, X is time, and A and B are real constants; the sample voltage V is extracted. sample , to predict the fuel cell stack; the fuel cell stack life test curve can be obtained in advance or set according to the factory parameters. The voltage-time model Y=AX+B is the linear relationship expression between the use time X and the voltage Y, reflecting the change of the use time and voltage of the fuel cell stack. sample , the comprehensive predicted voltage value Vpred2, and the preset polarization curve voltage are compared. When the extracted sample voltage V sample The voltage is less than the comprehensive predicted voltage value Vpred2 for three consecutive times, indicating that the voltage attenuation is clear and the extracted sample voltage V sampleIf the voltage is less than the preset polarization curve voltage three times in a row, it means that the attenuation is clear compared with the factory data, indicating that the fuel cell stack has decayed and the degree of decay can be predicted; finally, the attenuation percentage S=[1-(Vsample / Vpred2)]x100% is calculated, and according to the voltage-time model Y, the time X=(YB)A is calculated, where X is the time value of the consumed design life.
[0054] In this embodiment, the influencing parameters also include the vehicle fuel cell system operating time T and the cumulative air flow Q, the total cumulative mileage k of the fuel cell vehicle, the current / voltage ratio change curve, and the total power generation E of the fuel cell system.
[0055] In this embodiment, a vehicle-mounted fuel cell stack attenuation prediction device includes a data preprocessing module, a model building module, a prediction voltage module, and an attenuation calculation module which are connected in sequence, and the data preprocessing module, the model building module, the prediction voltage module, and the attenuation calculation module are integrated in the fuel cell controller and are electrically connected to the processor of the fuel cell controller respectively.
[0056] In this embodiment, the fuel cell vehicle is provided with an on-board fuel cell stack attenuation prediction device, and uses an on-board fuel cell stack attenuation prediction method to monitor the service life of the fuel cell stack, and prompts the driver or maintenance personnel to pay attention to the fuel cell health information by real-time display on the on-board display screen.
[0057] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.
[0058] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.
[0059] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.
[0060] The present invention is not limited to the above-mentioned embodiments. For ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications are also considered to be within the scope of protection of the present invention. The contents not described in detail in this specification belong to the prior art known to professional and technical personnel in this field.
Claims
1. A method for predicting the attenuation of a vehicle fuel cell stack. Features The following steps are involved: S1. The program starts, obtains the operating data of the fuel cell systems of the same model of all vehicles to be evaluated, and proceeds to the next step; S2, data extraction operation 1: extract the vehicle startup data L1 according to the vehicle running state, and delete the missing and abnormal data in the data L1 in advance, and proceed to the next step; S3, data extraction operation 2: extract valid data through the fuel cell stack start and stop flag, and proceed to the next step; S4, whether the start / stop flag is the set value, if not, delete the data corresponding to the non-set value, if yes, perform data cleaning, delete blank data, and select the current with the longest running time as the reference current I refer , proceed to the next step; S5. Get reference current I refer The voltage V refer And the influencing parameters, including power P, temperature T, humidity H, go to the next step; S6, respectively perform monotonic prediction of voltage V pred1 The influencing parameters are fitted with the one-variable multi-function Y1 and the one-variable linear function Y2, and the influencing parameters are respectively substituted into the fitted function model Y1 and model Y2 to obtain the monotonic prediction voltage V pred1 =(Y1+Y2) / 2, go to the next step; S7. Calculate the comprehensive predicted voltage value V pred2 , V pred2 =(V pred1 +V pred2 +V pred3 ,…,V predn ) / n, n is a positive integer, V pred1 、V pred2 、V pred3 ,…,V predn For each calculated monotonic predicted voltage value, proceed to the next step; S8. Determine the extracted sample voltage V sample Is it less than the comprehensive predicted voltage value V for three consecutive times? pred2 If not, SOH=100%, no attenuation; if yes, go to the next step; S9, determine the extracted sample voltage V sample Is it less than the preset polarization curve voltage for three consecutive times? If not, SOH=100%, no attenuation. If yes, substitute the comprehensive predicted voltage value V pred2 , obtain the predicted life distribution data and proceed to the next step; S10, calculate the attenuation percentage S = [1-(V sample / V pred2 )]x100%.
2. A method for predicting attenuation of a vehicle fuel cell stack according to claim 1, It is characterized in that The following data extraction steps are also included: S101, reading the vehicle operation data D1 of multiple fuel cell vehicles of the same model; S102, delete the row containing blank data to obtain data D2; S103, extracting data D3 in the vehicle start state according to the two running states of vehicle start and vehicle stop in data D2; S104, extracting data D4 in which the fuel cell start / stop position is started according to data D3; S105 . Delete data with a running time of less than 200 seconds according to data D4 to obtain data D5 .
