A fuel cell fault diagnosis method based on electric pile characteristic parameters

Abstract

The application discloses a fuel cell fault diagnosis method based on electric pile characteristic parameters and relates to the technical field of fuel cell fault diagnosis, and comprises the following steps: obtaining electric pile basic diagnosis data, dividing the electric pile according to the single cell arrangement order and forming a judgment threshold under the operation condition interval; collecting electric pile operation characteristic data, generating a partition electric pile characteristic parameter sequence according to an operation observation window, a diagnosis action window and an observation window after the action; calculating an end difference, a parameter change before and after the action, a residual rate after the action, an end residual, an insulation resistance drop rate and a current collection end contact resistance change slope; performing fault source classification judgment and generating a fault diagnosis conclusion; and updating the judgment threshold or the judgment condition according to a maintenance review result. The application improves the end area fault source classification and diagnosis record correction capability.

Description

Technical Field

[0001] This invention relates to the field of fuel cell fault diagnosis technology, and in particular to a fuel cell fault diagnosis method based on stack characteristic parameters. Background Technology

[0002] The fuel cell stack is the core energy conversion component of a fuel cell system. Its operating status is affected by factors such as gas supply, membrane electrode hydrothermal state, end current collector connection, sealing and insulation status, and cooling conditions. In vehicle power systems and stationary power generation systems, conventional fault diagnosis methods typically collect operating parameters such as single cell voltage, stack current, impedance, gas pressure, temperature, and insulation resistance. These parameters are then combined with controller alarms, operating conditions, and maintenance records to identify and classify stack anomalies, supporting operation and maintenance and safety control.

[0003] However, the existing methods still have two limitations: First, the differences between the end battery pack and the middle battery pack are not fully utilized, and abnormalities in end current collection, sealing, and gas supply and drainage are easily confused with each other under similar voltage or impedance performance; Second, there is insufficient connection between the parameter changes before and after the diagnostic action, the residual state after the action, and the maintenance review results, making it difficult to distinguish between recoverable operational abnormalities and persistent end structural abnormalities. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides a fuel cell fault diagnosis method based on stack characteristic parameters to solve the problems of inaccurate classification of fault sources in the end region and difficulty in writing back and updating the diagnosis results.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: This invention provides a fuel cell fault diagnosis method based on fuel cell stack characteristic parameters. The method includes: acquiring basic fuel cell stack diagnostic data; dividing the fuel cell stack according to the single cell arrangement order based on the basic diagnostic data; forming end-cell difference judgment thresholds, action response judgment thresholds, end-cell residue and sealing judgment thresholds, and contact resistance change slope thresholds according to operating condition intervals; collecting fuel cell stack operating characteristic data during operation; dividing the fuel cell stack operating characteristic data into time segments and validating them according to operating observation windows, diagnostic action windows, and post-action observation windows; generating a partitioned fuel cell stack characteristic parameter sequence; calculating end-cell difference calculation results based on the partitioned fuel cell stack characteristic parameter sequence; calculating parameter changes before and after diagnostic actions and post-action deviation residual rate based on parameter changes before and after diagnostic actions; and calculating the end-cell difference calculation results, insulation resistance, and current collector contact resistance within the continuous sampling window. The system calculates the end residual, insulation resistance decrease rate, and current collector contact resistance change slope. Based on the parameter change calculation results before and after the action, the deviation residual rate after the action, the end residual calculation results, insulation resistance decrease rate, and current collector contact resistance change slope, it performs fault source classification and judgment processing, classifying the fault sources into candidates for abnormal operating status, candidates for coexistence of abnormal operating status and end residual, candidates for abnormal end current collection, or candidates for end sealing or interface mismatch, and generates a fault source judgment record. Based on the fault source judgment record, it generates a fuel cell stack fault diagnosis conclusion, obtains the maintenance review results, and compares them with the fuel cell stack fault diagnosis conclusion. The comparison results are written into the confirmed fault sample set, the candidate correction sample set, or the misjudged correction sample set. Based on the confirmed fault sample set and the misjudged correction sample set, it updates the judgment threshold or fault source judgment conditions under the corresponding operating condition range, end direction, and parameter source marker.

[0007] As a preferred embodiment of the fuel cell fault diagnosis method based on stack characteristic parameters described in this invention, the step of acquiring the basic diagnostic data of the fuel cell stack and dividing the fuel cell stack according to the arrangement order of individual cells based on the basic diagnostic data includes: after acquiring the basic diagnostic data, reading the total number of individual cells, the arrangement order of individual cells, the position of the end current collector, the position of the end sealing interface, and the voltage sampling channel mapping relationship; continuously selecting individual cells from the first end of the stack to form a first end battery group, continuously selecting individual cells from the second end of the stack to form a second end battery group, continuously selecting individual cells from the middle position of the stack to form a middle battery group, and writing the individual cells between the first end battery group and the middle battery group, and between the second end battery group and the middle battery group, into the first transition battery group and the second transition battery group, respectively, to generate a partition marking record with end direction, individual cell number, and sampling channel number.

[0008] As a preferred embodiment of the fuel cell fault diagnosis method based on stack characteristic parameters described in this invention, the step of forming end difference judgment thresholds, action response judgment thresholds, end residue and sealing judgment thresholds, and contact resistance change slope thresholds according to the operating condition interval includes: when forming the judgment thresholds, determining the operating condition interval according to current density, coolant inlet temperature, hydrogen metering ratio, and air metering ratio; within the same operating condition interval, screening historical stable samples that have no communication interruptions, no range exceeding limits, no shutdown switching, no protection derating, and no incomplete maintenance records; after arranging the historical stable samples according to parameter values, combining factory calibration data, bench test data, sensor acquisition accuracy compensation values, and maintenance review allowable deviations, generating end difference judgment thresholds, action response judgment thresholds, end residue and sealing judgment thresholds, and contact resistance change slope thresholds.

[0009] As a preferred embodiment of the fuel cell fault diagnosis method based on stack characteristic parameters described in this invention, the step of collecting stack operation characteristic data during the operation of the fuel cell stack, and dividing and validating the stack operation characteristic data according to the operation observation window, diagnostic action window, and post-action observation window includes: when dividing the stack operation characteristic data by time, reading current loading change records, purging action records, drainage action records, gas metering ratio adjustment records, inlet humidity adjustment records, coolant flow rate adjustment records, and corresponding action feedback records; continuous sampling segments without diagnostic actions are marked as operation observation windows, sampling segments between the start time and end time of diagnostic actions are marked as diagnostic action windows, and sampling segments after the diagnostic action is completed and the feedback state reaches a stable condition are marked as post-action observation windows; sampling segments lacking action type, action start time, action end time, or action feedback state are written as incomplete action records, and the calculation of parameter changes before and after the action of the sampling segment is stopped.

[0010] As a preferred embodiment of the fuel cell fault diagnosis method based on stack characteristic parameters described in this invention, the step of dividing and validating the stack operation characteristic data according to the operation observation window, diagnostic action window, and post-action observation window to generate a partitioned stack characteristic parameter sequence includes: when validating the stack operation characteristic data, according to the sampling timestamp, sampling channel number, sensor range, operating condition interval number, and action record number, performing time consistency verification, range verification, sampling integrity verification, operating condition consistency verification, and action record integrity verification on voltage data, impedance data, gas pressure data, cooling data, temperature data, insulation detection data, and current collector contact resistance data; data that passes the verification is written into the partitioned stack characteristic parameter sequence according to the single cell partition mark, and data that fails the verification is written into the parameter record to be reviewed. The parameter record to be reviewed is not included in the calculation results of end difference, parameter change before and after action, end residual, insulation resistance decrease rate, and current collector contact resistance change slope.

[0011] As a preferred embodiment of the fuel cell fault diagnosis method based on stack characteristic parameters described in this invention, the following steps are included: calculating end-point difference calculation results based on the partitioned stack characteristic parameter sequence; calculating parameter change calculation results before and after the diagnostic action and the deviation residual rate after the action based on parameter changes before and after the action; calculating end-point residual calculation results, insulation resistance decrease rate, and current collector contact resistance change slope based on end-point difference calculation results, insulation resistance decrease rate, and current collector contact resistance change slope: when calculating end-point difference calculation results, comparing the first end battery pack with the middle battery pack to generate the first end-point difference calculation result; comparing the second end battery pack with the middle battery pack to generate the second end-point difference calculation result; sequentially comparing the differences of the first end battery pack, the first transition battery pack, and the middle battery pack to generate the first transition difference verification result; sequentially comparing the differences of the second end battery pack, the second transition battery pack, and the middle battery pack to generate the second transition difference verification result; and marking the end-point difference calculation results as end-point local difference or end-point extended difference based on the first transition difference verification result and the second transition difference verification result.

[0012] As a preferred embodiment of the fuel cell fault diagnosis method based on stack characteristic parameters described in this invention, the following steps are included: calculating the end-point difference calculation results according to the partitioned stack characteristic parameter sequence; calculating the parameter change calculation results before and after the diagnostic action and the deviation residual rate after the action based on the parameter changes before and after the action; and calculating the end-point residual calculation results, insulation resistance decrease rate, and current collector contact resistance change slope based on the end-point difference calculation results, insulation resistance decrease rate, and current collector contact resistance change slope. This includes: when calculating the parameter change calculation results before and after the action, recording the parameter changes corresponding to the purging and draining actions as purging and draining action response values, and recording the gas metering... The parameter changes corresponding to the adjustment are recorded as the gas adjustment action response value, and the parameter changes corresponding to the inlet humidity adjustment are recorded as the humidity adjustment action response value. The parameters before the action are taken from the nearest valid running observation window before the start time of the diagnostic action, and the parameters after the action are taken from the first valid post-action observation window after the end of the diagnostic action. The post-action deviation residual rate is generated based on the ratio of the voltage deviation rate at the end of the action to the voltage deviation rate at the end of the action. When there are more than two action types in the same diagnostic action window and the control record cannot be split into action time periods, the diagnostic action window is marked as a composite action window, and the calculation of single action response value is stopped.

[0013] As a preferred embodiment of the fuel cell fault diagnosis method based on stack characteristic parameters described in this invention, the following steps are included: calculating the end-point difference calculation results according to the partitioned stack characteristic parameter sequence; calculating the parameter change calculation results before and after the diagnostic action and the deviation residual rate after the action according to the parameter change before and after the action; and calculating the end-point residual calculation results, insulation resistance decrease rate, and current collector contact resistance change slope according to the end-point difference calculation results, insulation resistance decrease rate, and current collector contact resistance change slope within the continuous sampling window. Specifically, when calculating the end-point residual calculation results, insulation resistance decrease rate, and current collector contact resistance change slope, the deviation residual rate after the action, the current collector contact resistance deviation rate, and the end-point temperature deviation rate are written into the same end-point residual calculation record. The insulation resistance within the current sampling window is compared with the insulation resistance reference value within the same operating condition range to generate the insulation resistance decrease rate. The current collector contact resistances with consistent parameter source markings within the continuous sampling window are arranged in the order of sampling time, and the current collector contact resistance change slope is generated based on the change in current collector contact resistance between the first and last sampling windows and the time interval.

