A method and system for diagnosing reversible performance degradation of a high-temperature proton exchange membrane fuel cell system

Abstract

The application discloses a kind of plateau proton exchange membrane fuel cell system reversible performance degradation diagnosis method and system.The prior art is mainly concentrated on the internal fault identification of stack, and the system level reversible performance degradation caused by insufficient identification of altitude increase.The application generates identifiable loading process and unloading process by applying one or more controlled load change instructions to the PEMFC system, synchronously collects voltage response signal and load power response signal, extracts voltage transient deviation, loading time and unloading time;When the loading time change rate is greater than or equal to the preset first threshold value, the unloading time change rate is less than or equal to the negative preset second threshold value, and the voltage transient deviation is greater than or equal to the preset third threshold value, it is determined that there is reversible performance degradation caused by altitude increase.The application does not require additional sensors, and can realize rapid, in-situ altitude degradation diagnosis and grading evaluation using existing battery management system or controller.

Description

Technical Field

[0001] This invention relates to the field of fuel cell technology, and specifically to a method and system for diagnosing reversible performance degradation in a high-altitude proton exchange membrane fuel cell system. Background Technology

[0002] Proton exchange membrane fuel cells (PEMFCs), as a highly efficient and zero-emission power generation device, have shown broad application prospects in automobiles, rail transportation, and other fields. However, when PEMFC systems operate in high-altitude environments, atmospheric pressure and oxygen partial pressure decrease significantly with increasing altitude, leading to reversible performance degradation. According to the international standard atmospheric model, when the altitude increases from sea level to 2000 m, the ambient pressure decreases from approximately 101.3 kPa to approximately 79.5 kPa, and the oxygen partial pressure decreases proportionally by approximately 21% to 22%. This change exposes PEMFC systems to multiple coupled negative effects: intensified electrochemical polarization, reduced air compressor efficiency, and premature concentration polarization. It is worth noting that this performance degradation caused by altitude is, in principle, recoverable when the system returns to standard atmospheric pressure; therefore, it is a reversible performance degradation and should be distinguished from irreversible material aging in long-term durability tests.

[0003] In existing technologies, fuel cell fault diagnosis methods mainly focus on identifying internal faults within the fuel cell stack. For example, existing patents disclose fault diagnosis methods for membrane dryness and flooding based on individual cell voltage differences. For instance, Chinese invention patent CN104282925A discloses a method for fault judgment by calculating the difference and rate of change between the average and minimum individual voltages. Other patents disclose fault diagnosis methods for air starvation, flooding, and membrane dryness based on electrochemical impedance spectroscopy (EIS) and relaxation time distribution methods. For example, Chinese invention patent CN117154149B discloses a multi-fault joint diagnosis technology combining EIS testing and relaxation time distribution analysis. However, the aforementioned existing technologies have the following shortcomings: Existing fault diagnosis technologies primarily address internal faults in fuel cell stacks (such as flooding, membrane drying, flow channel blockage, and airtightness issues). These faults are fundamentally different from the reversible performance degradation caused by reduced external air pressure in high-altitude environments. Internal faults typically require individual cell voltage monitoring modules or dedicated electrochemical impedance spectroscopy (EIS) testing equipment, while altitude-induced degradation is a system-wide, gradual performance decline that current technologies cannot identify or quantify.

[0004] Studies have shown that the dynamic response of PEMFCs in high-altitude environments exhibits a "bidirectional asymmetry" characteristic that can be used for diagnosis: the loading time significantly increases with altitude, while the unloading time decreases, and the voltage transient deviation increases. This dynamic response characteristic, caused by changes in external air pressure, has not been fully utilized in existing schemes that primarily focus on diagnosing internal faults within the fuel cell stack. It is particularly important to emphasize that the "shortened unloading time" phenomenon is contrary to conventional expectations: those skilled in the art, when faced with system performance degradation, typically expect both loading and unloading times to increase, but experiments reveal this inverse trend of shortened unloading time. Existing publicly available technologies have not explicitly suggested using this inverse trend as a specific fingerprint for diagnosing altitude degradation.

