Fuel cell fault diagnosis method and system, electronic equipment and storage medium

Abstract

The invention provides a fuel cell fault diagnosis method and system, electronic equipment and a storage medium, and the method comprises the steps: obtaining a current operation condition parameter of a cell stack of a fuel cell and a current density distribution diagram corresponding to the current operation condition parameter at present, extracting a real-time multi-dimensional characteristic parameter of the cell stack based on the current density distribution diagram, calling a preset health reference current density distribution map matched with the current operation condition parameters from a preset multi-dimensional condition current density reference map library of the fuel cell stack in the health state, and comparing preset multi-dimensional characteristic parameters corresponding to the preset health reference current density distribution map with the real-time multi-dimensional characteristic parameters, when the comparison result meets a preset judgment condition, it is determined that the fuel cell stack breaks down, and the type and position of the fault are determined based on the real-time multi-dimensional characteristic parameters; on the basis of current density distribution scanning and multi-dimensional feature extraction, the current density is compared with a current density reference map library in a healthy state, and rapid and accurate fault diagnosis and fault positioning are achieved.

Description

Technical Field

[0001] This invention relates to the field of fuel cell technology, and in particular to a fuel cell fault diagnosis method, system, electronic device, and storage medium. Background Technology

[0002] A fuel cell is a chemical device that directly converts the chemical energy of fuel into electrical energy; it is also known as an electrochemical generator. Because fuel cells convert the Gibbs free energy portion of the chemical energy of fuel into electrical energy through an electrochemical reaction, they are not limited by the Carnot cycle effect and are therefore highly efficient. In addition, fuel cells use fuel and oxygen as raw materials and have no mechanical transmission parts, so they emit very few harmful gases and have a long service life. Fuel cells are widely used in vehicles.

[0003] To ensure the safety of fuel cells, real-time fault diagnosis is required. Currently, fault diagnosis of fuel cells mainly relies on monitoring the voltage and temperature of individual cells. However, these two parameters are a macroscopic, lagging, and highly coupled manifestation of the physicochemical processes inside the fuel cell stack, which makes it impossible to quickly and accurately locate faults and their positions when using these two parameters for fault diagnosis. Summary of the Invention

[0004] The present invention aims to at least solve one of the technical problems existing in related technologies. Therefore, the object of the present invention is to provide a fuel cell fault diagnosis method, system, electronic device, and storage medium.

[0005] This invention proposes a fuel cell fault diagnosis method, comprising the following steps: obtaining the current operating condition parameters of the fuel cell stack and its corresponding current density distribution map; extracting real-time multidimensional feature parameters of the fuel cell stack based on the current density distribution map, wherein the real-time multidimensional feature parameters include at least: a global uniformity index for characterizing overall uniformity, a local low current region feature for characterizing local insufficient response, a local high current region feature for characterizing local excessive response, and a symmetry index for characterizing distribution symmetry; calling a preset healthy benchmark current density distribution map matching the current operating condition parameters from a preset multidimensional operating condition current density benchmark map library of fuel cell stacks in a healthy state, comparing the preset multidimensional feature parameters corresponding to the preset healthy benchmark current density distribution map with the real-time multidimensional feature parameters; when the comparison result meets preset judgment conditions, determining that the fuel cell stack has failed, and determining the type and location of the failure based on the real-time multidimensional feature parameters.

[0006] According to the fuel cell fault diagnosis method of this invention, the current operating condition parameters of the fuel cell stack and its corresponding current density distribution map are first obtained. Then, real-time multi-dimensional feature parameters of the fuel cell stack are extracted based on the current density distribution map. Next, a preset healthy benchmark current density distribution map matching the current operating condition parameters is retrieved from a preset multi-dimensional operating condition current density benchmark map library of the fuel cell stack in a healthy state. The preset multi-dimensional feature parameters corresponding to the preset healthy benchmark current density distribution map are compared with the real-time multi-dimensional feature parameters. Finally, when the comparison result meets the preset judgment conditions, it is determined that the fuel cell stack has a fault, and the type and location of the fault are determined based on the real-time multi-dimensional feature parameters. Based on current density distribution scanning and multi-dimensional feature extraction, the internal state of the fuel cell stack can be visualized and monitored. By comparing with the current density benchmark map library in a healthy state, interference and lag problems can be avoided, and rapid and accurate fault diagnosis and fault location of the fuel cell can be achieved.

[0007] In addition, the fuel cell fault diagnosis method according to embodiments of the present invention may also have the following additional technical features: Furthermore, before obtaining the current operating condition parameters of the fuel cell stack and its corresponding current density distribution map, the method further includes: constructing and storing a multi-dimensional operating condition current density benchmark library of the fuel cell stack in a healthy state, including: constructing a multi-dimensional operating condition test matrix covering the typical operating range of the fuel cell stack, wherein the test parameters of the multi-dimensional operating condition test matrix include at least: load current, fuel cell stack temperature, anode inlet relative humidity, and cathode inlet relative humidity; controlling the fuel cell stack to operate stably at each set operating point, collecting the current density distribution data of all individual cells, and generating a standard healthy current density distribution map of each individual cell at the operating point; storing all the standard healthy current density distribution maps as a multi-dimensional lookup table to form the multi-dimensional operating condition current density benchmark library.

[0008] Further, the step of extracting real-time multidimensional feature parameters of the battery stack based on the current density distribution map includes: determining the ratio of the standard deviation to the mean of the current density distribution map as the global uniformity index; performing image segmentation on the current density distribution map to identify a first target connected region with an average current density lower than a first preset threshold, calculating the area ratio of the first target connected region and the centroid coordinates of the largest connected region, and determining the area ratio and the centroid coordinates as the local low current region features; performing image segmentation on the current density distribution map to identify a second target connected region with an average current density higher than a second preset threshold, calculating the peak current density and the maximum spatial gradient of the second target connected region, and determining the peak current density and the maximum spatial gradient as the local high current region features; dividing the real-time current density distribution map into two parts along the airflow direction, calculating the ratio of the average current densities of the two parts, and determining the ratio as the symmetry index.

[0009] Further, the step of determining that the fuel cell stack has failed when the comparison result meets the preset judgment conditions, and determining the type and location of the failure based on the real-time multidimensional feature parameters, includes: when the area ratio exceeds a preset area ratio threshold, and the first target connected region is a first preset shape and located in a first preset region, determining that the fuel cell stack has a cathode flooding failure, and determining the failure location information based on the characteristics of the local low current region; and / or, when the peak current density exceeds a preset current density threshold, and the maximum spatial gradient exceeds a preset spatial gradient threshold, determining that the fuel cell stack has a local membrane dryness failure or lack of [something]. The system identifies the following faults: 1) gas uniformity fault, and determines the fault location information based on the characteristics of the local high current region; 2) when the global uniformity index exceeds a preset uniformity threshold, and neither the local low current region characteristics nor the local high current region characteristics meet preset identification conditions, determines that the fuel cell stack has experienced an early performance uniformity degradation fault, and determines the fault location information based on the global uniformity index; 3) when the symmetry index is lower than a preset low symmetry threshold or higher than a preset high symmetry threshold, determines that the fuel cell stack has experienced a reactant gas uneven distribution fault or a unilateral flow channel blockage fault, and determines the fault location information based on the symmetry index.

[0010] Furthermore, the current density distribution map is obtained based on the current density sensor inside the fuel cell stack. The arrangement of the current density sensor includes: placing the current density sensor between the membrane electrode and the bipolar plate and contacting the diffusion layer of the membrane electrode; or embedding the current density sensor inside the bipolar plate and placing it in the cavity on the back side of the flow field of the bipolar plate; or attaching the current density sensor to the outside of the end plate of the fuel cell stack and attaching it to the current collector of the end cell.

[0011] Furthermore, after determining the type and location of the fault based on the real-time multidimensional feature parameters, the method further includes: outputting a diagnostic report based on the type and location of the fault and / or executing preset maintenance controls corresponding to the fault type and location, wherein the diagnostic report includes fault battery number, fault type, and fault location information.

[0012] Furthermore, after outputting a diagnostic report based on the type and location of the fault and / or executing preset maintenance controls corresponding to the fault type and location, the method further includes: performing self-learning based on the diagnostic results and / or maintenance control results to adaptively adjust and update the preset judgment conditions.