[0014] As a preferred embodiment of the fuel cell fault diagnosis method based on stack characteristic parameters described in this invention, the step of classifying fault sources into candidates for abnormal operating conditions, candidates for coexistence of abnormal operating conditions and end residues, candidates for abnormal end current collection, or candidates for end sealing or interface mismatch, and generating a fault source judgment record includes: when performing fault source classification and judgment processing, determining the current action response value according to the diagnostic action type; when the diagnostic action type is a purging action or a drainage action, comparing the purging / drainage action response value with the purging / drainage action response threshold; when the diagnostic action type is a gas metering ratio adjustment action, comparing the gas adjustment action response value with the gas adjustment action response threshold; when the diagnostic action type is an inlet humidity adjustment action, comparing the humidity adjustment action response value with the humidity adjustment action response threshold; the current action response value reaches the action response threshold corresponding to the diagnostic action type. When the post-action deviation residual rate does not reach the post-action deviation residual rate threshold, an abnormal operation status candidate is generated; when the current action response value reaches the action response threshold corresponding to the diagnostic action type, and the post-action deviation residual rate reaches the post-action deviation residual rate threshold, a candidate for coexistence of abnormal operation status and end residue is generated; when the purging and drainage action response value does not reach the purging and drainage action response threshold, the gas adjustment action response value does not reach the gas adjustment action response threshold, the humidity adjustment action response value does not reach the humidity adjustment action response threshold, and the post-action deviation residual rate reaches the post-action deviation residual rate threshold, the end residue calculation result is compared with the end residue judgment threshold, the insulation resistance decrease rate is compared with the insulation resistance decrease rate threshold, and the current collector contact resistance change slope is compared with the contact resistance change slope threshold. Based on the comparison results, an end current collector abnormality candidate or an end sealing or interface matching candidate is generated.

[0015] As a preferred embodiment of the fuel cell fault diagnosis method based on stack characteristic parameters described in this invention, the steps of generating a fuel cell stack fault diagnosis conclusion based on the fault source judgment record, obtaining maintenance review results and comparing them with the fuel cell stack fault diagnosis conclusion, writing the comparison results into a confirmed fault sample set, a candidate correction sample set, or a misjudged correction sample set, and updating the judgment threshold or fault source judgment condition under the corresponding operating condition range, end direction, and parameter source marker according to the confirmed fault sample set and the misjudged correction sample set include: after obtaining the maintenance review results, the actual fault type, actual fault location, and end direction in the maintenance review results are... The parameters retested after maintenance are compared with the fault diagnosis conclusions of the fuel cell stack. When the comparison results show that the diagnoses are consistent, the sampling window, end parameter calculation records, fault source judgment records, and maintenance verification results are written into the confirmed fault sample set. When the comparison results show that they are partially consistent, the corresponding records are written into the candidate correction sample set. When the comparison results show that the erroneous judgment is corrected, the corresponding records are written into the erroneous judgment correction sample set. When updating the judgment threshold or fault source judgment conditions, the confirmed fault sample set and the erroneous judgment correction sample set are screened according to the operating condition range, end direction, and parameter source marker. Samples with inconsistent matching conditions are not merged to generate the same threshold version or the same fault source judgment condition.

[0016] The beneficial effects of this invention are as follows: by calculating the end difference calculation results, the parameter change calculation results before and after the action, and the deviation residual rate after the action, the end operation changes and the end residual state are jointly included in the fault source classification and judgment process, forming candidates for abnormal operation status, candidates for abnormal end current collection, or candidates for end sealing or interface matching; by writing the maintenance review results into the corresponding sample set and updating the judgment threshold or fault source judgment conditions, the fault diagnosis conclusion of the fuel cell stack is made into a write-back record, improving the accuracy of fault source classification in the end area and the ability to correct diagnostic records. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart of a fuel cell fault diagnosis method based on stack characteristic parameters.

[0019] Figure 2 A flowchart is created to establish the basic records and effective sampling windows for fuel cell stack zoning diagnosis.

[0020] Figure 3This is a flowchart for calculating end differences, action response, and end residue.

[0021] Figure 4 Flowchart for classifying, judging, and updating fault sources and diagnostic records. Detailed Implementation

[0022] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0023] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0024] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0025] Reference Figures 1-4 This is one embodiment of the present invention, which provides a fuel cell fault diagnosis method based on fuel cell stack characteristic parameters, including the following steps: S1. Obtain the basic diagnostic data of the fuel cell stack. Based on the basic diagnostic data, divide the fuel cell stack according to the arrangement order of the single cells. Form the end difference judgment threshold, action response judgment threshold, end residue and sealing judgment threshold and contact resistance change slope threshold according to the operating condition range.

[0026] The basic diagnostic data for the fuel cell stack consists of stack structure data, assembly and connection data, operational baseline data, historical operational data, and historical maintenance data. The stack structure data records the stack number, total number of individual cells, cell arrangement order, effective reaction area of ​​each cell, membrane electrode type, bipolar plate flow channel type, end current collector material type, end conductive layer type, gas interface location, coolant interface location, drain channel location, end sealing interface location, and insulation detection terminal location. The end conductive layer type is recorded as a metal plating, carbon-based conductive layer, metal and carbon-based composite conductive layer, or other conductive layer material; for other conductive layer materials, the material name, thickness range, and bonding method are recorded simultaneously. The assembly and connection data records the end clamping force range, clamping component placement, current collector connection location, terminal block location, voltage sampling point location, end current sampling point location, four-wire measurement terminal configuration, temperature detection location, insulation detection location, end sealing component location, and water vapor interface protection structure location.

[0027] The operational baseline data records rated current density, allowable current density range, allowable loading rate, hydrogen metering ratio range, air metering ratio range, coolant inlet temperature range, coolant outlet temperature range, coolant flow rate range, inlet humidity range, purging cycle, and drain valve opening cycle. Historical operational data records historical operating voltage, historical operating impedance, historical gas differential pressure, historical end temperature, historical insulation resistance, and historical current collector contact resistance. Historical maintenance data records historical fault types, fault occurrence areas, actual maintenance locations, maintenance actions, parameter changes before and after maintenance, post-maintenance retest window number, and maintenance conclusion status. Maintenance conclusion status includes confirmed fault, misjudgment correction, pending review, and invalid record; maintenance records in the confirmed fault and misjudgment correction status participate in threshold correction, while records pending review and invalid records do not participate in threshold formation. After data acquisition, a basic record for fuel cell stack zoning diagnosis is generated and bound to the fuel cell stack number, vehicle number, or fixed application equipment number for storage.

[0028] Based on the arrangement of individual cells, the fuel cell stack is divided into a first end cell group, a first transition cell group, a middle cell group, a second transition cell group, and a second end cell group. The first end cell group is selected continuously from the first end of the stack, the second end cell group is selected continuously from the second end of the stack, and the middle cell group is selected continuously from the middle of the stack. The first transition cell group is located between the first end cell group and the middle cell group, and the second transition cell group is located between the second end cell group and the middle cell group. The first and second transition cell groups are not output as separate fault sources, but rather serve as verification partitions for the end-cell difference calculation results, participating in the judgment of local end-cell differences and extended end-cell differences.

[0029] The number of individual cells contained in each end battery pack is expressed as follows: ; in, This indicates the number of individual cells contained in each end battery pack; This indicates the total number of individual cells in a fuel cell stack; This indicates the proportion of battery cells taken from the end battery pack, and the value is no greater than 0.1; This indicates that the product of the proportion of cells taken from the end battery pack and the total number of individual cells is rounded up. This indicates that the calculation result is rounded down to 10% of the total number of single cells; Indicates rounding up. Indicates rounding down; This indicates taking the maximum value. This indicates that the minimum value is taken; when the total number of single batteries is less than 20, 2 batteries are taken from each end battery pack.

[0030] The proportion of cells taken from the end-cell stacks is determined by the stack structure design documents, the coverage area of ​​the end current collectors, the impact range of end compression, and historical maintenance review records of the same model of stack. Fuel cell stacks of the same model, with the same single-cell arrangement and the same end current collector connection structure, use the same proportion of cells taken from the end-cell stacks. If corresponding records are missing, the proportion of cells taken from the end-cell stacks is determined based on the range of continuous single cells where the end voltage deviation rate, end impedance deviation rate, and end temperature deviation rate show a stable difference relative to the central stack during bench testing, and this proportion is recorded in the stack zoning diagnostic baseline record for the state to be verified. When the total number of single cells is less than 20, 2 cells are taken from each end-cell stack; when the total number of single cells is greater than or equal to 20, the number of cells taken from each end-cell stack is not less than 2 and does not exceed 10% of the total number of single cells.

[0031] The number of individual cells in the middle battery pack shall not be less than the number of individual cells in any end battery pack. If the total number of individual cells is insufficient to simultaneously form end battery packs, middle battery packs, and transition battery packs, the end battery packs and middle battery packs shall be retained first, and the transition battery packs shall be marked as vacant. A partition label shall be established for each individual cell, recording the individual cell number, sequence number, partition to which it belongs, adjacent individual cell numbers, corresponding end direction, interface proximity, drainage channel proximity, and current collector connection proximity. When there is a discrepancy between the individual cell voltage sampling channel number and the actual sequence number, a channel mapping table shall be established according to the assembly document; individual cell voltage data for which channel mapping has not been completed shall not be included in the end voltage deviation rate calculation.

[0032] Thresholds are formed separately for each operating condition interval. The operating condition interval is determined jointly by current density, coolant inlet temperature, hydrogen metering ratio, and air metering ratio. When the current density, coolant inlet temperature, hydrogen metering ratio, and air metering ratio within a sampling window all fall within the same operating condition interval, the sampling window participates in the threshold formation for that operating condition interval. When a sampling window spans multiple operating condition intervals, the operating condition interval with the longest duration is marked, but it is not included as a historical stable sample in threshold formation.

[0033] The thresholds for judging end differences include the end voltage deviation rate threshold, high-frequency impedance deviation rate threshold, low-frequency impedance deviation rate threshold, gas pressure difference deviation rate threshold, end temperature deviation rate threshold, and collector contact resistance deviation rate threshold. The thresholds for judging action response include the purging / drainage action response threshold, gas adjustment action response threshold, and humidity adjustment action response threshold. The thresholds for judging end residue and sealing include the post-action deviation residue rate threshold, the end residue determination threshold, and the insulation resistance decrease rate threshold. The contact resistance change slope threshold is used to determine whether the change in collector contact resistance within a continuous sampling window meets the conditions for the formation of an end collector anomaly candidate.