[0005] Some existing technologies employ step load testing to extract dynamic response characteristics. They decompose the response curve into fast and slow components using mathematical decomposition, extract the time constant, and compare it with a benchmark value to determine irreversible aging states such as reduced catalyst activity and deterioration of the gas supply channel. However, these methods have the following limitations, fundamentally different from the present invention: First, they require complex signal decomposition algorithms, resulting in a high computational burden and making real-time execution difficult in vehicle battery management systems. Second, their diagnostic targets are irreversible material aging, which is entirely different from the reversible performance degradation caused by altitude increase addressed in this invention. Third, existing technologies have not yet combined the bidirectional asymmetric characteristics of "extended loading time + shortened unloading time" under high-altitude conditions as altitude degradation characteristics. Fourth, existing solutions have not established a threshold relationship between voltage transient deviation and the degree of altitude degradation.

[0006] Existing diagnostic methods generally rely on additional hardware (such as cell voltage monitoring chips, EIS testers, and drainage pressure sensors), resulting in high implementation costs and difficulty in achieving in-situ, rapid diagnostics in the vehicle environment. For fuel cell vehicles that must operate in high-altitude terrains such as the Qinghai-Tibet Plateau and the Andes Mountains, there is an urgent need for a method that can diagnose altitude degradation without additional sensors, utilizing only existing battery management system data. Summary of the Invention

[0007] The technical problem this invention aims to solve is that existing fuel cell fault diagnosis technologies mainly focus on internal fault identification, and are insufficient in identifying system-level reversible performance degradation caused by altitude. Therefore, this invention provides a method and system for diagnosing reversible performance degradation in high-altitude proton exchange membrane fuel cell systems. This method requires no additional sensors and can quickly and in-situ determine altitude degradation and its level through controlled load change tests (e.g., standard step load tests).

[0008] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is: a method for diagnosing reversible performance degradation of a proton exchange membrane fuel cell system at high altitudes, comprising: applying one or more controlled load change commands to the proton exchange membrane fuel cell system to induce identifiable loading and unloading processes in the system, and acquiring system voltage response signals and load power response signals; obtaining voltage transient deviation based on the voltage response signals, and obtaining actual loading time and actual unloading time based on the load power response signals; reading preset benchmark loading time and preset benchmark unloading time, and calculating loading time change rate and unloading time change rate respectively; when the loading time change rate is greater than or equal to a preset first threshold, the unloading time change rate is less than or equal to a negative preset second threshold, and the voltage transient deviation is greater than or equal to a preset third threshold, determining that the proton exchange membrane fuel cell system has reversible performance degradation caused by altitude increase, and generating or outputting diagnostic results; wherein the preset first threshold, preset second threshold, and preset third threshold are pre-calibrated based on benchmark test data of the same type of proton exchange membrane fuel cell system.

[0009] In a preferred embodiment, the preset first threshold is 15% to 25%, the preset second threshold is 8% to 20%, and the preset third threshold is 6% to 10%; the loading time change rate is greater than or equal to the preset first threshold to characterize an extension of the loading time, the unloading time change rate is less than or equal to the negative preset second threshold to characterize a shortening of the unloading time, and the voltage transient deviation is greater than or equal to the preset third threshold to characterize an increase in the transient deviation.

[0010] This invention also provides a diagnostic system for reversible performance degradation of a high-altitude proton exchange membrane fuel cell, comprising a data acquisition module, a feature extraction module, a reference storage module, a comparison calculation module, a degradation determination module, and a diagnostic result output module. The data acquisition module applies one or more controlled load change commands and acquires voltage response signals and load power response signals. The feature extraction module obtains voltage transient deviation, actual loading time, and actual unloading time. The reference storage module stores preset reference values ​​and preset first, second, and third threshold values. The comparison calculation module calculates the loading time change rate and the unloading time change rate. The degradation determination module performs a three-parameter threshold combination determination. The diagnostic result output module outputs a diagnostic result characterizing the presence or absence of reversible performance degradation. The diagnostic system may further include a level determination module, a level output module, and a pre-pressurization control signal output module, used to output a pre-pressurization control signal to the air compressor controller based on degradation level information.