[0013] To address the aforementioned problems, this invention also proposes a fuel cell fault diagnosis system, comprising: a data acquisition module for acquiring the current operating condition parameters of the fuel cell stack and its corresponding current density distribution map; an extraction module for extracting real-time multidimensional feature parameters of the fuel cell stack based on the current density distribution map, wherein the real-time multidimensional feature parameters include at least: a global uniformity index for characterizing overall uniformity, a local low current region feature for characterizing local insufficient response, a local high current region feature for characterizing local excessive response, and a symmetry index for characterizing distribution symmetry; a comparison module for retrieving a preset healthy benchmark current density distribution map matching the current operating condition parameters from a preset multidimensional operating condition current density benchmark map library of fuel cell stacks in a healthy state, and comparing the preset multidimensional feature parameters corresponding to the preset healthy benchmark current density distribution map with the real-time multidimensional feature parameters; and a diagnosis module for determining that the fuel cell stack has failed when the comparison result meets preset judgment conditions, and determining the type and location of the failure based on the real-time multidimensional feature parameters.

[0014] According to the fuel cell fault diagnosis system of the present invention, the fuel cell fault diagnosis method of the above embodiment is executed. First, the current operating condition parameters of the fuel cell stack and its corresponding current density distribution map are obtained. Then, the real-time multi-dimensional feature parameters of the fuel cell stack are extracted based on the current density distribution map. Next, a preset healthy reference current density distribution map matching the current operating condition parameters is called from a preset multi-dimensional operating condition current density reference map library of the fuel cell stack in a healthy state. The preset multi-dimensional feature parameters corresponding to the preset healthy reference current density distribution map are compared with the real-time multi-dimensional feature parameters. Finally, when the comparison result meets the preset judgment conditions, it is determined that the fuel cell stack has a fault, and the type and location of the fault are determined based on the real-time multi-dimensional feature parameters. In this way, based on current density distribution scanning and multi-dimensional feature extraction, the internal state of the fuel cell stack can be visualized and monitored. By comparing with the current density reference map library in a healthy state, interference and lag problems can be avoided, and rapid and accurate fault diagnosis and fault location of the fuel cell can be achieved.

[0015] To address the aforementioned problems, the present invention also proposes an electronic device, comprising: a fuel cell fault diagnosis system as described in the above embodiments of the present invention; or, the electronic device comprises: a processor, a memory, and a fuel cell fault diagnosis program stored in the memory and executable on the processor, wherein the fuel cell fault diagnosis program, when executed by the processor, implements the fuel cell fault diagnosis method as described in the above embodiments of the present invention.

[0016] An electronic device according to an embodiment of the present invention is equipped with the fuel cell fault diagnosis system described in the above embodiment. The fuel cell fault diagnosis method described in the above embodiment is executed by first acquiring the current operating condition parameters of the fuel cell stack and its corresponding current density distribution map. Then, based on the current density distribution map, real-time multi-dimensional feature parameters of the fuel cell stack are extracted. Next, a preset healthy reference current density distribution map matching the current operating condition parameters is retrieved from a preset multi-dimensional operating condition current density reference map library of the fuel cell stack in a healthy state. The preset multi-dimensional feature parameters corresponding to the preset healthy reference current density distribution map are compared with the real-time multi-dimensional feature parameters. Finally, when the comparison result meets the preset judgment conditions, a fault is determined to have occurred in the fuel cell stack, and the type and location of the fault are determined based on the real-time multi-dimensional feature parameters. Based on current density distribution scanning and multi-dimensional feature extraction, the internal state of the fuel cell stack can be visualized and monitored. By comparing with the current density reference map library in a healthy state, interference and lag problems can be avoided, achieving rapid and accurate fault diagnosis and fault location of the fuel cell.

[0017] To address the aforementioned problems, the present invention also proposes a computer-readable storage medium storing a fuel cell fault diagnosis program, which, when executed by a processor, implements the fuel cell fault diagnosis method as described in the above embodiments of the present invention.

[0018] According to an embodiment of the present invention, when a computer-readable storage medium storing a fuel cell fault diagnosis program thereon is executed by a processor, the fuel cell fault diagnosis method of the above embodiment is executed. First, the current operating condition parameters of the fuel cell stack and its corresponding current density distribution map are obtained. Then, real-time multi-dimensional feature parameters of the fuel cell stack are extracted based on the current density distribution map. Next, a preset healthy reference current density distribution map matching the current operating condition parameters is retrieved from a preset multi-dimensional operating condition current density reference map library of the fuel cell stack in a healthy state. The preset multi-dimensional feature parameters corresponding to the preset healthy reference current density distribution map are compared with the real-time multi-dimensional feature parameters. Finally, when the comparison result meets preset judgment conditions, a fault is determined to have occurred in the fuel cell stack, and the type and location of the fault are determined based on the real-time multi-dimensional feature parameters. Thus, based on current density distribution scanning and multi-dimensional feature extraction, the internal state of the fuel cell stack can be visualized and monitored. By comparing it with the current density reference map library in a healthy state, interference and lag problems can be avoided, achieving rapid and accurate fault diagnosis and fault location of the fuel cell.

[0019] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0020] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the description of the embodiments taken in conjunction with the following drawings, in which: Figure 1 This is a flowchart of a fuel cell fault diagnosis method according to an embodiment of the present invention; Figure 2 This is a flowchart of a fuel cell fault diagnosis method according to another embodiment of the present invention; Figure 3 This is a flowchart of a fuel cell fault diagnosis method according to another embodiment of the present invention; Figure 4 This is a flowchart of a fuel cell fault diagnosis method according to yet another embodiment of the present invention; Figure 5 This is an overall architecture diagram of a fuel cell fault diagnosis system according to a specific embodiment of the present invention; Figure 6 This is a schematic diagram of the core arrangement scheme of a current density sensor according to a specific embodiment of the present invention; Figure 7 This is a flowchart of a fuel cell fault diagnosis method according to a specific embodiment of the present invention; Figure 8 This is a structural block diagram of a fuel cell fault diagnosis system according to an embodiment of the present invention.

[0021] Figure label: 100 - Fuel cell fault diagnosis system; 110 - Acquisition module; 120 - Extraction module; 130 - Comparison module; 140 - Multi-diagnosis module. Detailed Implementation

[0022] The embodiments of the present invention are described in detail below. The embodiments described with reference to the accompanying drawings are exemplary. The embodiments of the present invention are described in detail below.

[0023] Current fault diagnosis in fuel cells primarily relies on monitoring the voltage and temperature of individual cells. However, these two parameters represent a macroscopic, lagging, and highly coupled set of physicochemical processes within the fuel cell stack. Their inherent drawback lies in: (1) Diagnostic lag: When the voltage or temperature is obviously abnormal, the fault has often developed to a certain extent and may have caused irreversible damage to the battery, missing the best maintenance window.

[0024] (2) Positioning ambiguity: When the voltage or temperature of a single battery is abnormal, it can only tell that the battery is "sick", but it cannot accurately indicate "where the problem is" (such as anode water flooding, cathode gas deficiency, local blockage, etc.), making it difficult to guide precise maintenance.

[0025] (3) Susceptible to interference: Global changes in operating conditions (such as sudden load changes) will affect the voltage of all batteries at the same time, which can easily mask weak signals caused by local faults.

[0026] Current density distribution is a direct, local, and visual parameter reflecting the internal reaction state of a fuel cell. Local imbalances in water, gas, and heat management can directly lead to abnormal current density in that area. While current density scanners can measure distribution experimentally, transforming them from laboratory tools into a reliable early diagnostic method applicable to automotive or commercial systems, and establishing a precise mapping from current density images to fault types, remains a significant technical challenge.

[0027] To address the aforementioned problems in related technologies, embodiments of the present invention provide a fuel cell fault diagnosis method, system, electronic device, and storage medium, as described below. Figures 1-8 A method, system, electronic device, and storage medium for diagnosing fuel cell faults according to embodiments of the present invention are described.

[0028] Figure 1 This is a flowchart of a fuel cell fault diagnosis method according to an embodiment of the present invention. Figure 1 As shown, a fuel cell fault diagnosis method according to an embodiment of the present invention includes the following steps: Step S1: Obtain the current operating parameters of the fuel cell stack and its corresponding current density distribution map.