[0034] Before threshold formation, a historical stable sample set is established. This set selects only sampling windows within the same operating condition range where parameter acquisition is complete, communication is uninterrupted, sensor over-range is not observed, there is no shutdown switching, protection load reduction, incomplete maintenance records, and no unclear fault confirmation status. When the number of historical stable samples within a single operating condition range is no less than 30, the samples are arranged according to parameter values, and the grading boundaries are determined based on factory calibration documents, bench test documents, or historical operating documents of the same model of fuel cell stack. After the grading boundaries are determined, sensor acquisition accuracy compensation values ​​and maintenance verification allowable deviations are superimposed to form the corresponding parameter thresholds. When the number of historical stable samples is less than 30, factory calibration data and bench test data are used to form the initial threshold, and the threshold status is marked as the initial threshold status.

[0035] The action response thresholds are generated in two stages: initial formation and sample updating. Before confirmed fault samples are formed, factory calibration documents, bench test documents, and historical maintenance records of the same model fuel cell stack are read. Changes in end voltage deviation rate, end impedance deviation rate, and gas differential pressure deviation rate before and after the diagnostic action are extracted and arranged according to the calculated results of the corresponding action response values ​​to generate initial action response thresholds, which are then marked as pending verification. After confirmed fault samples are formed, the purging and drainage action response thresholds are updated based on confirmed flooding samples, confirmed drainage anomaly samples, and corresponding misjudgment correction samples; the gas adjustment action response thresholds are updated based on confirmed gas supply anomaly samples and corresponding misjudgment correction samples; and the humidity adjustment action response thresholds are updated based on confirmed membrane moisture anomaly samples and corresponding misjudgment correction samples. When there is no overlap between the action response values ​​of confirmed fault samples and misjudgment correction samples, the median value of the adjacent boundaries of the two types of samples is selected as the action response threshold; if there is an overlap, the overlap is marked as a verification interval.

[0036] The contact resistance change slope threshold is determined based on normal aging samples and confirmed manifold corrosion samples. Normal aging samples are derived from historical operating records where no end manifold maintenance or end seal maintenance was performed, and post-maintenance retests did not confirm end manifold failure. Confirmed manifold corrosion samples are derived from records confirming corrosion, conductive layer failure, or contact interface deterioration after disassembly, maintenance, or replacement of the end manifold. When the contact resistance change slopes of the two types of samples do not overlap, the median value between the two boundaries is selected as the contact resistance change slope threshold; when overlapping intervals exist, the overlapping interval is marked as the verification interval.

[0037] The threshold record includes the threshold name, applicable parameters, operating condition interval number, sample source, sample quantity, sample selection rules, quantile boundaries, sensor acquisition accuracy compensation value, allowable deviation for maintenance and review, threshold status, threshold version number, generation time, and effective status. Threshold status includes initial threshold status and maintenance sample correction threshold status; effective status includes pending verification, effective, deactivated, and replaced. Thresholds in the pending verification status are used to assist in review and do not directly generate fault source conclusions; thresholds in the effective status participate in the current fault source judgment; thresholds in the deactivated or replaced status are retained in the historical record and do not participate in the current diagnostic judgment.

[0038] When pressure sensors, temperature sensors, voltage acquisition modules, insulation detection modules, or contact resistance measurement terminals are replaced, the corresponding parameter thresholds re-enter the verification state. The threshold to be verified is verified after collecting no fewer than 30 new stable samples within the same operating condition range and obtaining no fewer than 5 maintenance review samples. If the fault identification result of the threshold to be verified is not lower than the replaced threshold, the number of misjudgment correction triggers is not higher than the replaced threshold, and no similar misjudgments occur in at least 5 consecutive maintenance review results, then the threshold to be verified is updated to the effective state. After completing the establishment of the basic records for fuel cell zoning diagnosis and threshold formation rules, a basic record version number is generated. The basic record version number is bound to the fuel cell number, operating condition range number, zoning mark, channel mapping table, threshold version number, and maintenance sample version number; when any zoning rule, channel mapping relationship, threshold version, or maintenance sample set changes, a new basic record version number is generated.

[0039] S2. Collect fuel cell stack operation characteristic data during the operation of the fuel cell stack, divide the fuel cell stack operation characteristic data into time segments and verify their validity according to the operation observation window, diagnostic action window and post-action observation window, and generate a partitioned fuel cell stack characteristic parameter sequence.

[0040] After the basic record for fuel cell stack regional diagnosis is established, the basic record version number, operating condition interval number, single cell partition mark, channel mapping table, threshold version number, threshold status, and parameter source mark are read, and fuel cell stack operating characteristic data are collected according to the sampling time. The fuel cell stack operating characteristic data consists of voltage data, impedance data, gas pressure data, cooling data, temperature data, insulation detection data, current collector contact resistance data, drainage status data, and diagnostic action data. Specifically, voltage data records single cell voltage, total stack voltage, and stack output current; impedance data records high-frequency impedance and low-frequency impedance; gas pressure data records hydrogen inlet pressure, hydrogen outlet pressure, air inlet pressure, and air outlet pressure; cooling data records coolant inlet temperature, coolant outlet temperature, and coolant flow feedback value; temperature data records the temperature of the first end region, the second end region, and the middle region; and diagnostic action data records the action type, start time, end time, and feedback status of purging actions, drainage actions, gas metering ratio adjustment actions, inlet humidity adjustment actions, and coolant flow adjustment actions. Each data item is associated with the sampling time, sampling channel number, sensor number, parameter unit, sampled value, sampling status, and data source.

[0041] The single-cell voltage is matched with the single-cell number according to the channel mapping table. When the single-cell voltage sampling channel number matches the actual sequence number of the single cell, it is written to the corresponding partition according to the single-cell arrangement order; when they do not match, it is converted according to the channel mapping table and written to the corresponding partition; the single-cell voltage data for which channel mapping has not been completed is written to the parameter record to be reviewed and is not included in the calculation of end voltage deviation rate and single-cell voltage dispersion. The single-cell voltage data is stored separately for the first end battery group, the first transition battery group, the middle battery group, the second transition battery group, and the second end battery group, and is bound to the stack output current at the same sampling time.

[0042] High-frequency impedance characterization values ​​are preferably obtained through online impedance detection data from the fuel cell stack control system. When online impedance detection data is unavailable, the zoned equivalent ohmic impedance is calculated based on the current change and the corresponding voltage change, provided that the current change record is complete, the voltage sampling before and after the current change is complete, and no purging or draining actions occur within the sampling window.

[0043] The partitioned equivalent ohmic impedance is expressed as: ; in, This represents the equivalent ohmic impedance in the high-frequency impedance characterization value of the partition; This represents the average single-cell voltage change within the same sampling window for a target zone, which includes the first end battery pack, the second end battery pack, or the middle battery pack. This represents the change in the output current of the fuel cell within the same sampling window; This represents a positive number to prevent the denominator from being zero, and its value is determined based on the resolution of the current acquisition.

[0044] Low-frequency impedance characterization values ​​are primarily obtained through online low-frequency impedance detection data from the fuel cell stack control system. When online low-frequency impedance detection data is unavailable, the average single-cell voltage recovery and average stack output current during the current holding phase are read to calculate the equivalent low-frequency impedance characterization value; the equivalent low-frequency impedance characterization value is the ratio of the average single-cell voltage recovery to the average stack output current. The start time of the current holding phase is the first sampling moment after the current control command is issued, when the stack output current enters the target current allowable range; the end time is the last valid sampling moment before the next current control command, purging action, drainage action, gas metering ratio adjustment, inlet humidity adjustment, or coolant flow rate adjustment occurs. If the current holding phase is missing, the holding time is less than the minimum holding time, the target current allowable range is missing, the average single-cell voltage recovery cannot be obtained, or the valid current sampling value is insufficient, the sampling window is marked as having no low-frequency impedance data and is not included in the calculation of low-frequency impedance deviation rate and purging / draining action response value. High-frequency impedance characterization values ​​and low-frequency impedance characterization values ​​are saved separately, and the parameter source markers are recorded separately for each.

[0045] The method for obtaining the collector contact resistance is determined based on the collector measurement configuration. When a four-wire measurement terminal is provided at the collector, the contact resistance measurement value is read. When no four-wire measurement terminal is provided, the voltage difference between the collector connection point and the terminal voltage reference point, as well as the terminal current sampling value, are read to calculate the equivalent contact resistance. If the voltage sampling line includes the voltage drop of external conductors, the external conductor voltage drop is subtracted from the conductor resistance and terminal current sampling value in the factory calibration file. If the terminal current sampling value is less than the minimum calculated current value, the equivalent contact resistance is not calculated, and an insufficient terminal current record is generated.

[0046] The equivalent contact resistance is expressed as: ; in, Indicates the contact resistance of the current collector; This represents the voltage difference between the collector connection point and the end voltage reference point within the same sampling window; This indicates the end current sampling value within the same sampling window. The hydrogen-side pressure difference is calculated based on the hydrogen inlet and outlet pressures, while the air-side pressure difference is calculated based on the air inlet and outlet pressures. Before pressure data is included in the calculation, it is determined whether the inlet pressure sampling time, outlet pressure sampling time, and stack output current sampling time belong to the same sampling period. Pressure data exceeding the time synchronization allowable value is recorded as a timestamp inconsistency record and is not included in the gas pressure difference calculation. When the outlet pressure is greater than the inlet pressure and the difference exceeds the pressure sensor's accuracy compensation range, a pressure sampling verification record is generated and is not included in the gas pressure difference deviation rate calculation.

[0047] The hydrogen-side pressure difference, air-side pressure difference, and coolant temperature difference are expressed as follows: ; ; ; in, This indicates the pressure difference on the hydrogen side; Indicates the hydrogen inlet pressure; Indicates the hydrogen outlet pressure; Indicates the air pressure difference; Indicates the air inlet pressure; Indicates the air outlet pressure; Indicates the coolant temperature difference; Indicates the coolant outlet temperature; This represents the coolant inlet temperature. The above calculation results are respectively bound and saved with the fuel cell output current, operating condition interval number, and sampling time within the same sampling window.

[0048] The temperatures of the first end region, the second end region, and the middle region are written to the corresponding partitions according to the temperature detection location. When no temperature sensor is configured in the middle region, the coolant temperature near the middle battery pack or the adjacent casing temperature is read as a substitute value for the middle region temperature, and the substitute source marker is recorded; if a substitute value cannot be obtained, the end temperature deviation rate is not calculated for this sampling window. Insulation resistance is read by the insulation detection module and is bound and saved with the insulation detection location, detection time, and detection status. The drainage status records the drain valve opening time, drain valve closing time, drainage duration, drainage action number, and drainage action completion status.

[0049] Sampling windows are divided based on the fuel cell stack operation control records. These windows include an operational observation window, a diagnostic action window, and a post-action observation window. The operational observation window is for samples where the stack output current, hydrogen metering ratio, air metering ratio, coolant inlet temperature, and inlet humidity are all within the same operating range, and where no purging, drainage, gas metering ratio adjustment, inlet humidity adjustment, coolant flow rate adjustment, protection load reduction, or shutdown switching has occurred within this window. The diagnostic action window is for samples where purging, drainage, gas metering ratio adjustment, inlet humidity adjustment, or coolant flow rate adjustment has occurred. The post-action observation window records changes in single-cell voltage, impedance, gas pressure difference, end temperature, insulation resistance, and current collector contact resistance after the diagnostic action has been completed.