[0011] The aforementioned diagnostic methods and systems can be implemented in controllers or electronic devices via computer programs. When executed by a processor, the computer program can perform functions such as triggering controlled load change commands, signal acquisition, feature extraction, rate of change calculation, threshold comparison, degradation determination, and diagnostic result output.

[0012] It should be noted that the standard step load command is only a preferred embodiment of the present invention. In other embodiments, the controlled load change command may also be a ramp loading command, a pulse loading command, a segmented loading command, or other forms of controlled power change command; the load increase process and the unload process may be generated continuously by the same controlled load change command, or they may be generated separately by multiple controlled load change commands. As long as response data that can be obtained for calculating the loading time change rate, the unloading time change rate, and the voltage transient deviation can be obtained, the diagnostic purpose of the present invention can be achieved.

[0013] It is particularly important to note that existing diagnostic methods based on dynamic response are mainly used to detect internal faults (such as flooding and membrane drying) or assess irreversible aging. The trends of their dynamic time parameters (loading time and unloading time) are typically unidirectional (both increasing or decreasing simultaneously). For example, flooding faults are characterized by a simultaneous increase in loading and unloading times, while membrane drying faults are characterized by a simultaneous decrease or near-unchanged loading and unloading times. However, this invention discovers that under the specific external conditions of high-altitude low-pressure environments, loading and unloading times exhibit an anomalous change in opposite directions—that is, a prolonged loading time and a shortened unloading time. This inverse combination of features can serve as a specific diagnostic feature combination for altitude-induced reversible degradation. Existing publicly available technologies have not explicitly suggested using this inverse combination of changes to diagnose altitude degradation.

[0014] The above technical solution utilizes the "bidirectional asymmetric" characteristic combination of the dynamic response of PEMFC in high-altitude environments, which is fundamentally different from internal faults. Experiments show that under increasing altitude conditions, the loading time of the PEMFC system is significantly prolonged relative to the sea-level reference, while the unloading time is shortened, and the transient voltage deviation increases significantly. The above-mentioned three-parameter inverse change characteristic of "prolonged loading + shortened unloading + increased transient deviation" is an important diagnostic feature combination of altitude-induced reversible degradation. Detailed experimental data can be found in Example 1.

[0015] The physical mechanism behind the aforementioned "shortened unloading time" phenomenon can be explained as follows: In the low-pressure environment of high altitude, the cathode back pressure regulation strategy of the PEMFC system changes. To maintain a suitable pressure difference between the cathode inlet and outlet, the back pressure valve opening is adjusted accordingly, reducing the resistance to gas discharge on the cathode side when the load drops from high power, thus shortening the time required for the system power to return to a low-load steady state. Simultaneously, the lower ambient back pressure directly reduces the exhaust resistance at the cathode outlet, accelerating the pressure relief process during load shedding. Meanwhile, during the loading phase, due to the reduced ambient oxygen partial pressure, the air compressor needs more time to establish sufficient supply pressure to maintain high power output, resulting in a prolonged loading time. The increased voltage transient deviation stems from the intensified cathode oxygen starvation in the initial stage of loading transients, causing a larger transient deviation of the stack voltage relative to the steady-state value, manifested as voltage undershoot and subsequent recovery / rebound. The coupling of these three physical effects forms a dynamic fingerprint of altitude degradation, which has not yet been utilized as a diagnostic feature for altitude degradation in current technologies.

[0016] In contrast, internal flooding faults are characterized by a simultaneous increase in loading and unloading times and a continuous voltage drop, without exhibiting the characteristic of "shortened unloading." Membrane dryness faults are characterized by an increased slope in the ohmic region and an increase in high-frequency impedance; the change in dynamic response time does not exhibit the aforementioned bidirectional asymmetry. Therefore, this invention achieves specific identification of reversible altitude degradation by using a reverse combination of three parameters: "the rate of change of loading time is greater than or equal to a first threshold, the rate of change of unloading time is less than or equal to a negative second threshold, and the transient voltage deviation is greater than or equal to a third threshold," effectively eliminating interference from internal faults. To increase diagnostic margin and avoid boundary misjudgments, the third threshold is preferably set to 6.5% (with an allowable range of 6% to 10%) to ensure sufficient reliability even when the measured transient voltage deviation approaches the threshold.