[0029] In a specific embodiment, the current operating parameters of the fuel cell stack and its corresponding current density distribution map are first obtained. Specifically, for example, key operating parameters such as load current and temperature of the fuel cell stack are read in real time, and stable current density data is collected through current density scanning sensors (such as segmented current collectors) to generate a standard healthy current density distribution map for each individual cell.

[0030] Specifically, according to the fuel cell fault diagnosis method of the present invention, the current operating condition parameters of the fuel cell stack and its corresponding current density distribution map are first obtained to characterize the internal reaction state of the fuel cell and provide an intuitive basis for subsequent fault diagnosis.

[0031] Step S2: Extract real-time multidimensional feature parameters of the battery stack based on the current density distribution map. The real-time multidimensional feature parameters include at least: a global uniformity index for characterizing overall uniformity, a local low current region feature for characterizing local insufficient response, a local high current region feature for characterizing local excessive response, and a symmetry index for characterizing distribution symmetry.

[0032] In a specific embodiment, real-time multidimensional feature parameters of the battery stack are then extracted based on the current density distribution map. Specifically, the real-time multidimensional feature parameters include at least: a global uniformity index for characterizing overall uniformity, features of local low-current regions for characterizing insufficient local response, features of local high-current regions for characterizing excessive local response, and a symmetry index for characterizing distribution symmetry.

[0033] Specifically, according to the fuel cell fault diagnosis method of the present invention, real-time multidimensional feature parameters of the battery stack are extracted based on the current density distribution map, thereby obtaining real-time multidimensional feature parameters characterizing the state of the battery stack as the core of diagnosis, which facilitates subsequent comprehensive and accurate fault diagnosis based on the real-time multidimensional feature parameters.

[0034] Step S3: Retrieve a preset healthy benchmark current density distribution map that matches the current operating parameters from the preset multi-dimensional operating condition current density benchmark map library of fuel cell stacks under healthy conditions, and compare the preset multi-dimensional feature parameters corresponding to the preset healthy benchmark current density distribution map with the real-time multi-dimensional feature parameters.

[0035] In a specific embodiment, a preset healthy benchmark current density distribution map matching the current operating condition parameters is then retrieved from a preset multi-dimensional operating condition current density benchmark map library of fuel cell stacks under healthy conditions. The preset multi-dimensional feature parameters corresponding to the preset healthy benchmark current density distribution map are then compared with the real-time multi-dimensional feature parameters. Specifically, the preset multi-dimensional operating condition current density benchmark map library of fuel cell stacks under healthy conditions includes multiple multi-dimensional operating condition current density benchmark maps of fuel cell stacks under healthy conditions, and each multi-dimensional operating condition current density benchmark map under healthy conditions includes corresponding preset multi-dimensional feature parameters.

[0036] Specifically, according to the fuel cell fault diagnosis method of the present invention, a preset healthy reference current density distribution map matching the current operating condition parameters is then retrieved from a preset multi-dimensional operating condition current density reference map library of fuel cell stacks in a healthy state. The preset multi-dimensional feature parameters corresponding to the preset healthy reference current density distribution map are compared with the real-time multi-dimensional feature parameters to quickly determine the healthy reference current density distribution map and its corresponding preset multi-dimensional feature parameters corresponding to the current operating condition parameters. Comparing the preset multi-dimensional feature parameters with the real-time multi-dimensional feature parameters facilitates subsequent rapid fault diagnosis and fault location determination based on the comparison results.

[0037] Step S4: When the comparison result meets the preset judgment conditions, it is determined that the fuel cell stack has failed, and the type and location of the failure are determined based on real-time multidimensional feature parameters.

[0038] In a specific embodiment, when the comparison result meets the preset judgment conditions, a fault is determined to have occurred in the fuel cell stack, and the type and location of the fault are determined based on real-time multi-dimensional feature parameters. Specifically, the preset judgment conditions are set as needed and can be adaptively updated based on factors such as diagnostic accuracy.

[0039] Specifically, according to the fuel cell fault diagnosis method of the present invention, when the comparison result meets the preset judgment conditions, it is determined that the fuel cell stack has failed, and the type and location of the fault are determined based on real-time multidimensional feature parameters, thereby enabling rapid and accurate fault diagnosis and determination of the fault type and location.

[0040] Therefore, the fuel cell fault diagnosis method according to embodiments of the present invention first obtains the current operating condition parameters of the fuel cell stack and its corresponding current density distribution map. Then, based on the current density distribution map, it extracts the real-time multidimensional feature parameters of the fuel cell stack. Next, it calls a preset healthy benchmark current density distribution map that matches the current operating condition parameters from a preset multidimensional operating condition current density benchmark map library of the fuel cell stack in a healthy state. It compares the preset multidimensional feature parameters corresponding to the preset healthy benchmark current density distribution map with the real-time multidimensional feature parameters. Finally, when the comparison result meets the preset judgment conditions, it is determined that the fuel cell stack has a fault, and the type and location of the fault are determined based on the real-time multidimensional feature parameters. In this way, based on current density distribution scanning and multidimensional feature extraction, the internal state of the fuel cell stack can be visualized and monitored. By comparing it with the current density benchmark map library in a healthy state, interference and lag problems can be avoided, and rapid and accurate fault diagnosis and fault location of the fuel cell can be achieved.

[0041] Figure 2 This is a flowchart of a fuel cell fault diagnosis method according to another embodiment of the present invention. Figure 2 As shown, in one embodiment of the present invention, before obtaining the current operating condition parameters of the fuel cell stack and its corresponding current density distribution map in step S1, the method further includes: step S0: constructing and storing a multi-dimensional operating condition current density benchmark library of the fuel cell stack in a healthy state, including: constructing a multi-dimensional operating condition test matrix covering the typical operating range of the fuel cell stack, wherein the test parameters of the multi-dimensional operating condition test matrix include at least: load current, fuel cell stack temperature, anode inlet relative humidity and cathode inlet relative humidity; controlling the fuel cell stack to operate stably at each set operating point, collecting the current density distribution data of all single cells, and generating a standard healthy current density distribution map of each single cell at the operating point; storing all the standard healthy current density distribution maps as a multi-dimensional lookup table to form a multi-dimensional operating condition current density benchmark library.

[0042] In a specific embodiment, before obtaining the current operating condition parameters of the fuel cell stack and its corresponding current density distribution map, a multi-dimensional operating condition current density benchmark library of the fuel cell stack in a healthy state is constructed and stored. This includes: firstly, constructing a multi-dimensional operating condition test matrix covering the typical operating range of the fuel cell stack; then, controlling the stable operation of the fuel cell stack at each set operating point, collecting the current density distribution data of all individual cells, generating a standard healthy current density distribution map of each individual cell at the operating point; and finally, storing all the standard healthy current density distribution maps as a multi-dimensional lookup table to form a multi-dimensional operating condition current density benchmark library. Specifically, the test parameters of the multi-dimensional operating condition test matrix include at least: load current, fuel cell stack temperature, anode inlet relative humidity, and cathode inlet relative humidity. For example, the load current is set with at least 5 step points from idle speed to rated power; the fuel cell stack temperature is set with at least 3 temperature points within ±15°C of the rated operating point; the anode inlet relative humidity covers low, medium, and high humidity settings; and the cathode inlet relative humidity covers low, medium, and high humidity settings.

[0043] Specifically, according to the fuel cell fault diagnosis method of the present invention, before obtaining the current operating condition parameters of the fuel cell stack and its corresponding current density distribution map, a multi-dimensional operating condition current density benchmark library of the fuel cell stack in a healthy state is constructed and stored. This includes: firstly, constructing a multi-dimensional operating condition test matrix covering the typical operating range of the fuel cell stack; then, at each set operating point, controlling the fuel cell stack to operate stably, collecting the current density distribution data of all individual cells, generating a standard healthy current density distribution map of each individual cell at the operating point; and finally, storing all the standard healthy current density distribution maps as a multi-dimensional lookup table to form a multi-dimensional operating condition current density benchmark library. Thus, before fault diagnosis, the multi-dimensional operating condition test matrix is ​​used to construct and store the multi-dimensional operating condition current density benchmark library of the fuel cell stack in a healthy state, facilitating the direct retrieval of preset healthy benchmark current density distribution maps and enabling rapid comparison of preset multi-dimensional characteristic parameters with other multi-dimensional characteristic parameters, thereby ensuring the speed of fault diagnosis.