[0050] The boundaries of the sampling window are determined based on the control records. The target range is the feedback judgment range formed by the target value issued by the controller and the corresponding allowable deviation. The target value and allowable deviation are derived from the fuel cell stack operating reference data, controller control strategy documents, factory calibration documents, or bench test documents. If the control records already contain the target upper limit and target lower limit, these two are directly used as the target range. For current load changes, the window boundary is the moment when the change in fuel cell stack output current reaches the current change judgment value. For purging actions, the start time of the diagnostic action window is the moment the purging valve opens, and the end time is the moment the purging valve closes. For drainage actions, the start time of the diagnostic action window is the moment the drain valve opens, and the end time is the moment the drain valve closes. For gas metering ratio adjustment, inlet humidity adjustment, and coolant flow rate adjustment, the start time of the diagnostic action window is the moment the corresponding control command is issued, and the end time is the moment the corresponding feedback value enters the target range. Sampling windows lacking action start time, action end time, action type records, or target range records only generate operating parameter records and do not participate in the calculation of parameter changes before and after the action.

[0051] After the diagnostic action is completed, the duration of the observation window after the action is determined according to the type of control action. The duration of the observation window after a purging action or gas metering ratio adjustment is not less than the gas flow feedback stabilization time; the duration of the observation window after a drainage action is not less than the drainage status confirmation time after the drain valve is closed; the duration of the observation window after an inlet humidity adjustment is not less than the humidity feedback stabilization time; and the duration of the observation window after a coolant flow adjustment is not less than the coolant temperature feedback stabilization time. The above durations are derived from the fuel cell stack control calibration document or bench test document.

[0052] The validity of the acquired parameters is verified. Validity verification includes time consistency verification, range verification, sampling integrity verification, operating condition consistency verification, and action record integrity verification. Time consistency verification determines whether the single-cell voltage, stack output current, impedance, pressure, temperature, insulation resistance, and current collector contact resistance are within the same sampling period; range verification determines whether the sampled values ​​are within the calibrated range of the corresponding sensor; sampling integrity verification determines whether there are missing channels, duplicate frozen values, or communication interruptions within the same sampling window; operating condition consistency verification determines whether the current density, coolant inlet temperature, hydrogen metering ratio, and air metering ratio within the same sampling window are within the same operating condition range; action record integrity verification determines whether the diagnostic action window has the action type, action start time, action end time, and action feedback status.

[0053] When a collected parameter fails validity verification, a parameter record to be reviewed is generated. This record includes the window number, parameter name, sampling channel number, sensor number, anomaly type, anomaly occurrence time, and processing flag. Anomaly types include communication interruption, sampling freeze, inconsistent timestamps, sampled values ​​exceeding range, missing sampling channels, missing action records, operating conditions crossing ranges, insufficient end current, reverse pressure sampling, and missing mid-section temperature. Parameter records to be reviewed are not included in the corresponding formula calculations but are retained in the fuel cell stack zonal diagnostic basic record for maintenance review and threshold sample screening.

[0054] After validity verification, a sequence of characteristic parameters for each zoned battery pack is generated. This sequence is saved according to window number and zone label, including the parameter sequences for the first end battery pack, the first transition battery pack, the middle battery pack, the second transition battery pack, and the second end battery pack. Each zoned battery pack parameter sequence records the single-cell voltage sequence, impedance characterization value, end temperature or middle region temperature, gas pressure difference, insulation resistance, current collector contact resistance, drainage status, diagnostic action type, action start time, action end time, action feedback status, parameter validity status, and parameter source label. The zoned battery pack characteristic parameter sequence corresponding to the first transition battery pack is bound to the first end direction, and the zoned battery pack characteristic parameter sequence corresponding to the second transition battery pack is bound to the second end direction. When a transition battery pack lacks valid sampling data, a transition difference verification record is generated, and the transition difference verification status is marked in the end parameter calculation record.

[0055] When the single-cell voltage data of the first or second end battery pack is valid, but the single-cell voltage data of the middle battery pack is missing, the sampling window does not calculate the end voltage deviation rate; when the middle battery pack is valid and the single-sided end battery pack is valid, the end difference calculation result is only calculated for the valid end battery pack; when both end battery packs are valid, the first end difference calculation result and the second end difference calculation result are calculated respectively.

[0056] S3. Calculate the end difference calculation results based on the characteristic parameter sequence of the partitioned stack. Calculate the parameter change calculation results before and after the diagnostic action and the deviation residual rate after the action based on the parameter change before and after the action. Calculate the end residual calculation results, insulation resistance decrease rate and current collector contact resistance change slope based on the end difference calculation results, insulation resistance decrease rate and current collector contact resistance change slope within the continuous sampling window.

[0057] After a valid sampling window is formed, the sequence of characteristic parameters of the partitioned fuel cell stack, the operating condition interval number, the threshold version number, the threshold status, and the parameter source marker are read. If the parameter validity status of the sampling window is valid and the threshold status used for comparison is active, the calculations for end-point differences, parameter changes before and after the action, and end-point residues are performed. If the threshold status is pending verification, the calculation results only generate a verification record and do not directly enter the fault source judgment; if the threshold status is disabled or replaced, that threshold is not used in the calculation. Sampling windows with missing sampling channels, inconsistent timestamps, out-of-range measurements, operating conditions spanning different intervals, or missing parameter sources will not enter the corresponding formula calculations and will generate a parameter record pending verification.

[0058] The end-point difference calculation results include end voltage deviation rate, high-frequency impedance deviation rate, low-frequency impedance deviation rate, end temperature deviation rate, collector contact resistance deviation rate, and gas pressure difference deviation rate. The parameter change calculation results before and after the action include purge / drainage action response values, gas adjustment action response values, and humidity adjustment action response values. The end-point residue calculation results are obtained based on the post-action deviation residue rate, collector contact resistance deviation rate, and end temperature deviation rate. The insulation resistance decrease rate is calculated based on the insulation resistance within the current sampling window and the baseline insulation resistance value within the same operating condition range. The collector contact resistance change slope is calculated based on the collector contact resistance within the continuous sampling window and the operating time interval.

[0059] The end-difference calculation results are calculated separately for the first end battery pack and the middle battery pack, and for the second end battery pack and the middle battery pack. The first end battery pack generates the first end-difference calculation result, and the second end battery pack generates the second end-difference calculation result. When both end battery packs are valid, two sets of calculation results are saved respectively; when only one end battery pack is valid, only the end-difference calculation result for that side is saved; when the middle battery pack lacks valid reference data, the end-difference calculation result is not calculated. After the end-difference calculation results are generated, the transition difference verification results are calculated. The first end battery pack is compared with the first transition battery pack, and the first transition battery pack is compared with the middle battery pack to generate the first transition difference verification result; the second end battery pack is compared with the second transition battery pack, and the second transition battery pack is compared with the middle battery pack to generate the second transition difference verification result. When the end-difference calculation result reaches the corresponding judgment threshold and the difference between the same-side transition battery pack and the middle battery pack does not reach the corresponding judgment threshold, an end local difference marker is generated; when the end-difference calculation result reaches the corresponding judgment threshold and the difference between the same-side transition battery pack and the middle battery pack reaches the corresponding judgment threshold, an end extended difference marker is generated.

[0060] The terminal voltage deviation rate is expressed as: ; in, Indicates the deviation rate of the terminal voltage; This represents the average single-cell voltage of either the first-end battery pack or the second-end battery pack within the same sampling window. This represents the average single-cell voltage of the middle battery pack within the same sampling window. When calculating the average single-cell voltage, only single-cell voltage data with valid parameter states are used. If the number of valid single cells in the middle battery pack is less than the number of valid single cells in the corresponding end battery pack, or if the average single-cell voltage of the middle battery pack is lower than the lower limit of voltage calculation, the end voltage deviation rate is not calculated, and a middle voltage insufficient reference record is generated.

[0061] The lower limit of voltage calculation is determined based on the effective range, sampling resolution, protection control lower limit of the single cell voltage sampling module, and the lowest effective voltage in the stable operation samples of the same type of battery stack. It is used to determine whether the average single cell voltage of the middle battery pack meets the conditions for participating in the calculation of the end voltage deviation rate.

[0062] The end impedance deviation rate is expressed as: ; in, Indicates the deviation rate of the end impedance; This represents the impedance characterization value corresponding to the first or second end battery pack. This represents the impedance characteristic value corresponding to the middle battery pack. When... and When all values ​​are taken as high-frequency impedance characterization values, the calculation result is recorded as the high-frequency impedance deviation rate; when and When all impedance values ​​are low-frequency, the calculation result is recorded as the low-frequency impedance deviation rate. High-frequency and low-frequency impedance deviation rates are stored separately. If an impedance value is missing, the impedance source markings are inconsistent, or the mid-range impedance value is below the lower limit for impedance calculation, the corresponding impedance deviation rate is not calculated.

[0063] The lower limit of impedance calculation is determined based on the effective range of the impedance measurement module, the impedance sampling resolution, the impedance calculation noise boundary, and the lowest effective impedance of stable samples within the same operating condition range. It is used to determine whether the impedance of the middle battery pack meets the conditions for participating in the calculation of the end impedance deviation rate.

[0064] Before calculating the end temperature deviation rate, the temperatures of the first end region, the second end region, and the middle region are all converted to Kelvin. The end temperature deviation rate is expressed as: ; in: Indicates the deviation rate of the end temperature; This represents the Kelvin temperature characterization value of the first or second end region within the same sampling window. This represents the Kelvin temperature characterization value of the central region within the same sampling window. The central region temperature is obtained from the central temperature sensor; if the central temperature sensor is missing, a substitute value for the central region temperature is read according to the parameter source marking and included in the calculation after being converted to Kelvin temperature. If both the central region temperature and the substitute value are missing, the end temperature deviation rate is not calculated.

[0065] The deviation rate of the collector contact resistance is expressed as: ; in: Indicates the deviation rate of the contact resistance at the current collector; This indicates the collector contact resistance within the current sampling window; This represents the reference value of the collector contact resistance within the same operating condition range. Before calculating the collector contact resistance deviation rate, the reference value of the collector contact resistance is compared with the lower limit of the contact resistance reference calculation. When the reference value of the collector contact resistance is lower than the lower limit of the contact resistance reference calculation, the collector contact resistance deviation rate is not calculated, and a record of insufficient contact resistance reference is generated.

[0066] The gas pressure difference deviation rate is expressed as: ; in: Indicates the deviation rate of gas pressure difference; This indicates the pressure difference on the hydrogen side or the air side within the current sampling window; This represents the reference value for hydrogen-side or air-side differential pressure within the same operating condition range. Before calculating the gas differential pressure deviation rate, the reference value for gas differential pressure is compared with the lower limit of the gas differential pressure reference calculation. When the reference value for gas differential pressure is lower than the lower limit of the gas differential pressure reference calculation, the gas differential pressure deviation rate is not calculated, and a record of insufficient gas differential pressure reference is generated.