[0017] To facilitate engineering implementation and logic encapsulation in the vehicle controller, this invention introduces the Degradation Index (DI) as an auxiliary concept in the specification. It is used to describe the comprehensive judgment logic after the three parameters of loading time change rate, unloading time change rate and voltage transient deviation are fused by threshold.

[0018] Compared with existing technologies, this invention has the following advantages. First, it utilizes the "bidirectional asymmetry" characteristic of dynamic response for altitude degradation diagnosis, overcoming the shortcomings of existing technologies that focus on internal fault diagnosis and struggle to identify external altitude degradation. This provides a new technical path for online condition monitoring of high-altitude PEMFCs. Second, it eliminates the need for additional hardware such as cell voltage monitoring modules, electrochemical impedance spectroscopy, or drainage pressure sensors. It can acquire the required signals using only existing battery management systems and vehicle controllers, resulting in lower implementation costs and easy integration into mass-produced vehicles. Third, it completes diagnosis within tens of seconds (the test process takes approximately 40 to 50 seconds) through a standard step load test, far faster than polarization curve tests or stability tests that require tens of minutes or even hours, meeting the real-time requirements of on-board online diagnostics. Fourth, by fusing three parameters—load time change rate, unload time change rate, and voltage transient deviation—this invention not only determines the presence of degradation but also quantifies the degradation level (mild / moderate / severe) based on the threshold range of each parameter, providing a decision-making basis for adaptive adjustment of the vehicle's energy management strategy. After diagnosing reversible degradation, this invention can further output a signal to trigger the pre-pressurization control of the air compressor. By increasing the air compressor speed in advance, it can overcome the lag in dynamic response at high altitudes, realize a "diagnosis-control" closed loop, and improve the vehicle's power and safety in high-altitude areas.

[0019] It should be noted that the preset time window is determined based on the typical voltage response time of a PEMFC system (the maximum transient deviation usually occurs within 1 to 3 seconds) and with sufficient margin. For other application scenarios or specially designed PEMFC systems, this time window can be adaptively adjusted according to the time from the start of a step jump to the first obvious inflection point in the voltage, for example, set to 2 to 3 times the nominal response time of the system. The adjusted window is still within the protection scope of this invention. Attached Figure Description

[0020] Figure 1 This is an overall flowchart of the method for diagnosing reversible performance degradation of the PEMFC system in a high-altitude environment, as described in this embodiment of the invention. Figure 2 This is a schematic diagram comparing the step load response of the system in this embodiment of the invention at sea level and at an altitude of 2000 m, showing the bidirectional asymmetric characteristics of increased loading time, shortened unloading time, and increased voltage transient deviation. Figure 3 This is a block diagram illustrating the calculation logic of the Degeneration Diagnostic Index (DI) in an embodiment of the present invention. Figure 4 This is a schematic diagram of the calibration mapping curves of the absolute values ​​of the loading time change rate and the unloading time change rate at different altitudes in an embodiment of the present invention; Figure 5This is a schematic diagram of the modular structure of the diagnostic system in an embodiment of the present invention.

[0021] It should be noted that D_V in the attached figures and text both represent the voltage transient deviation. Detailed Implementation

[0022] To make the objectives, features, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. Those skilled in the art will understand that the following embodiments are for illustrative purposes only and should not be considered as limiting the scope of the invention.