[0044] In one embodiment of the present invention, step S2 extracts real-time multidimensional feature parameters of the battery stack based on the current density distribution map, including: determining the ratio of the standard deviation to the average value of the current density distribution map as the global uniformity index; performing image segmentation on the current density distribution map to identify a first target connected region with an average current density lower than a first preset threshold, calculating the area ratio of the first target connected region and the centroid coordinates of the largest connected region, and determining the area ratio and centroid coordinates as features of a local low current region; performing image segmentation on the current density distribution map to identify a second target connected region with an average current density higher than a second preset threshold, calculating the peak current density and the maximum spatial gradient of the second target connected region, and determining the peak current density and the maximum spatial gradient as features of a local high current region; dividing the real-time current density distribution map into two parts along the airflow direction, calculating the ratio of the average current densities of the two parts, and determining the ratio as the symmetry index.

[0045] In a specific embodiment, the ratio of the standard deviation to the average value of the current density distribution map is determined as the global uniformity index; the current density distribution map is segmented to identify first target connected regions with average current densities lower than a first preset threshold, and the area ratio of the first target connected regions and the centroid coordinates of the largest connected region are calculated, determining the area ratio and centroid coordinates as features of local low-current regions; the current density distribution map is segmented to identify second target connected regions with average current densities higher than a second preset threshold, and the peak current density and maximum spatial gradient of the second target connected regions are calculated, determining the peak current density and maximum spatial gradient as features of local high-current regions; the real-time current density distribution map is divided into two parts along the airflow direction, and the ratio of the average current densities of the two parts is calculated, determining the ratio as the symmetry index. Specifically, the first and second preset thresholds are set as needed, for example, the first preset threshold is 0.7 times the average current density of the current density distribution map, and the second preset threshold is for example, 1.3 times the average current density of the current density distribution map.

[0046] Specifically, according to the fuel cell fault diagnosis method of the present invention, the ratio of the standard deviation to the average value of the current density distribution map is determined as the global uniformity index; the current density distribution map is segmented to identify a first target connected region with an average current density lower than a first preset threshold, and the area ratio of the first target connected region and the centroid coordinates of the largest connected region are calculated to determine the area ratio and centroid coordinates as features of a local low current region; the current density distribution map is segmented to identify a second target connected region with an average current density higher than a second preset threshold, and the peak current density and maximum spatial gradient of the second target connected region are calculated to determine the peak current density and maximum spatial gradient as features of a local high current region; the real-time current density distribution map is divided into two parts along the airflow direction, and the ratio of the average current densities of the two parts is calculated to determine the ratio as the symmetry index; in this way, real-time multidimensional feature parameters of the fuel cell stack can be accurately extracted based on the current density distribution map, thereby obtaining real-time multidimensional feature parameters characterizing the state of the fuel cell stack as the core of diagnosis, which facilitates subsequent comprehensive and accurate fault diagnosis based on the real-time multidimensional feature parameters.

[0047] In one embodiment of the present invention, step S4, when the comparison result meets the preset judgment conditions, determines that the fuel cell stack has failed, and determines the type and location of the failure based on real-time multidimensional feature parameters, including: when the area ratio exceeds the preset area ratio threshold, and the first target connected region is a first preset shape and located in the first preset region, it is determined that the fuel cell stack has a cathode flooding failure, and the failure location information is determined based on the characteristics of the local low current region; and / or, when the peak current density exceeds the preset current density threshold, and the maximum spatial gradient exceeds the preset spatial gradient threshold, it is determined that the fuel cell stack has a local membrane dryness failure or a gas shortage failure, and the failure location information is determined based on the characteristics of the local high current region; and / or, when the global uniformity index exceeds the preset uniformity threshold, and neither the local low current region characteristics nor the local high current region characteristics meet the preset identification conditions, it is determined that the fuel cell stack has an early performance uniformity degradation failure, and the failure location information is determined based on the global uniformity index; and / or, when the symmetry index is lower than the preset low symmetry threshold or higher than the preset high symmetry threshold, it is determined that the fuel cell stack has a reactant gas uneven distribution failure or a single-sided flow channel blockage failure, and the failure location information is determined based on the symmetry index.

[0048] In a specific embodiment, when the area ratio exceeds a preset area ratio threshold, and the first target connected region has a first preset shape and is located in a first preset region, it is determined that a cathode flooding fault has occurred in the fuel cell stack, and the fault location information is determined based on the characteristics of the local low current region. Specifically, the preset area ratio threshold is set as needed, the first preset shape is, for example, a block shape, and the first preset region is, for example, the downstream of the flow channel.

[0049] In a specific embodiment, when the peak current density exceeds a preset current density threshold and the maximum spatial gradient exceeds a preset spatial gradient threshold, a local membrane dryness fault or gas shortage fault is determined in the fuel cell stack, and the fault location information is determined based on the characteristics of the local high current region. Specifically, the preset current density threshold and preset spatial gradient threshold are set as needed. When the peak current density exceeds the preset current density threshold, the maximum spatial gradient exceeds the preset spatial gradient threshold, and the temperature in the corresponding region is relatively high, it is more likely to determine that a local membrane dryness fault or gas shortage fault has occurred in the fuel cell stack.

[0050] In a specific embodiment, when the global uniformity index exceeds a preset uniformity threshold, and neither the local low-current region characteristics nor the local high-current region characteristics meet the preset identification conditions, it is determined that an early performance uniformity degradation fault has occurred in the fuel cell stack, and the fault location information is determined based on the global uniformity index. Specifically, the preset uniformity threshold is set as needed, and meeting the preset identification conditions is, for example, a significant local low-current region characteristic or a significant local high-current region characteristic.

[0051] In a specific embodiment, when the symmetry index is lower than a preset low symmetry threshold or higher than a preset high symmetry threshold, it is determined that the fuel cell stack has experienced a reactant gas uneven distribution fault or a unilateral flow channel blockage fault, and the fault location information is determined based on the symmetry index. Specifically, the low symmetry threshold and the high symmetry threshold are set as needed.

[0052] Specifically, according to the fuel cell fault diagnosis method of the present invention, when the area ratio exceeds a preset area ratio threshold, and the first target connected region is a first preset shape and located in the first preset region, it is determined that the fuel cell stack has a cathode flooding fault, and the fault location information is determined based on the characteristics of the local low current region; when the peak current density exceeds a preset current density threshold, and the maximum spatial gradient exceeds a preset spatial gradient threshold, it is determined that the fuel cell stack has a local membrane dryness fault or a gas shortage fault, and the fault location information is determined based on the characteristics of the local high current region; when the global uniformity index exceeds a preset uniformity threshold, and the characteristics of the local low current region and the local high current region are both present, the fault location information is determined based on the characteristics of the local high current region. If none of the region features meet the preset identification conditions, it is determined that the fuel cell stack has experienced an early performance uniformity degradation fault, and the fault location information is determined based on the global uniformity index. When the symmetry index is lower than the preset low symmetry threshold or higher than the preset high symmetry threshold, it is determined that the fuel cell stack has experienced a reactant gas uneven distribution fault or a unilateral flow channel blockage fault, and the fault location information is determined based on the symmetry index. In this way, when the comparison results meet the preset judgment conditions, it is possible to accurately determine that the fuel cell stack has experienced a fault, and accurately determine the type and location of the fault based on real-time multidimensional feature parameters, thereby enabling rapid and accurate fault diagnosis and determination of fault type and location.

[0053] In one embodiment of the present invention, the current density distribution map is obtained based on the current density sensor inside the fuel cell stack. The arrangement of the current density sensor includes: placing the current density sensor between the membrane electrode and the bipolar plate and contacting the diffusion layer of the membrane electrode; or embedding the current density sensor inside the bipolar plate and placing it in the cavity on the back side of the flow field of the bipolar plate; or attaching the current density sensor to the outside of the end plate of the fuel cell stack and attaching it to the current collector of the end cell.

[0054] In a specific embodiment, the current density distribution map is obtained based on current density sensors inside the fuel cell stack. Specifically, for example, the current density distribution map can be obtained through an array of current density sensors.

[0055] In a specific embodiment, the current density sensor is arranged by placing it between the membrane electrode assembly (MEA) and the bipolar plate, and in contact with the diffusion layer of the MEA. Specifically, the sensor directly replaces the traditional current collector, located between the MEA and the graphite bipolar plate, in direct contact with the MEA diffusion layer to ensure current conduction efficiency. Its thickness is comparable to that of a traditional current collector, and a sealing groove is provided around the sensor to ensure compatibility with the existing sealing system of the battery stack.