[0067] The lower limits for calculation of contact resistance and gas pressure difference are determined by the fuel cell stack operation reference data, sensor acquisition accuracy, factory calibration documents, bench test documents, and stable operation samples of the same model fuel cell stack, and are stored in conjunction with the threshold version number.

[0068] The gas pressure difference deviation rate is expressed as: ; in, Indicates the deviation rate of gas pressure difference; This indicates the pressure difference on the hydrogen side or the air side within the current sampling window; This represents the reference value for the corresponding gas pressure difference within the same operating condition range. The pressure difference on the hydrogen side and the air side are calculated and stored separately. The corresponding gas pressure difference deviation rate is not calculated when pressure sampling is reversed, pressure sampling times are inconsistent, pressure data is missing, or the reference value for gas pressure difference is missing.

[0069] The calculation results of parameter changes before and after an action are formed based on the operation observation window, the diagnostic action window, and the post-action observation window. The diagnostic action window must include the action type, action start time, action end time, and action feedback status. When multiple operation observation windows exist before the action starts, the operation observation window closest to the action start time and with valid parameters is selected as the pre-action parameter source; when multiple post-action observation windows exist after the action ends, the first post-action observation window with valid parameters after the action ends is selected as the post-action parameter source. The pre-action parameter source and the post-action parameter source must belong to the same operating condition interval; when crossing operating condition intervals, an action window cross-interval record is generated, and the calculation results of parameter changes before and after the action are not calculated. When two or more action types exist simultaneously in the same diagnostic action window and the control record cannot be split into individual action time periods, it is marked as a composite action window. The composite action window is used for operation parameter recording and not for single action response value calculation.

[0070] Before calculating the post-action deviation residual rate, the terminal voltage deviation rate before the diagnostic action is compared with the minimum effective deviation rate of the terminal voltage. When the terminal voltage deviation rate before the diagnostic action is lower than the minimum effective deviation rate of the terminal voltage, the post-action deviation residual rate is not calculated, and a record of insufficient terminal voltage deviation before the action is generated. The post-action deviation residual rate is expressed as: ; in: Indicates the deviation residual rate after the action; This indicates the terminal voltage deviation rate before the diagnostic action; This indicates the deviation rate of the terminal voltage after the diagnostic action; This indicates the positive number representing the deviation rate of the terminal voltage from zero.

[0071] Before calculating the parameter changes before and after the action, the deviation rates of the end voltage, low-frequency impedance, high-frequency impedance, and gas pressure difference before the diagnostic action are compared with the corresponding minimum effective deviation rates. Parameters below the corresponding minimum effective deviation rate are not included in the action response value calculation, and a record of insufficient deviation before the action is generated. The purging and drainage action response value is expressed as follows: ; The gas adjustment response value is expressed as follows: ; The humidity adjustment response value is expressed as follows: ; in: This indicates the response value of the purging and drainage action; This indicates the response value of the gas adjustment action; This indicates the response value of the humidity adjustment action; and These represent the low-frequency impedance deviation rates before and after the diagnostic procedure, respectively. and These represent the high-frequency impedance deviation rates before and after the diagnostic procedure, respectively. and These represent the deviation rates of the gas pressure difference before and after the diagnostic procedure; This indicates a positive number representing the low-frequency impedance deviation rate to prevent zero deviation. This indicates a positive number representing the high-frequency impedance deviation rate to prevent zero deviation. This indicates the deviation rate of the gas pressure difference from zero, represented by a positive number.

[0072] The calculation results of the end residue are expressed as follows: ; in: This indicates the calculation results for end-piece residue; This indicates a deviation from the residual rate threshold after the action; This indicates the threshold value for the deviation rate of the current collector contact resistance; This indicates the threshold for the rate of deviation of the end temperature. This indicates a positive number indicating a deviation from the residual rate after the action, preventing it from reaching zero. This indicates the deviation rate of the collector contact resistance from zero, represented by a positive number. This indicates the positive number representing the deviation rate of the end temperature from zero.

[0073] The minimum effective deviation rate of end voltage, the minimum effective deviation rate of low-frequency impedance, the minimum effective deviation rate of high-frequency impedance, the minimum effective deviation rate of gas pressure difference, and each zero-positive value are all determined by the sensor acquisition accuracy, parameter calculation resolution, factory calibration documents, bench test documents, and stable operation samples of the same type of fuel cell stack, and are bound to the threshold version number for storage. Each component in the action response value is a dimensionless change ratio, and each component in the end residual calculation result is normalized to the corresponding threshold before entering the same comparison domain; the equal division calculation rule is verified by bench samples and maintenance review samples. When the verification fails, the corresponding action response threshold or end residual judgment threshold is marked as pending verification, and the relevant records do not directly generate a conclusion to determine the source of the fault.

[0074] The rate of decrease in insulation resistance is expressed as: ; in, Indicates the rate of decrease in insulation resistance; Indicates the insulation resistance within the current sampling window; This represents the baseline value of insulation resistance within the same operating condition range. When the insulation resistance within the current sampling window is greater than or equal to the baseline value within the same operating condition range, the insulation resistance decrease rate is recorded as 0; when the insulation resistance within the current sampling window is less than the corresponding baseline value, the insulation resistance decrease rate is calculated. If the insulation detection module is in an abnormal state or the insulation resistance baseline value is missing, the insulation resistance decrease rate is not calculated.

[0075] The slope of the change in the contact resistance at the current collector is expressed as: ; in, This indicates the slope of the change in the contact resistance at the current collector; Indicates the first The contact resistance of the current collector corresponding to each sampling window; This indicates the contact resistance of the current collector corresponding to the first sampling window; This indicates the time interval between two sampling windows. Indicates the sequence number of the continuous sampling window. Indicates the first The sampling time corresponding to each sampling window This indicates the sampling time corresponding to the first sampling window. The two sampling windows involved in the calculation must belong to the same operating condition range, the same current collector connection location, and the same parameter source marker. When the operating time interval is less than the minimum operating time interval, the slope of the current collector contact resistance change is not calculated. The minimum operating time interval is determined based on the sampling period, the current collector contact resistance sampling stabilization time, the temperature influence stabilization time, and the maintenance review records of the same type of fuel cell stack. When the time interval of consecutive sampling windows is less than the minimum operating time interval, the slope of the current collector contact resistance change is not calculated, and a record of insufficient operating time interval is generated.

[0076] After completing all calculations, an end-point parameter calculation record is generated. This record includes the sampling window number, end-point direction, operating condition interval number, threshold version number, parameter source marker, end-point voltage deviation rate, high-frequency impedance deviation rate, low-frequency impedance deviation rate, end-point temperature deviation rate, current collector contact resistance deviation rate, gas pressure difference deviation rate, post-action deviation residual rate, purging / drainage action response value, gas adjustment action response value, humidity adjustment action response value, end-point residual calculation result, insulation resistance decrease rate, current collector contact resistance change slope, and calculation validity status. Valid calculation statuses include valid, missing parameters, unavailable thresholds, inconsistent operating conditions, mismatched sampling windows, and inconsistent parameter sources. End-point parameter calculation records in valid status are entered into the fault source judgment; end-point parameter calculation records in other statuses are written into the parameter record to be reviewed.

[0077] S4. Based on the calculation results of parameter changes before and after the action, the deviation residual rate after the action, the calculation results of end residual, the insulation resistance decrease rate and the slope of the current collector contact resistance change, perform fault source classification and judgment processing, classify the fault source into candidates of abnormal operation status, candidates of coexistence of abnormal operation status and end residual, candidates of abnormal end current collection, or candidates of end sealing or interface matching, and generate fault source judgment records.

[0078] The fault source classification and judgment process refers to comparing the calculated results of parameter changes before and after the action with the action response judgment threshold, comparing the deviation residual rate after the action with the deviation residual rate threshold after the action, and combining the end residual calculation results, insulation resistance decrease rate and current collector contact resistance change slope to classify and mark the fault source and generate a fault source judgment record.

[0079] The system reads the end-point parameter calculation record, operating condition interval number, threshold version number, threshold status, parameter source marker, and parameter record to be reviewed. If the end-point parameter calculation record is valid and the threshold being compared is active, the system proceeds to fault source determination. If the end-point parameter calculation record is in a state of parameter missing, threshold unavailable, inconsistent operating conditions, mismatched sampling window, or inconsistent parameter source, no fault source conclusion is generated; instead, a fault source review record is generated. The fault source review record includes the sampling window number, end-point direction, missing parameter name, unavailable threshold name, parameter source marker, and review reason.

[0080] The fault source classification and judgment process reads the transition difference verification results, end local difference markers, and end extended difference markers. If the end local difference marker is valid, and the deviation residual rate, end residual calculation result, insulation resistance decrease rate, or current collector contact resistance change slope after the action meet the corresponding judgment conditions, then end current collector anomaly candidates or end sealing or interface matching candidates are preferentially generated. If the end extended difference marker is valid, and the parameter change calculation result before and after the action reaches the corresponding action response judgment threshold, then operating status anomaly candidates or candidates with both operating status anomaly and end residual are preferentially generated. When the transition difference verification status is pending review, the fault source judgment record is marked as a candidate pending review.

[0081] Fault source determination is based on action response, post-action deviation residual rate, and end-point residual calculation results. The hydrogen-side pressure difference deviation rate and the air-side pressure difference deviation rate are compared with gas pressure difference deviation rate thresholds. To determine if the gas pressure difference deviation condition is met, either the hydrogen-side or air-side pressure difference deviation rate must reach the gas pressure difference deviation rate threshold. Conversely, to determine if the gas pressure difference deviation condition is not met, neither the hydrogen-side nor air-side pressure difference deviation rate must reach the gas pressure difference deviation rate threshold. Fault source determination records are generated separately for the first and second ends; candidate records from the two ends are not merged.

[0082] Candidates for operational anomalies include candidates for gas supply anomalies, flooding, drainage anomalies, membrane moisture anomalies, and thermal management anomalies. Candidates for coexistence of operational anomalies and end-point residue refer to records where the action response value reaches the corresponding action response threshold, and the post-action deviation from the residue rate threshold also reaches the post-action deviation from the residue rate threshold. Candidates for end-point manifold anomalies include candidates for manifold contact anomalies, manifold corrosion candidates, and composite conductive layer failure candidates. Candidates for end-point sealing or interface mismatch include end-point sealing anomalies and interface mismatch anomalies.

[0083] When the action response value reaches the corresponding action response threshold, and the post-action deviation residual rate does not reach the post-action deviation residual rate threshold, a candidate record for abnormal operating status is generated. When the action response value reaches the corresponding action response threshold, and the post-action deviation residual rate also reaches the post-action deviation residual rate threshold, a candidate record for both abnormal operating status and end residue is generated, and the end source is further determined based on the calculation results of the current collector contact resistance deviation rate, end temperature deviation rate, insulation resistance decrease rate, and end residue. When none of the three types of action response values ​​reach the corresponding action response threshold, but the post-action deviation residual rate reaches the post-action deviation residual rate threshold, an end current collector candidate record or an end sealing candidate record is generated. When the action response value is negative, it is saved according to the actual calculation result, and this action response value does not meet the corresponding action response threshold.