3. A method for predicting attenuation of a vehicle fuel cell stack according to claim 2, It is characterized in that The following influencing parameter extraction steps are also included: S201, extracting data distribution according to data D5; S202, using a neural network model to perform data feature screening, extracting current distribution data in data D5 and using the most distributed current value as the reference current I with the longest running time refer ; S203, further extract data D5, and convert the current value I refer All data within the range of ±5 amperes are extracted and used as data D6.
4. A method for predicting attenuation of a vehicle fuel cell stack according to claim 3, Features: The data D6 includes voltage, current, power, temperature, humidity, and time.
5. A method for predicting attenuation of a vehicle-mounted fuel cell stack according to claim 2, It is characterized in that The following model building steps are also included: S301, establishing a multidimensional model according to data D6; S302, fitting the model, dividing the extracted data into a training set and a test set; S303, extracting voltage, power, temperature, and humidity to build a voltage-power model, a voltage-temperature model, and a voltage-humidity model, respectively, and calculating the voltage-power model voltage V1, the voltage-temperature model voltage V2, and the voltage-humidity model voltage V3; S304, calculate the monotonic prediction voltage V pred1 , set the univariate multi-time function Y1=a 0 +a 1 x+a 2 x 2 +a 3 x 3 , where a 0 、a 1 、a 2 、a 3 is a fitting parameter and takes values in the real number range, x is the power or temperature or humidity value; and a one-dimensional linear model Y2=ax+b is set, where a and b are fitting parameters and take values in the real number range, x is the power or temperature or humidity; corresponding to the monotonic prediction voltage V pred1 =(Y1+Y2) / 2; S305, calculate the comprehensive predicted voltage value V pred2 , the voltage-power model voltage V1, the voltage-temperature model voltage V2, and the voltage-humidity model voltage V3 are averaged to obtain the comprehensive predicted voltage value V pred2 =(V1+V2+V3) / 3.
6. A method for predicting attenuation of a vehicle-mounted fuel cell stack according to claim 1, It is characterized in that The following prediction result output steps are also included: S401. According to the fuel cell stack life test curve, establish a voltage-time model Y, Y=AX+B, where Y is voltage, X is time, and A and B are real constants; S402, extracting sample voltage V sample , to predict the fuel cell stack; S403: extract the sample voltage V sample , the comprehensive predicted voltage value Vpred2, and the preset polarization curve voltage are compared. When the extracted sample voltage V sample The voltage is less than the comprehensive predicted voltage value V for three consecutive times pred2 , and the extracted sample voltage V sample If the voltage is less than the preset polarization curve voltage three times in a row, it means that the fuel cell stack has decayed and the decay degree can be predicted; S404, calculate the attenuation percentage S = [1-(V sample / V pred2 )]x100%, according to the voltage-time model Y, calculate the time X=(YB)A, where X is the time value of the consumed design life.
7. A method for predicting attenuation of a vehicle-mounted fuel cell stack according to claim 1, Features: The influencing parameters also include operating time T, cumulative air flow Q, cumulative mileage k, current / voltage ratio, and power generation E.
8. A vehicle-mounted fuel cell stack attenuation prediction device, comprising a vehicle-mounted fuel cell stack attenuation prediction method according to any one of claims 1 to 7, Features: It includes a data preprocessing module, a model building module, a voltage prediction module, and an attenuation calculation module which are connected in sequence, and the data preprocessing module, the model building module, the voltage prediction module, and the attenuation calculation module are integrated in the fuel cell controller and are electrically connected to the processor of the fuel cell controller respectively.
9. A fuel cell vehicle, comprising a vehicle-mounted fuel cell stack attenuation prediction method or device according to any one of claims 1 to 8, Features: The fuel cell vehicle is provided with the on-board fuel cell stack attenuation prediction device, and uses the on-board fuel cell stack attenuation prediction method to monitor the service life of the fuel cell stack.
Citation Information
Patent Citations
Fuel cell life prediction method and device and computer readable storage medium
CN109683093A
Fuel cell residual life prediction method and device based on hybrid model
CN114994555A
Method, system, equipment and terminal for predicting remaining service life of fuel cell
CN115983084A
Fuel cell stack life prediction method and engine system and vehicle using same
CN118209865A
Method and apparatus for estimating battery life
KR101949449B1