[0056] In a specific embodiment, the current density sensor is arranged by embedding it inside a bipolar plate and placing it within a cavity on the back side of the bipolar plate's flow channel. Specifically, the sensor function is integrated into the bipolar plate, and the sensor cavity is fabricated on the back side of the bipolar plate's flow channel to ensure that the flow field performance is not affected.

[0057] In a specific embodiment, the current density sensor is arranged by attaching it to the outside of the battery stack end plate and abutting it against the current collector of the last single cell. Specifically, the sensor is attached to the outside of the battery stack end plate and tightly fitted against the current collector of the last single cell. This method is suitable for upgrading existing battery stacks and utilizes a high-sensitivity magnetoresistive sensor array.

[0058] Specifically, according to the fuel cell fault diagnosis method of the present invention, the current density distribution map is obtained based on the current density sensor inside the fuel cell stack. The arrangement of the current density sensor includes: placing the current density sensor between the membrane electrode and the bipolar plate and contacting the diffusion layer of the membrane electrode; embedding the current density sensor inside the bipolar plate and placing it in the cavity on the back side of the flow field of the bipolar plate; or attaching the current density sensor to the outside of the end plate of the fuel cell stack and attaching it to the current collector of the end cell. In this way, different arrangement methods can be used to arrange the current density sensor. The current density distribution map is obtained by collecting the current density sensor through different arrangement methods. Not only can the internal reaction state of the fuel cell be characterized by the current density distribution map, but the applicability of the current density distribution map can also be improved.

[0059] Figure 3 This is a flowchart of a fuel cell fault diagnosis method according to another embodiment of the present invention. Figure 3 As shown, in one embodiment of the present invention, after determining the type and location of the fault based on real-time multidimensional feature parameters in step S4, the method further includes: step S5: outputting a diagnostic report based on the type and location of the fault and / or executing preset maintenance controls corresponding to the fault type and location, wherein the diagnostic report includes fault battery number, fault type and fault location information.

[0060] In a specific embodiment, after determining the type and location of the fault based on real-time multidimensional feature parameters, a diagnostic report is output and / or preset maintenance controls corresponding to the fault type and location are executed based on the fault type and location. Specifically, the diagnostic report includes the faulty battery number, fault type, and fault location information, and the maintenance controls include, but are not limited to, drying and humidifying.

[0061] Specifically, according to the fuel cell fault diagnosis method of the present invention, after determining the type and location of the fault based on real-time multidimensional feature parameters, a diagnostic report is output based on the type and location of the fault and / or preset maintenance control corresponding to the fault type and location is executed; this facilitates users to understand the fault situation in a timely manner and to automatically perform maintenance when a fault occurs, which helps to ensure the safety of the fuel cell.

[0062] Figure 4 This is a flowchart of a fuel cell fault diagnosis method according to another embodiment of the present invention. Figure 4 As shown, in one embodiment of the present invention, after step S5 outputs a diagnostic report based on the type and location of the fault and / or executes a preset maintenance control corresponding to the fault type and location, the method further includes: step S6: performing self-learning based on the diagnostic results and / or maintenance control results to adaptively adjust and update the preset judgment conditions.

[0063] In a specific embodiment, after outputting a diagnostic report based on the fault type and location and / or executing preset maintenance controls corresponding to the fault type and location, self-learning is performed based on the diagnostic results and / or maintenance control results to adaptively adjust and update the preset judgment conditions. Specifically, for example, self-learning is performed through a learning model to update the adjusted parameters, strategies, and new system knowledge, and to continuously collect current density data and perform fault diagnosis.

[0064] Specifically, according to the fuel cell fault diagnosis method of the present invention, after outputting a diagnostic report based on the fault type and location and / or executing preset maintenance controls corresponding to the fault type and location, self-learning is performed based on the diagnostic results and / or maintenance control results to adaptively adjust and update the preset judgment conditions; thereby, fault diagnosis can be made more realistic through self-learning, which helps to improve the intelligence of fault diagnosis.

[0065] The following specific embodiment further illustrates the fuel cell fault diagnosis method of the present invention. In this specific embodiment, a fuel cell fault diagnosis system and method are provided, which can detect the initial faults of the fuel cell earlier than traditional voltage / temperature diagnosis methods, accurately locate the specific location of the fault within the active area of ​​a single cell, and intelligently identify the root cause of the fault to provide clear guidance for subsequent accurate maintenance (such as directional purging), while improving the reliability of diagnosis and reducing false alarms.

[0066] Figure 5 This is an overall architecture diagram of a fuel cell fault diagnosis system according to a specific embodiment of the present invention. Figure 5 As shown in this specific embodiment, the fuel cell fault diagnosis system includes: a hardware layer, an offline learning module, an online monitoring module, an intelligent diagnosis module, and a decision output module.

[0067] In this specific embodiment, the hardware layer includes a current density sensor, a multi-condition sensor, and a signal acquisition unit for current density scanning, multi-condition and other signal acquisition.

[0068] In this specific embodiment, the offline learning module is used to construct a "multi-dimensional operating condition matrix of current, temperature, humidity, etc.", scan and obtain a "healthy current density distribution map" under healthy conditions, and establish a "healthy benchmark library stored as a multi-dimensional lookup table".

[0069] In this specific embodiment, the offline learning module constructs a multi-dimensional operating condition matrix by defining a test plan covering the typical operating range of the battery stack. Specifically, the matrix includes at least: load current (I): at least 5 step points from idle speed to rated power; battery stack temperature (T): at least 3 temperature points within ±15°C of the rated operating point; anode inlet relative humidity (RH_an): covering low, medium, and high humidity settings; cathode inlet relative humidity (RH_ca): similarly covering different humidity levels.

[0070] In this specific embodiment, the offline learning module collects the healthy current density distribution map by allowing the battery stack to operate stably at each set operating point (I, T, RH_an, RH_ca). Specifically, a current density scanning sensor (such as a segmented current collector) is used to collect the stabilized current density data, and a "standard healthy current density distribution map" J_healthy(x, y) is generated for each individual cell.

[0071] In this specific embodiment, the offline learning module establishes a health benchmark library by storing all collected J_healthy graphs in a multidimensional lookup table, using their corresponding operating condition parameters (I, T, RH_an, RH_ca) as index keys; this library serves as a diagnostic benchmark and can be updated subsequently.

[0072] In this specific embodiment, the online monitoring module is used to collect operating condition data and current density distribution map in real time and extract a multi-dimensional feature parameter system from the distribution map.

[0073] In this specific embodiment, the real-time data synchronous acquisition of the online monitoring module includes: acquisition of operating conditions: real-time reading of key operating parameters such as load current I_real and temperature T_real of the battery stack; acquisition of distribution map: the current density scanning sensor synchronously acquires the real-time current density distribution map J_real(x, y) of all individual cells at a frequency of not less than 1Hz.

[0074] In this specific embodiment, the online monitoring module extracts a multidimensional feature parameter system, which includes: calculating a set of quantitative feature parameters from each J_real(x, y). This system is the core of the diagnosis and must simultaneously include the following four categories: 1) Global uniformity index U: The ratio of the standard deviation σ of the current density in the entire distribution map to the average value J_avg (U = σ / J_avg), which reflects the overall uniformity of the response.

[0075] 2) Characteristics of the local low current region F_low: Location: Set the threshold Th_low = 0.7 × J_avg, and identify connected low-current regions using an image segmentation algorithm.

[0076] Quantification: Calculate its area percentage A_low% and the centroid coordinates (X_low, Y_low) of the largest region.

[0077] 3) Characteristics of the local high current region F_high: Location: Set a threshold Th_high = 1.3 × J_avg to identify hotspot areas.

[0078] Quantization: Calculate its peak current density J_peak and spatial gradient Grad_max.

[0079] 4) Distribution symmetry index S: Divide the left and right regions along the airflow direction and calculate the average current density ratio S = J_avg_left / J_avg_right, which is used to detect asymmetric faults.

[0080] In this specific embodiment, the intelligent diagnostic module is used to call the "health baseline map" that matches the current working condition and execute the fault identification rule base to determine the fault type and location.