[0084] When the terminal voltage deviation rate reaches the terminal voltage deviation rate threshold, the low-frequency impedance deviation rate reaches the low-frequency impedance deviation rate threshold, and the gas pressure difference deviation condition is met, a candidate record for mass transfer or drainage anomaly is generated. The judgment condition is: ; ; ; in, Indicates the deviation rate of the terminal voltage; Indicates the threshold value for the terminal voltage deviation rate; Indicates the low-frequency impedance deviation rate; This indicates the threshold for low-frequency impedance deviation. This indicates the deviation rate of the pressure difference on the hydrogen side or the deviation rate of the pressure difference on the air side. This represents the threshold for gas pressure difference deviation rate.

[0085] In the candidate records of mass transfer or drainage anomalies, if the purging and drainage action response value reaches the purging and drainage action response threshold, and the post-action deviation from the residual rate does not reach the post-action deviation from the residual rate threshold, then the fault source is marked as flooding or drainage anomaly. The judgment condition is: ; ; in, This indicates the response value of the purging and drainage action; Indicates the response threshold for purging and drainage actions; Indicates the deviation residual rate after the action; This indicates the deviation from the residual rate threshold after the action. If the drain valve opening time, drain valve closing time, drainage duration, or drainage action feedback status does not meet the drainage control record, the fault source is marked as drainage abnormality; if the end voltage deviation rate, low frequency impedance deviation rate, and gas pressure difference deviation rate decrease after the purging action, and no abnormal record is formed for the drain valve status, the fault source is marked as flooding.

[0086] When the terminal voltage deviation rate reaches the terminal voltage deviation rate threshold, and the gas pressure difference deviation condition is met, a candidate record for gas supply anomaly is generated. If the gas adjustment action response value reaches the gas adjustment action response threshold, and the post-action deviation residual rate does not reach the post-action deviation residual rate threshold, then the fault source is marked as gas supply anomaly. The judgment condition is: ; ; ; ; in, This indicates the response value of the gas adjustment action; This indicates the response threshold for gas adjustment actions. When the hydrogen-side differential pressure deviation rate reaches the gas differential pressure deviation rate threshold but the air-side differential pressure deviation rate does not, the gas supply anomaly location is marked as the hydrogen side; when the air-side differential pressure deviation rate reaches the gas differential pressure deviation rate threshold but the hydrogen-side differential pressure deviation rate does not, the gas supply anomaly location is marked as the air side; when both the hydrogen-side and air-side differential pressure deviation rates reach the gas differential pressure deviation rate threshold, the gas supply anomaly location is marked as a dual-side gas passage. If the gas adjustment action only includes hydrogen metering ratio adjustment, the gas supply anomaly location is limited to the hydrogen side; if the gas adjustment action only includes air metering ratio adjustment, the gas supply anomaly location is limited to the air side.

[0087] When the high-frequency impedance deviation rate reaches the high-frequency impedance deviation rate threshold, and the gas pressure difference deviation condition is not met, candidate records for membrane moisture anomaly or ohmic internal resistance anomaly are generated. The gas pressure difference deviation condition not being met means that neither the hydrogen-side pressure difference deviation rate nor the air-side pressure difference deviation rate reaches the gas pressure difference deviation rate threshold. The judgment condition is: ; ; ; in, Indicates the high-frequency impedance deviation rate; This indicates the threshold for high-frequency impedance deviation rate; This indicates the deviation rate of the pressure difference on the hydrogen side; Indicates the deviation rate of air-side pressure difference; This represents the threshold for gas pressure difference deviation rate.

[0088] If the humidity adjustment response value reaches the humidity adjustment response threshold, but the post-action deviation residual rate does not reach the post-action deviation residual rate threshold, the fault source is marked as membrane moisture anomaly. If the humidity adjustment response value does not reach the humidity adjustment response threshold, but the high-frequency impedance deviation rate still reaches the high-frequency impedance deviation rate threshold, the fault source is marked as ohmic internal resistance anomaly or fuel cell stack aging candidate record. If the hydrogen-side pressure difference deviation rate or air-side pressure difference deviation rate reaches the gas pressure difference deviation rate threshold, it will not proceed to the membrane moisture anomaly or ohmic internal resistance anomaly candidate judgment, and will instead proceed to the fault source classification judgment and processing corresponding to gas supply anomaly, flooding, or drainage anomaly.

[0089] Gas pressure difference deviation rates are stored separately for the gas side and the hydrogen side pressure difference deviation rate is denoted as... Air pressure difference deviation rate is denoted as When the condition for judging gas pressure difference deviation is met, it satisfies or When the gas pressure difference deviation condition is not met, and simultaneously the following conditions are satisfied: and The judgment condition is ; ;or ; ;in, This indicates the response value of the humidity adjustment action; This indicates the threshold for responding to humidity adjustment actions.

[0090] When the end temperature deviation rate reaches the end temperature deviation rate threshold, a cooling-side anomaly flag is established. If the collector contact resistance deviation rate does not reach the collector contact resistance deviation rate threshold, a thermal management anomaly candidate record is generated. The cooling-side anomaly flag is determined based on whether the coolant inlet temperature, coolant outlet temperature, and coolant flow rate feedback value exceed the corresponding operating reference range. If any parameter exceeds the corresponding operating reference range, the cooling-side anomaly flag is established; if none exceed the corresponding operating reference range, the cooling-side anomaly flag is invalid. The judgment condition is: ; ; ; in, Indicates the deviation rate of the end temperature; This indicates the threshold for the rate of deviation of the end temperature. Indicates the deviation rate of the contact resistance at the current collector; This indicates the threshold value for the deviation rate of the current collector contact resistance; This indicates an abnormality on the cooling side. It is set to 1 when the coolant inlet temperature, coolant outlet temperature, or coolant flow feedback value exceeds the corresponding operating reference range, and to 0 when it does not exceed the corresponding operating reference range.

[0091] When the cooling-side anomaly marker is 0, no thermal management anomaly candidate record is generated; when the collector contact resistance deviation rate is missing, it is not considered as not reaching the collector contact resistance deviation rate threshold, and a collector contact resistance deviation rate record to be reviewed is generated.

[0092] When the terminal voltage deviation rate reaches the terminal voltage deviation rate threshold, the collector contact resistance deviation rate reaches the collector contact resistance deviation rate threshold, and the post-action deviation residual rate reaches the post-action deviation residual rate threshold, a candidate terminal collector record is generated. The judgment condition is: ; ; ; In the candidate records for end-collection, if the purge and drainage response values, gas adjustment response values, and humidity adjustment response values ​​all fail to reach their corresponding response thresholds, and the end-residue calculation result reaches the end-residue judgment threshold, then the fault source is marked as an abnormal contact at the collector end. The judgment condition is: ; ; ; ; in, This indicates the calculation results for end-piece residue; This indicates the threshold for determining end-point residual resistance. When the source of the collector contact resistance is marked as direct measurement, the comparison is made according to the threshold corresponding to the direct measurement source; when the source is marked as equivalent calculation, the comparison is made according to the threshold corresponding to the equivalent calculation source. If the parameter source markings are inconsistent, a record of inconsistent collector contact resistance source is generated, but no conclusion of abnormal collector contact is generated.

[0093] When the slope of the current collector contact resistance change reaches a contact resistance change slope threshold, the current collector contact resistance deviation rate reaches a current collector contact resistance deviation rate threshold, and the end temperature deviation rate reaches an end temperature deviation threshold, a candidate record for current collector corrosion or composite conductive layer failure is generated. The judgment conditions are: ; ; ; in, This indicates the slope of the change in the contact resistance at the current collector; This indicates the threshold for the slope of the contact resistance change. If there are existing disassembly / inspection records, maintenance records, or retest records after replacing the end current collector before diagnosis, and the records indicate conductive layer peeling, conductive layer damage, or abnormal contact between the conductive layer and the current collector substrate, then the candidate record is refined and marked as composite conductive layer failure. If the records indicate current collector substrate corrosion or end conductive interface corrosion, then the candidate record is refined and marked as current collector corrosion. If no confirmation record is available, it is retained as a candidate record for current collector corrosion or composite conductive layer failure.

[0094] When the end voltage deviation rate reaches the end voltage deviation rate threshold, the end temperature deviation rate reaches the end temperature deviation rate threshold, the insulation resistance decrease rate reaches the insulation resistance decrease rate threshold, and the gas pressure difference deviation condition is met, a candidate record for end sealing or interface matching is generated. The judgment condition is: ; ; ; ; in, Indicates the rate of decrease in insulation resistance; This indicates the threshold for the rate of decrease in insulation resistance. If the occurrence time of the leak detection record, insulation detection record, or end gas pressure recovery record falls within the same sampling window or the observation window after an adjacent action, and the end direction in the record matches the end direction in the end seal or interface matching candidate record, the end seal or interface matching candidate record is retained. If the maintenance record or inspection record points to seal damage, insufficient seal compression, seal aging, or seal groove contamination, the fault source is marked as end seal anomaly. If the maintenance record or assembly verification record points to deviation in the position of the manifold, end plate, water / gas interface, or pipeline interface, the fault source is marked as interface matching anomaly. If a maintenance record or assembly verification record is missing, it is retained as an end seal or interface matching candidate record. If the occurrence time or end direction is inconsistent, an end seal anomaly conclusion is not generated, and the corresponding record is transferred to the end manifold candidate record verification.

[0095] When both the end-sealing or interface matching candidate and the current collector corrosion or composite conductive layer failure candidate are met, if both the insulation resistance decrease rate and gas pressure difference deviation conditions are met, and the leakage detection record, insulation detection record, or end gas pressure recovery record is consistent with the end direction, then the end-sealing or interface matching candidate is retained; if the time and end direction correspondence is not met, the current collector corrosion or composite conductive layer failure candidate is retained. When both the current collector contact anomaly and the current collector corrosion or composite conductive layer failure candidate are met, the current collector contact anomaly candidate is retained, depending on whether the current collector contact resistance change slope reaches the contact resistance change slope threshold.

[0096] After determining the fault source, a fault source determination record is generated. This record includes the sampling window number, end direction, operating condition interval number, fault candidate type, fault source marker, satisfied judgment conditions, unsatisfied judgment conditions, action type, action response value, post-action deviation residual rate, end residual calculation result, parameter source marker, threshold version number, and judgment status. Judgment status includes classified, candidate pending verification, and insufficient parameters; when the threshold status is pending verification, the judgment status is marked as candidate pending verification. Classified records are entered into the fault diagnosis conclusion output, while candidate pending verification records and insufficient parameter records are written into the fault source verification record.

[0097] S5. Generate a fuel cell stack fault diagnosis conclusion based on the fault source judgment record, obtain the maintenance review result and compare it with the fuel cell stack fault diagnosis conclusion, write the comparison result into the confirmed fault sample set, the candidate correction sample set or the misjudged correction sample set, and update the judgment threshold or fault source judgment condition under the corresponding operating condition range, end direction and parameter source mark according to the confirmed fault sample set and the misjudged correction sample set.

[0098] After determining the fault source, the system reads the fault source determination record, fault source verification record, end parameter calculation record, parameter record to be verified, operating condition interval number, threshold version number, and basic record version number to generate a fuel cell stack fault diagnosis conclusion. When the fault source determination record's determination status is "classified," a definitive fault diagnosis conclusion is generated; when the determination status is "candidate to be verified," a candidate fault diagnosis conclusion and verification prompt are generated; when the determination status is "parameters insufficient," a parameter insufficient record is generated, and no definitive fault source is output.

[0099] The fault diagnosis conclusion includes the conclusion number, stack number, sampling window number, end direction, operating condition interval number, associated single cell number, fault candidate type, fault source marker, met judgment conditions, unmet judgment conditions, action type, action response value, post-action deviation residual rate, end residual calculation result, current collector contact resistance deviation rate, current collector contact resistance change slope, end temperature deviation rate, insulation resistance decrease rate, hydrogen side pressure difference deviation rate, air side pressure difference deviation rate, parameter source marker, threshold version number, and conclusion status. The conclusion status includes confirmed, candidate pending verification, and insufficient parameters. The fault diagnosis conclusion is output separately for the first and second ends, and the results from the two end directions are not merged into the same fault source.

[0100] Fault source markers include abnormal gas supply, flooding, abnormal drainage, abnormal membrane moisture, abnormal ohmic internal resistance, fuel cell aging, abnormal thermal management, abnormal current collector contact, current collector corrosion or composite conductive layer failure, candidate end seal or interface mismatch, and end seal and interface mismatch. Candidate markers are updated to confirm the corresponding fault source after obtaining disassembly and inspection records, maintenance records, retest records after component replacement, seal inspection records, leak detection records, or assembly verification records. A fault handling output record is generated based on the fault source markers. The fault handling output record records the fault source marker, the direction of the fault occurrence end, the name of the associated parameter, the value of the associated parameter, the object to be verified, and the candidate confirmation marker.

[0101] After troubleshooting or maintenance inspection is completed, obtain the maintenance review results. The maintenance review results include the review record number, fuel cell number, sampling window number, end direction, actual fault type, actual fault location, maintenance actions, disassembly and inspection results, replaced component numbers, single cell voltage before maintenance, single cell voltage after maintenance, high-frequency impedance before maintenance, high-frequency impedance after maintenance, low-frequency impedance before maintenance, low-frequency impedance after maintenance, hydrogen-side pressure difference before maintenance, hydrogen-side pressure difference after maintenance, air-side pressure difference before maintenance, air-side pressure difference after maintenance, end temperature before maintenance, end temperature after maintenance, current collector contact resistance before maintenance, current collector contact resistance after maintenance, insulation resistance before maintenance, insulation resistance after maintenance, post-maintenance retest window number, and review conclusion status. Review conclusion status includes diagnostic consistency, partial consistency, misjudgment correction, and review invalid.

[0102] The maintenance review results are compared with the fault diagnosis conclusions. For abnormal gas supply, flooding, drainage, membrane moisture, and thermal management, if the actual fault type, actual fault location, and end direction are consistent with the fault diagnosis conclusion, and the deviation rate of the fault source judgment within the post-maintenance retest window has not reached the corresponding threshold for at least two consecutive sampling windows, the review conclusion status is marked as diagnostically consistent. For abnormal collector contact, collector plate corrosion, composite conductive layer failure, end sealing abnormality, and interface matching abnormality, if the disassembly and inspection records, maintenance records, replacement records, or post-repair retest records are consistent with the fault diagnosis conclusion, the review conclusion status is marked as diagnostically consistent. If the actual fault type is consistent but the actual fault location or end direction is inconsistent, the review conclusion status is marked as partially consistent. If the fault type confirmed by the maintenance review results is inconsistent with the fault diagnosis conclusion, the review conclusion status is marked as misjudgment correction. If the review record lacks maintenance actions, retest windows, or key parameters, the review conclusion status is marked as review invalid.

[0103] When multiple fault diagnosis conclusions correspond to the same maintenance review result, they are compared separately according to end direction, fault type, and associated parameters, and are not merged into a single review result. When the maintenance review result corresponds to multiple actual fault types, multiple actual fault types are retained, and each actual fault type is compared separately with the fault diagnosis conclusion.

[0104] When the verification conclusion status is "diagnostic consistency," the corresponding sampling window, end parameter calculation record, fault source judgment record, and maintenance verification result are written into the confirmed fault sample set, and saved separately according to stack model, operating condition range, end direction, fault source marker, and parameter source marker. When the verification conclusion status is "partially consistent," the corresponding record is written into the candidate correction sample set; when the verification conclusion status is "misjudgment correction," the corresponding record is written into the misjudgment correction sample set, and the misjudgment source parameter, misjudgment threshold name, misjudgment occurrence window, and actual fault type are recorded. When the verification conclusion status is "verification invalid," only the verification record is saved, and it does not participate in threshold update and judgment condition update.

[0105] When the number of confirmed fault samples under the same operating condition range, the same end direction, and the same fault source marker reaches the preset update quantity, and the consistency of post-maintenance retesting meets the requirements, the corresponding threshold is updated based on the set of confirmed fault samples. Post-maintenance retesting consistency means that the post-maintenance retesting window is consistent with the operating condition range corresponding to the fault diagnosis conclusion, and the parameters involved in judging the fault source no longer meet the original fault judgment conditions within at least two consecutive sampling windows. For current collector corrosion, composite conductive layer failure, end sealing abnormalities, or interface mismatch abnormalities, after replacement, repair, or reassembly, if the corresponding current collector contact resistance, insulation resistance, gas pressure difference, or end temperature detection results no longer meet the original candidate judgment conditions, the post-maintenance retesting consistency is deemed to meet the requirements.

[0106] When updating thresholds or fault source judgment conditions based on sample sets, the operating condition range, end direction, and parameter source markers are used as matching conditions for the update objects. Samples with inconsistent matching conditions are not merged for the same threshold version or the same fault source judgment condition. The preset update quantity, the most recent preset quantity, and the preset number of times are determined by the factory calibration document, bench test document, or maintenance statistics document of the same model of fuel cell stack. If there is no corresponding document, a threshold to be verified is allowed to be generated when the number of confirmed fault samples in a single operating condition range is not less than 10. If the same parameter combination has been misjudged and corrected no less than 3 times in the most recent 30 maintenance review results, the parameter combination is adjusted to the review condition. Threshold update objects include end voltage deviation rate threshold, low frequency impedance deviation rate threshold, high frequency impedance deviation rate threshold, gas pressure difference deviation rate threshold, end temperature deviation rate threshold, current collector contact resistance deviation rate threshold, post-operation deviation residual rate threshold, end residual judgment threshold, insulation resistance decrease rate threshold, and contact resistance change slope threshold. When updating thresholds, sample values ​​are rearranged according to the classification results of historical stable samples, confirmed fault samples, and misjudged correction samples. The updated thresholds are generated using the quantile boundaries of the corresponding operating condition intervals, sensor acquisition accuracy compensation values, and maintenance review allowable deviations. The updated thresholds are initially marked as pending verification and do not directly replace the effective thresholds. Once the pending verification thresholds meet the verification requirements in the new stable samples and maintenance review samples, their effective status is updated to effective. Action response thresholds are updated based on confirmed fault samples and misjudged correction samples. Purging and drainage action response thresholds are updated based on flooding samples, drainage anomaly samples, and corresponding misjudged correction samples; gas adjustment action response thresholds are updated based on gas supply anomaly samples and corresponding misjudged correction samples; humidity adjustment action response thresholds are updated based on membrane moisture anomaly samples and corresponding misjudged correction samples. When there is no overlapping range between the action response values ​​of two types of samples, the median value between adjacent boundaries is selected as the updated action response threshold; when there is an overlapping range, the overlapping range is marked as a review range, and records falling into the review range are not directly output to determine the source of the fault.

[0107] When the same parameter combination results in at least a preset number of misjudgments in the most recent preset number of maintenance review results, the parameter combination will be changed from a direct classification condition to a review condition. The parameter combination includes at least two parameters from the following: end voltage deviation rate, end impedance deviation rate, gas pressure difference deviation rate, end temperature deviation rate, current collector contact resistance deviation rate, post-action deviation residual rate, and end residual calculation results. After being adjusted to a review condition, this parameter combination must be satisfied together with the action response value, post-action deviation residual rate, current collector contact resistance change slope, insulation resistance decrease rate, or maintenance review record before a definitive fault source can be generated.

[0108] After the threshold or judgment condition is updated, a diagnostic record update result is generated. The diagnostic record update result includes the name of the updated object, the version number before the update, the version number after the update, the number of confirmed fault samples involved in the update, the number of misjudged and corrected samples involved in the update, the operating condition interval number, the end direction, the parameter source marker, the update reason, the generation time, and the effective status. The effective status includes pending verification, effective, disabled, and replaced. Replaced thresholds and judgment conditions retain their historical versions and are not used in the current fault source judgment.

[0109] After completing the output of fault diagnosis conclusions, comparison of maintenance review results, and update of diagnostic records, the fault diagnosis conclusions, fault handling output records, maintenance review results, confirmed fault sample sets, candidate correction sample sets, misjudgment correction sample sets, and diagnostic record update results are stored in the fuel cell stack diagnostic database and bound to the stack number, basic record version number, threshold version number, sampling window number, and end direction for storage.

[0110] In summary, this invention calculates the end-point difference calculation results, the parameter change calculation results before and after the action, and the deviation residual rate after the action, so that the end-point operation changes and end-point residual states are jointly included in the fault source classification and judgment process, forming candidates for abnormal operation status, candidates for abnormal end current collection, or candidates for end sealing or interface matching; by writing the maintenance review results into the corresponding sample set and updating the judgment threshold or fault source judgment conditions, the fuel cell stack fault diagnosis conclusions are made into a write-back record, which improves the accuracy of end-point fault source classification and the ability to correct diagnostic records.