[0081] In this specific embodiment, the intelligent diagnosis module dynamically calls the health benchmark map by: using the nearest neighbor interpolation algorithm to call the most matching J_healthy from the health benchmark map library based on the real-time collected operating condition parameters (I_real, T_real, ...).

[0082] In this specific embodiment, the intelligent diagnostic module determines the fault type and location by comparing real-time features with J_healthy and applying a preset fault identification rule base for judgment; the core mapping relationship of this rule base includes: 1) IF(A_low%>A_th) AND (region shape is blocky) AND (location is downstream of the flow channel) THEN is diagnosed as "cathode flooding", with high confidence.

[0083] 2) IF(J_peak>J_th) AND (Grad_max>G_th) AND (corresponding area temperature is too high) THEN is diagnosed as "local membrane dryness / reaction gas deficiency".

[0084] 3) If (U>U_th) AND (F_low and F_high features are not significant) THEN, issue an early warning of "early performance uniformity degradation".

[0085] 4) IF(S<S_low_th) OR (S> S_high_th)THEN indicates "uneven air / hydrogen distribution or blockage on one side of the flow path".

[0086] In this specific embodiment, the decision output module is used to generate and output a diagnostic report containing the battery ID (IdentityDocument), fault type, location, and level, triggering precise maintenance actions or system parameter adjustments.

[0087] In this specific embodiment, the decision output module generates a diagnostic report including: outputting structured diagnostic results, the content of which includes at least: faulty battery ID, fault type (such as flooding, lack of gas), fault location (two-dimensional coordinates), severity level, timestamp, etc.

[0088] In this specific embodiment, the decision output module triggers precise maintenance by directly transmitting diagnostic results to the fuel cell main controller, which can automatically trigger targeted maintenance actions. Specifically, for example, for a "flooded" battery, a targeted high-pressure purging is performed; for a "membrane dryness" trend, the system humidification strategy is automatically adjusted.

[0089] In this specific embodiment, the decision output module archives all data from this diagnosis for use in optimizing the diagnostic algorithm and updating the health benchmark library, thereby enabling the system to learn itself.

[0090] Figure 6 This is a schematic diagram of the core arrangement scheme of a current density sensor according to a specific embodiment of the present invention. Figure 6 As shown in this specific embodiment, three core arrangement schemes for the current density sensor are also provided to ensure that the current density sensor can effectively collect data without negatively impacting the performance of the battery stack: Option 1: Alternative Current Collector Layout. Specifically, the current density sensor directly replaces the traditional current collector, located between the membrane electrode assembly (MEA) and the graphite bipolar plate, in direct contact with the MEA diffusion layer to ensure current conduction efficiency. Its thickness is comparable to that of the traditional current collector, and a sealing groove is set around the current density sensor to ensure compatibility with the original sealing system of the battery stack.

[0091] Option 2: Embedded bipolar plate arrangement. Specifically, the current density sensor function is integrated into the bipolar plate, and the sensor cavity is machined on the back of the bipolar plate flow channel to ensure that the flow field performance is not affected.

[0092] Option 3: External endplate arrangement. Specifically, the current density sensor is attached to the outside of the battery stack endplate, closely fitting the current collector of the last single cell. This option is suitable for upgrading existing battery stacks and uses a high-sensitivity magnetoresistive sensor array.

[0093] Figure 7 This is a flowchart of a fuel cell fault diagnosis method according to a specific embodiment of the present invention. Figure 7 As shown in this specific embodiment, the fuel cell fault diagnosis method includes the following steps: Step S10: System Startup. The fuel cell system starts up, the data acquisition system, multi-condition sensors, current density sensors and other controllers are powered on and initialized, and preset diagnostic threshold parameters are loaded.

[0094] Step S20: Load the offline database. Load the established offline database "Health Benchmark Library Multidimensional Lookup Table".

[0095] Step S30: Real-time data acquisition. Real-time reading of key operating parameters of the battery stack, such as load current I_real and temperature T_real, as well as current density distribution.

[0096] Step S40: Data preprocessing. Extract a multidimensional feature parameter system from the distribution map.

[0097] Step S50: Intelligent Diagnosis. The system retrieves the "health baseline map" matching the current operating conditions and executes the fault identification rule base to determine the fault type and location.

[0098] Step S60: Generate a diagnostic report. Output structured diagnostic results, including at least: faulty battery ID, fault type (e.g., flooding, low air pressure), fault location (two-dimensional coordinates), severity level, and timestamp.

[0099] Step S70: Trigger baseline maintenance. The diagnostic results are directly transmitted to the fuel cell main controller, which can automatically trigger targeted maintenance actions. For example, for a "flooded" battery, a targeted high-pressure purging is performed; for a "membrane dryness" trend, the system humidification strategy is automatically adjusted.

[0100] Step S80: Effectiveness Evaluation. Evaluate the diagnostic results and maintenance, with results categorized as "Excellent," "Good," "Average," and "Poor." Record successful experiences, parameter adjustments, strategy optimizations, system retraining, and other operations for each result.

[0101] Step S90: Knowledge Update. Update the knowledge of the adjusted parameters, strategies, and new system, and continuously collect current density data and diagnose faults.

[0102] As can be seen in this specific embodiment, the fuel cell fault diagnosis system and method aim to overcome the lag and ambiguity of existing diagnostic technologies and solve the following core problems: (1) How to achieve "early warning" of faults: before a significant change in the voltage or temperature of a single cell occurs, potential faults can be identified by capturing minute anomalies in the current density distribution; (2) How to achieve "precise location and type identification" of faults: not only can it be determined which cell is faulty, but also which area inside the cell the fault is located in, and what type it is (e.g., flooding, membrane dryness, gas shortage); (3) How to reduce the false alarm rate: distinguish between global current density changes caused by normal load changes and abnormal distribution patterns caused by local faults. The fuel cell fault diagnosis system and method are based on current density distribution scanning technology to visualize and monitor the internal state of the fuel cell stack, thereby achieving early and accurate fault diagnosis at the single cell level.

[0103] In summary, in this specific embodiment, the fuel cell fault diagnosis method has the following significant advantages compared with related technologies: 1. Foresight: It can issue early warnings when water and gas management are just beginning to show slight imbalances, before causing significant changes in voltage / temperature, achieving true early diagnosis. 2. Accuracy: It provides two-dimensional spatial location and type identification of faults, advancing maintenance from "replacing the entire battery" to "targeted treatment of specific areas of the battery," guiding precise maintenance. 3. High reliability: Current density is a direct reflection of local reactions, with strong anti-interference capabilities, effectively distinguishing between global operating condition changes and local real faults, significantly reducing the false alarm rate. 4. Visualization: The current density distribution map provides an intuitive "electrocardiogram," greatly improving the observability of the system status and facilitating fault analysis and system optimization.

[0104] In summary, the fuel cell fault diagnosis method according to embodiments of the present invention first obtains the current operating condition parameters of the fuel cell stack and its corresponding current density distribution map. Then, it extracts real-time multidimensional feature parameters of the fuel cell stack based on the current density distribution map. Next, it retrieves a preset healthy benchmark current density distribution map that matches the current operating condition parameters from a preset multidimensional operating condition current density benchmark map library of the fuel cell stack in a healthy state. It compares the preset multidimensional feature parameters corresponding to the preset healthy benchmark current density distribution map with the real-time multidimensional feature parameters. Finally, when the comparison result meets the preset judgment conditions, it determines that the fuel cell stack has a fault, and determines the type and location of the fault based on the real-time multidimensional feature parameters. In this way, based on current density distribution scanning and multidimensional feature extraction, the internal state of the fuel cell stack can be visualized and monitored. By comparing it with the current density benchmark map library in a healthy state, interference and lag problems can be avoided, and rapid and accurate fault diagnosis and fault location of the fuel cell can be achieved.

[0105] A further embodiment of the present invention discloses a fuel cell fault diagnosis system. Figure 8 This is a structural block diagram of a fuel cell fault diagnosis system according to an embodiment of the present invention, as shown below. Figure 8 As shown, the fuel cell fault diagnosis system 100 includes: a data acquisition module 110, an extraction module 120, a comparison module 130, and a diagnosis module 140.