[0111] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for fault diagnosis of fuel cells based on stack characteristic parameters, characterized in that, include: Acquire basic diagnostic data of fuel cell stacks, divide fuel cell stacks according to the arrangement of individual cells based on basic diagnostic data, and form end difference judgment thresholds, action response judgment thresholds, end residue and sealing judgment thresholds and contact resistance change slope thresholds according to operating condition intervals. Collect fuel cell stack operation characteristic data during the operation of the fuel cell stack, divide the stack operation characteristic data into time and verify its validity according to the operation observation window, diagnostic action window and post-action observation window, and generate a partitioned stack characteristic parameter sequence; Based on the characteristic parameter sequence of the partitioned stack, the end difference calculation results are calculated. Based on the parameter changes before and after the diagnostic action, the parameter changes before and after the action and the deviation residual rate after the action are calculated. Based on the end difference calculation results, insulation resistance and current collector contact resistance within the continuous sampling window, the end residual calculation results, insulation resistance decrease rate and current collector contact resistance change slope are calculated. Based on the calculation results of parameter changes before and after the action, the deviation residual rate after the action, the calculation results of end residual, the insulation resistance decrease rate and the slope of the current collector contact resistance change, the fault source classification and judgment process is performed. The fault source is classified into candidates of abnormal operation status, candidates of coexistence of abnormal operation status and end residual, candidates of abnormal end current collection, or candidates of end sealing or interface matching, and a fault source judgment record is generated. The fault diagnosis conclusion of the fuel cell stack is generated based on the fault source judgment record. The maintenance review result is obtained and compared with the fault diagnosis conclusion of the fuel cell stack. The comparison result is written into the confirmed fault sample set, the candidate correction sample set, or the misjudged correction sample set. The judgment threshold or fault source judgment condition under the corresponding operating condition range, end direction, and parameter source mark is updated according to the confirmed fault sample set and the misjudged correction sample set.

2. The fuel cell fault diagnosis method based on stack characteristic parameters as described in claim 1, characterized in that, The process of acquiring basic diagnostic data for the fuel cell stack and dividing the fuel cell stack according to the arrangement of individual cells based on the basic diagnostic data includes: After obtaining the basic diagnostic data of the fuel cell stack, read the total number of individual cells, the arrangement order of individual cells, the position of the end current collector, the position of the end sealing interface, and the voltage sampling channel mapping relationship. Single cells are continuously selected from the first end of the battery stack to form a first end battery group, single cells are continuously selected from the second end of the battery stack to form a second end battery group, and single cells are continuously selected from the middle position of the battery stack to form a middle battery group. The single cells between the first end battery group and the middle battery group, and the single cells between the second end battery group and the middle battery group are respectively written into the first transition battery group and the second transition battery group, generating a partition marking record with end direction, single cell number and sampling channel number.

3. The fuel cell fault diagnosis method based on stack characteristic parameters as described in claim 1 or 2, characterized in that, The thresholds for judging end differences, judging action response, judging end residue and sealing, and judging contact resistance change slope based on the operating condition range include: When forming the judgment threshold, the operating condition range is determined according to the current density, coolant inlet temperature, hydrogen metering ratio and air metering ratio. Within the same operating condition range, historical stable samples with no communication interruption, no range over-limit, no shutdown switching, no protection derating and no incomplete maintenance records are selected. After arranging historical stable samples according to parameter values, and combining factory calibration data, bench test data, sensor acquisition accuracy compensation values, and maintenance review allowable deviations, thresholds for end difference judgment, action response judgment, end residue and sealing judgment, and contact resistance change slope are generated.

4. The fuel cell fault diagnosis method based on stack characteristic parameters as described in claim 3, characterized in that, The process of collecting fuel cell stack operation characteristic data during operation, and dividing and validating the stack operation characteristic data according to the operation observation window, diagnostic action window, and post-action observation window, includes: When dividing the fuel cell stack operation characteristic data into time segments, read the current loading change records, purging action records, drainage action records, gas metering ratio adjustment records, inlet humidity adjustment records, coolant flow rate adjustment records, and corresponding action feedback records. The continuous sampling segment in which no diagnostic action occurs is marked as the running observation window; the sampling segment between the start time and the end time of the diagnostic action is marked as the diagnostic action window; and the sampling segment after the diagnostic action is completed and the feedback status reaches a stable condition is marked as the post-action observation window. If a sampling segment lacks action type, action start time, action end time, or action feedback status, write it into the incomplete action record and stop calculating the parameter changes before and after the action in the sampling segment.

5. The fuel cell fault diagnosis method based on stack characteristic parameters as described in claim 4, characterized in that, The step of dividing and validating the stack operation characteristic data according to the operation observation window, diagnostic action window, and post-action observation window, and generating a partitioned stack characteristic parameter sequence includes: When validating the operating characteristic data of the fuel cell stack, the following checks are performed on voltage data, impedance data, gas pressure data, cooling data, temperature data, insulation detection data, and current collector contact resistance data: time consistency check, range check, sampling integrity check, operating condition consistency check, and action record integrity check, according to the sampling timestamp, sampling channel number, sensor range, operating condition interval number, and action record number. Data that passes verification is written into the characteristic parameter sequence of the partitioned stack according to the single-cell partition mark. Data that fails verification is written into the parameter record to be reviewed. The parameter record to be reviewed is not included in the calculation results of end difference, parameter change before and after operation, end residue, insulation resistance decrease rate and current collector contact resistance change slope.

6. The fuel cell fault diagnosis method based on stack characteristic parameters as described in claim 5, characterized in that, The calculation of end-point differences based on the characteristic parameter sequence of the partitioned stack, the calculation of parameter changes before and after diagnostic actions, the calculation of deviation residual rate after actions based on parameter changes before and after diagnostic actions, and the calculation of end-point residual results, insulation resistance decrease rate, and current collector contact resistance change slope based on end-point difference calculation results, insulation resistance decrease rate, and current collector contact resistance change slope include: When calculating the end difference calculation results, the first end battery pack is compared with the middle battery pack to generate the first end difference calculation result, and the second end battery pack is compared with the middle battery pack to generate the second end difference calculation result; The differences of the first end battery pack, the first transition battery pack, and the middle battery pack are compared sequentially to generate the first transition difference verification result. The differences between the second end battery pack, the second transition battery pack, and the middle battery pack are compared sequentially to generate the second transition difference verification result. Based on the first transition difference verification result and the second transition difference verification result, the end difference calculation result is marked as end local difference or end extended difference.

7. The fuel cell fault diagnosis method based on stack characteristic parameters as described in claim 6, characterized in that, The calculation of end-point differences based on the characteristic parameter sequence of the partitioned stack, the calculation of parameter changes before and after diagnostic actions, the calculation of deviation residual rate after actions based on parameter changes before and after diagnostic actions, and the calculation of end-point residual results, insulation resistance decrease rate, and current collector contact resistance change slope based on end-point difference calculation results, insulation resistance decrease rate, and current collector contact resistance change slope include: When calculating the parameter changes before and after the action, the parameter changes corresponding to the purging action and the drainage action are recorded as the purging and drainage action response value, and the parameter changes corresponding to the gas metering ratio adjustment are recorded as the gas adjustment action response value. The parameters before the action are taken from the nearest valid running observation window before the start time of the diagnostic action, and the parameters after the action are taken from the first valid post-action observation window after the end of the diagnostic action. The post-action deviation residual rate is generated based on the ratio of the post-action voltage deviation rate to the pre-action voltage deviation rate. If there are more than two action types within the same diagnostic action window and the control record cannot be split into action time periods, the diagnostic action window will be marked as a composite action window, and the calculation of single action response values ​​will be stopped.

8. The fuel cell fault diagnosis method based on stack characteristic parameters as described in claim 7, characterized in that, The calculation of end-point differences based on the characteristic parameter sequence of the partitioned stack, the calculation of parameter changes before and after diagnostic actions, the calculation of deviation residual rate after actions based on parameter changes before and after diagnostic actions, and the calculation of end-point residual results, insulation resistance decrease rate, and current collector contact resistance change slope based on end-point difference calculation results, insulation resistance decrease rate, and current collector contact resistance change slope include: When calculating the end residual calculation results, insulation resistance decrease rate and current collector contact resistance change slope, write the post-action deviation residual rate, current collector contact resistance deviation rate and end temperature deviation rate into the same end residual calculation record. The insulation resistance within the current sampling window is compared with the reference value of insulation resistance within the same operating condition range to generate the insulation resistance decrease rate. The collector contact resistances with consistent parameter source markings within the continuous sampling window are arranged in chronological order of sampling time. The slope of the collector contact resistance change is generated based on the change in collector contact resistance between the first and last sampling windows and the time interval.

9. The fuel cell fault diagnosis method based on stack characteristic parameters as described in claim 8, characterized in that, The process of classifying fault sources into candidates for abnormal operating status, candidates for coexistence of abnormal operating status and end-end residue, candidates for abnormal end-end current collection, or candidates for end-end sealing or interface mismatch, and generating fault source judgment records includes: When performing fault source classification and judgment processing, the current action response value is determined according to the diagnostic action type; when the diagnostic action type is a purging action or a drainage action, the purging and drainage action response value is compared with the purging and drainage action response threshold; when the diagnostic action type is a gas metering ratio adjustment action, the gas adjustment action response value is compared with the gas adjustment action response threshold; when the diagnostic action type is an inlet humidity adjustment action, the humidity adjustment action response value is compared with the humidity adjustment action response threshold. When the current action response value reaches the action response threshold corresponding to the diagnostic action type, and the post-action deviation residual rate does not reach the post-action deviation residual rate threshold, an abnormal operation status candidate is generated. When the current action response value reaches the action response threshold corresponding to the diagnostic action type, and the post-action deviation residual rate reaches the post-action deviation residual rate threshold, a candidate for coexistence of abnormal operation status and end residue is generated. If the purge and drainage response value, the gas adjustment response value, or the humidity adjustment response value do not reach the purge and drainage response threshold, and the post-action deviation rate reaches the post-action deviation rate threshold, the end residue calculation result is compared with the end residue judgment threshold, the insulation resistance decrease rate is compared with the insulation resistance decrease rate threshold, and the current collector contact resistance change slope is compared with the contact resistance change slope threshold. Based on the comparison results, end current collector abnormality candidates or end sealing or interface matching candidates are generated.

10. The fuel cell fault diagnosis method based on stack characteristic parameters as described in claim 9, characterized in that, The process of generating a fuel cell stack fault diagnosis conclusion based on the fault source judgment record, obtaining maintenance review results and comparing them with the fuel cell stack fault diagnosis conclusion, writing the comparison results into the confirmed fault sample set, the candidate correction sample set, or the misjudged correction sample set, and updating the judgment threshold or fault source judgment condition under the corresponding operating condition range, end direction, and parameter source marker according to the confirmed fault sample set and the misjudged correction sample set includes: After obtaining the maintenance review results, the actual fault type, actual fault location, end direction, and post-maintenance retest parameters in the maintenance review results are compared with the fuel cell stack fault diagnosis conclusions. When the comparison results show that the diagnoses are consistent, the sampling window, end parameter calculation record, fault source judgment record, and maintenance review result are written into the confirmed fault sample set; If the comparison result is partially consistent, the corresponding record will be written into the candidate correction sample set; When the comparison result indicates a misjudgment correction, the corresponding record is written into the misjudgment correction sample set; When updating the judgment threshold or the judgment condition for the source of the fault, the fault sample set and the misjudgment correction sample set are screened and confirmed according to the operating condition range, end direction and parameter source mark. Samples with inconsistent matching conditions are not merged to generate the same threshold version or the same judgment condition for the source of the fault.

Images