[0106] Specifically, the acquisition module 110 is used to acquire the current operating condition parameters of the fuel cell stack and its corresponding current density distribution map; the extraction module 120 is used to extract the real-time multidimensional feature parameters of the fuel cell stack based on the current density distribution map, wherein the real-time multidimensional feature parameters include at least: a global uniformity index for characterizing overall uniformity, a local low current region feature for characterizing local insufficient response, a local high current region feature for characterizing local excessive response, and a symmetry index for characterizing distribution symmetry; the comparison module 130 is used to call a preset health benchmark current density distribution map that matches the current operating condition parameters from a preset multidimensional operating condition current density benchmark map library of fuel cell stacks in a healthy state, and compare the preset multidimensional feature parameters corresponding to the preset health benchmark current density distribution map with the real-time multidimensional feature parameters; the diagnosis module 140 is used to determine that the fuel cell stack has failed when the comparison result meets the preset judgment conditions, and to determine the type and location of the failure based on the real-time multidimensional feature parameters.

[0107] In one embodiment of the present invention, the acquisition module 110 is further configured to: before acquiring the current operating condition parameters of the fuel cell stack and its corresponding current density distribution map, construct and store a multi-dimensional operating condition current density benchmark library of the fuel cell stack in a healthy state, including: constructing a multi-dimensional operating condition test matrix covering the typical operating range of the fuel cell stack, wherein the test parameters of the multi-dimensional operating condition test matrix include at least: load current, fuel cell stack temperature, anode inlet relative humidity and cathode inlet relative humidity; controlling the fuel cell stack to operate stably at each set operating point, acquiring the current density distribution data of all individual cells, generating a standard healthy current density distribution map of each individual cell at the operating point; and storing all the standard healthy current density distribution maps as a multi-dimensional lookup table to form a multi-dimensional operating condition current density benchmark library.

[0108] In one embodiment of the present invention, the extraction module 120 extracts real-time multidimensional feature parameters of the battery stack based on the current density distribution map, including: determining the ratio of the standard deviation to the average value of the current density distribution map as the global uniformity index; performing image segmentation on the current density distribution map to identify a first target connected region with an average current density lower than a first preset threshold, calculating the area ratio of the first target connected region and the centroid coordinates of the largest connected region, and determining the area ratio and centroid coordinates as features of a local low current region; performing image segmentation on the current density distribution map to identify a second target connected region with an average current density higher than a second preset threshold, calculating the peak current density and the maximum spatial gradient of the second target connected region, and determining the peak current density and the maximum spatial gradient as features of a local high current region; dividing the real-time current density distribution map into two parts along the airflow direction, calculating the ratio of the average current densities of the two parts, and determining the ratio as the symmetry index.

[0109] In one embodiment of the present invention, when the comparison result meets the preset judgment conditions, the diagnostic module 140 determines that the fuel cell stack has failed, and determines the type and location of the failure based on real-time multidimensional feature parameters, including: when the area ratio exceeds the preset area ratio threshold, and the first target connected region is a first preset shape and located in the first preset region, it determines that the fuel cell stack has a cathode flooding failure, and determines the failure location information based on the characteristics of the local low current region; and / or, when the peak current density exceeds the preset current density threshold, and the maximum spatial gradient exceeds the preset spatial gradient threshold, it determines that the fuel cell stack has a local membrane dryness failure or a gas shortage failure, and determines the failure location information based on the characteristics of the local high current region; and / or, when the global uniformity index exceeds the preset uniformity threshold, and neither the local low current region characteristics nor the local high current region characteristics meet the preset identification conditions, it determines that the fuel cell stack has an early performance uniformity degradation failure, and determines the failure location information based on the global uniformity index; and / or, when the symmetry index is lower than the preset low symmetry threshold or higher than the preset high symmetry threshold, it determines that the fuel cell stack has a reactant gas uneven distribution failure or a unilateral flow channel blockage failure, and determines the failure location information based on the symmetry index.

[0110] In one embodiment of the present invention, the current density distribution map is obtained based on the current density sensor inside the fuel cell stack. The arrangement of the current density sensor includes: placing the current density sensor between the membrane electrode and the bipolar plate and contacting the diffusion layer of the membrane electrode; or embedding the current density sensor inside the bipolar plate and placing it in the cavity on the back side of the flow field of the bipolar plate; or attaching the current density sensor to the outside of the end plate of the fuel cell stack and attaching it to the current collector of the end cell.

[0111] In one embodiment of the present invention, the diagnostic module 140 is further configured to, after determining the type and location of the fault based on real-time multidimensional feature parameters, output a diagnostic report based on the type and location of the fault and / or execute preset maintenance controls corresponding to the fault type and location, wherein the diagnostic report includes fault battery number, fault type and fault location information.

[0112] In one embodiment of the present invention, the diagnostic module 140 is further configured to perform self-learning based on the diagnostic results and / or maintenance control results after outputting a diagnostic report based on the fault type and location and / or executing preset maintenance control corresponding to the fault type and location, so as to adaptively adjust and update the preset judgment conditions.

[0113] It should be noted that the specific implementation of the fuel cell fault diagnosis system 100 in this embodiment of the invention is similar to the specific implementation of the fuel cell fault diagnosis method described in the above embodiment of the invention. For details, please refer to the description in the fuel cell fault diagnosis method section. To reduce redundancy, it will not be repeated here.

[0114] The fuel cell fault diagnosis system 100 according to an embodiment of the present invention is used to implement the fuel cell fault diagnosis method of the above embodiment of the present invention. First, the current operating condition parameters of the fuel cell stack and its corresponding current density distribution map are obtained. Then, the real-time multi-dimensional feature parameters of the fuel cell stack are extracted based on the current density distribution map. Next, a preset healthy reference current density distribution map matching the current operating condition parameters is called from a preset multi-dimensional operating condition current density reference map library of the fuel cell stack in a healthy state. The preset multi-dimensional feature parameters corresponding to the preset healthy reference current density distribution map are compared with the real-time multi-dimensional feature parameters. Finally, when the comparison result meets the preset judgment conditions, it is determined that the fuel cell stack has a fault, and the type and location of the fault are determined based on the real-time multi-dimensional feature parameters. In this way, based on current density distribution scanning and multi-dimensional feature extraction, the internal state of the fuel cell stack can be visualized and monitored. By comparing with the current density reference map library in a healthy state, interference and lag problems can be avoided, and rapid and accurate fault diagnosis and fault location of the fuel cell can be achieved.

[0115] Further embodiments of the present invention also disclose an electronic device.

[0116] In some embodiments, the electronic device includes a fuel cell fault diagnosis system 100 as described in any of the above embodiments of the present invention.

[0117] In other embodiments, the electronic device includes a processor, a memory, and a fuel cell fault diagnosis program stored in the memory and executable on the processor, wherein the fuel cell fault diagnosis program, when executed by the processor, implements the fuel cell fault diagnosis method as described in any of the above embodiments of the present invention.

[0118] It should be noted that the specific implementation of the electronic device in the embodiments of the present invention is similar to the specific implementation described in the fuel cell fault diagnosis method of the above embodiments of the present invention. For details, please refer to the description in the fuel cell fault diagnosis method section. In order to reduce redundancy, it will not be repeated here.

[0119] According to an embodiment of the present invention, an electronic device is provided with a fuel cell fault diagnosis system 100 as described in the above embodiment, which executes the fuel cell fault diagnosis method described in the above embodiment. First, the current operating condition parameters of the fuel cell stack and its corresponding current density distribution map are obtained. Then, real-time multi-dimensional feature parameters of the fuel cell stack are extracted based on the current density distribution map. Next, a preset healthy reference current density distribution map matching the current operating condition parameters is retrieved from a preset multi-dimensional operating condition current density reference map library of the fuel cell stack in a healthy state. The preset multi-dimensional feature parameters corresponding to the preset healthy reference current density distribution map are compared with the real-time multi-dimensional feature parameters. Finally, when the comparison result meets the preset judgment conditions, a fault is determined to have occurred in the fuel cell stack, and the type and location of the fault are determined based on the real-time multi-dimensional feature parameters. Based on current density distribution scanning and multi-dimensional feature extraction, the internal state of the fuel cell stack can be visualized and monitored. By comparing with the current density reference map library in a healthy state, interference and lag problems can be avoided, achieving rapid and accurate fault diagnosis and fault location of the fuel cell.

[0120] A further embodiment of the present invention discloses a computer-readable storage medium storing a fuel cell fault diagnosis program, which, when executed by a processor, implements the fuel cell fault diagnosis method as described in any of the above embodiments of the present invention.

[0121] It should be noted that the specific implementation of the computer-readable storage medium in the embodiments of the present invention is similar to the specific implementation described in the fuel cell fault diagnosis method of the above embodiments of the present invention. For details, please refer to the description in the fuel cell fault diagnosis method section. To reduce redundancy, it will not be repeated here.

[0122] According to an embodiment of the present invention, when a computer-readable storage medium storing a fuel cell fault diagnosis program thereon is executed by a processor, the fuel cell fault diagnosis method of the above embodiment is executed. First, the current operating condition parameters of the fuel cell stack and its corresponding current density distribution map are obtained. Then, real-time multi-dimensional feature parameters of the fuel cell stack are extracted based on the current density distribution map. Next, a preset healthy reference current density distribution map matching the current operating condition parameters is retrieved from a preset multi-dimensional operating condition current density reference map library of the fuel cell stack in a healthy state. The preset multi-dimensional feature parameters corresponding to the preset healthy reference current density distribution map are compared with the real-time multi-dimensional feature parameters. Finally, when the comparison result meets preset judgment conditions, a fault is determined to have occurred in the fuel cell stack, and the type and location of the fault are determined based on the real-time multi-dimensional feature parameters. Thus, based on current density distribution scanning and multi-dimensional feature extraction, the internal state of the fuel cell stack can be visualized and monitored. By comparing it with the current density reference map library in a healthy state, interference and lag problems can be avoided, achieving rapid and accurate fault diagnosis and fault location of the fuel cell.

[0123] In the description of this specification, references to terms such as "one embodiment," "some embodiments," "illustrative embodiment," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example.

[0124] Although embodiments of the invention have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the claims and their equivalents.

Claims

1. A method for diagnosing fuel cell faults, characterized in that, Includes the following steps: Obtain the current operating parameters of the fuel cell stack and its corresponding current density distribution map; Based on the current density distribution map, real-time multidimensional feature parameters of the battery stack are extracted, wherein the real-time multidimensional feature parameters include at least: a global uniformity index for characterizing overall uniformity, a local low current region feature for characterizing local insufficient response, a local high current region feature for characterizing local excessive response, and a symmetry index for characterizing distribution symmetry. The preset health benchmark current density distribution map that matches the current operating condition parameters is retrieved from the preset multi-dimensional operating condition current density benchmark map library of fuel cell stacks under healthy conditions, and the preset multi-dimensional feature parameters corresponding to the preset health benchmark current density distribution map are compared with the real-time multi-dimensional feature parameters. When the comparison result meets the preset judgment conditions, it is determined that the fuel cell stack has failed, and the type and location of the failure are determined based on the real-time multidimensional feature parameters.

2. The fuel cell fault diagnosis method according to claim 1, characterized in that, Before obtaining the current operating parameters of the fuel cell stack and its corresponding current density distribution map, the process also includes: Construct and store a multi-dimensional operating condition current density benchmark library of the battery stack under healthy conditions, including: A multi-dimensional operating condition test matrix covering the typical operating range of the battery stack is constructed, wherein the test parameters of the multi-dimensional operating condition test matrix include at least: load current, battery stack temperature, anode inlet relative humidity and cathode inlet relative humidity; At each set operating point, the battery stack is controlled to operate stably, and the current density distribution data of all individual cells are collected to generate a standard healthy current density distribution map of each individual cell at the operating point. All the aforementioned standard healthy current density distribution maps are stored as multidimensional lookup tables to form the multidimensional operating condition current density benchmark map library.

3. The fuel cell fault diagnosis method according to claim 1, characterized in that, The extraction of real-time multidimensional feature parameters of the battery stack based on the current density distribution map includes: The ratio of the standard deviation to the mean of the current density distribution map is determined as the global uniformity index; The current density distribution map is segmented to identify a first target connected region with an average current density lower than a first preset threshold. The area ratio of the first target connected region and the centroid coordinates of the largest connected region are calculated. The area ratio and the centroid coordinates are determined as features of the local low current region. The current density distribution map is segmented to identify second target connected regions with average current density higher than a second preset threshold. The peak current density and maximum spatial gradient of the second target connected regions are calculated, and the peak current density and maximum spatial gradient are determined as features of the local high current region. The real-time current density distribution map is divided into two parts along the airflow direction, and the ratio of the average current density of the two parts is calculated. The ratio is then determined as the symmetry index.

4. The fuel cell fault diagnosis method according to claim 3, characterized in that, When the comparison result meets the preset judgment conditions, it is determined that the fuel cell stack has failed, and the type and location of the failure are determined based on the real-time multidimensional feature parameters, including: When the area ratio exceeds a preset area ratio threshold, and the first target connected region has a first preset shape and is located within a first preset region, it is determined that the fuel cell stack has experienced a cathode flooding fault, and the fault location information is determined based on the characteristics of the local low current region; and / or, When the peak current density exceeds a preset current density threshold and the maximum spatial gradient exceeds a preset spatial gradient threshold, it is determined that the fuel cell stack has experienced a local membrane dry fault or gas shortage fault, and the fault location information is determined based on the characteristics of the local high current region; and / or, When the global uniformity index exceeds a preset uniformity threshold, and neither the local low-current region feature nor the local high-current region feature meets the preset identification conditions, it is determined that the fuel cell stack has experienced an early performance uniformity degradation fault, and the fault location information is determined based on the global uniformity index; and / or, When the symmetry index is lower than a preset low symmetry threshold or higher than a preset high symmetry threshold, it is determined that the fuel cell stack has a reaction gas uneven distribution fault or a single-sided flow channel blockage fault, and the fault location information is determined according to the symmetry index.

5. The fuel cell fault diagnosis method according to claim 1, characterized in that, The current density distribution map is obtained based on current density sensors inside the fuel cell stack, and the current density sensors are arranged in the following ways: The current density sensor is positioned between the membrane electrode and the bipolar plate, and in contact with the diffusion layer of the membrane electrode; or, The current density sensor is embedded inside the bipolar plate and disposed within a cavity on the back side of the flow channel field of the bipolar plate; or, The current density sensor is attached to the outside of the battery stack end plate and is attached to the current collector of the end cell.

6. The fuel cell fault diagnosis method according to claim 1, characterized in that, After determining the type and location of the fault based on the real-time multidimensional feature parameters, the process further includes: Based on the type and location of the fault, a diagnostic report is output and / or preset maintenance controls corresponding to the fault type and location are executed, wherein the diagnostic report includes faulty battery number, fault type and fault location information.

7. The fuel cell fault diagnosis method according to claim 6, characterized in that, After outputting a diagnostic report based on the fault type and location and / or executing preset maintenance controls corresponding to the fault type and location, the process further includes: The system performs self-learning based on diagnostic results and / or maintenance control results to adaptively adjust and update the preset judgment conditions.

8. A fuel cell fault diagnosis system, characterized in that, include: The acquisition module is used to acquire the current operating condition parameters of the fuel cell stack and its corresponding current density distribution map. The extraction module is used to extract real-time multidimensional feature parameters of the battery stack based on the current density distribution map. The real-time multidimensional feature parameters include at least: a global uniformity index for characterizing overall uniformity, a local low current region feature for characterizing local insufficient response, a local high current region feature for characterizing local excessive response, and a symmetry index for characterizing distribution symmetry. The comparison module is used to call up a preset health benchmark current density distribution map that matches the current operating condition parameters from a preset multi-dimensional operating condition current density benchmark map library of fuel cell stacks in a healthy state, and compare the preset multi-dimensional feature parameters corresponding to the preset health benchmark current density distribution map with the real-time multi-dimensional feature parameters. The diagnostic module is used to determine that the fuel cell stack has failed when the comparison result meets the preset judgment conditions, and to determine the type and location of the failure based on the real-time multidimensional feature parameters.

9. An electronic device, characterized in that, include: The fuel cell fault diagnosis system as described in claim 8; or, A processor, a memory, and a fuel cell fault diagnosis program stored in the memory and executable on the processor, wherein the fuel cell fault diagnosis program, when executed by the processor, implements the fuel cell fault diagnosis method as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a fuel cell fault diagnosis program, which, when executed by a processor, implements the fuel cell fault diagnosis method as described in any one of claims 1-7.

Images