[0023] Example 1: Diagnostic Method for Reversible Performance Degradation of PEMFC System in High-Altitude Environment

[0024] This embodiment uses a vehicle-grade PEMFC stack with a rated net power of 80 kW and a peak stack power of 93 kW as the subject. Each cell contains 360 cells, with an active area of ​​300 cm². The proton exchange membrane is Nafion 211, and the catalyst is Pt / C. The catalyst loadings at both the cathode and anode are 0.4 mg·cm⁻². The test bench is equipped with an electronic load ranging from 0 to 150 kW and an environmental chamber capable of simulating temperatures from −40 to 60 °C and altitudes from 0 to 5500 m. Figure 1 As shown, the overall process of the diagnostic method of the present invention includes: applying a standard step load command, synchronously acquiring voltage / power response, extracting voltage transient deviation / loading time / unloading time, calculating the rate of change by comparing with a sea level reference, performing a three-parameter reverse combination judgment, and outputting degradation level and pre-compensation signal. The following provides a detailed description of each step.

[0025] Standard step load tests were performed at 10% to 90% of rated power under standard sea-level conditions (101.3 kPa, 25 °C, 50% relative humidity, RH) and simulated 2000 m altitude conditions (79.5 kPa, 25 °C, 50% RH). A dynamic load response procedure applied a step change, simultaneously recording transient voltage and current; each transient was repeated five times, and the arithmetic mean and standard deviation were reported. The data acquisition system recorded the stack voltage and current at a sampling frequency of at least 10 Hz, and temperature, pressure, and flow rate at 1 Hz. Furthermore, the step load test lasted approximately 40 to 50 seconds (including loading, steady-state holding, and unloading), and was performed only when the vehicle was stationary or idling. The duration of the current surge to the stack was extremely short, far below the durability test conditions specified in GB / T 24554-2022, and therefore its impact on stack life was negligible.

[0026] The experimental results, shown in Table 1, reveal the "bidirectional asymmetry" characteristic of the dynamic response of PEMFCs in high-altitude environments. At an altitude of 2000 m, the system loading time of the 93 kW automotive PEMFC stack increased from approximately 24 s at sea level to approximately 32 s (an increase of approximately 33%), while the unloading time decreased from approximately 5 s at sea level to approximately 4 s (a decrease of approximately 20%). Simultaneously, the voltage transient deviation increased from approximately 5% to approximately 12%. This inverse relationship between the three parameters—"prolonged loading + shortened unloading + increased transient deviation"—is a crucial diagnostic feature of altitude-induced reversible degradation. The data in the table are the arithmetic mean of five repeated tests, with the standard deviation in parentheses.

[0027] Table 1 Comparison of dynamic response parameters at sea level and 2000 m altitude (mean ± standard deviation) System loading time (10%→90% rated power) 24 s (±1.2 s) 32 s (±1.5 s) +33.3% System unloading time (90%→10% rated power) 5 s (±0.3 s) 4 s (±0.2 s) −20.0% Voltage transient deviation D_V 5%（±0.8%） 12%（±1.2%） +140% As shown in Table 1, at an altitude of 2000 m, the system loading time is significantly prolonged, while the unloading time is shortened, and the voltage transient deviation increases substantially. Experiments indicate that there is a difference of approximately 20 times between the system loading time (24 s) and the 10% to 90% rise time of the stack voltage (approximately 1.2 s, independently measured by high-frequency voltage sampling), suggesting that the gas supply subsystem is the main bottleneck in the transient response. The shortened unloading time is attributed to the lower ambient back pressure reducing the exhaust resistance at the cathode outlet, while the adjustment of the back pressure valve opening further reduces the gas venting resistance on the cathode side, jointly accelerating the depressurization process during load shedding. This combination of three inverse changes—"prolonged loading + shortened unloading + increased transient deviation"—constitutes an important diagnostic characteristic of altitude-induced reversible degradation.

[0028] Based on the above findings, the diagnostic method of this embodiment is specifically implemented as follows: Step (1): Through the vehicle controller or test bench electronic load, a standard step load command is applied to the PEMFC system, which increases from 10% of the rated power to 90% of the rated power and then drops back to 10% of the rated power after reaching a steady state. Simultaneously, the total stack voltage V_stack(t) and load power P(t) are collected at a sampling frequency of not less than 10 Hz. When implemented on a vehicle, this virtual load request is not used to drive the wheels to output actual traction torque, and its corresponding electrical energy is absorbed by the power battery, DC / DC converter, or vehicle energy consumption unit; when implemented on a test bench, its corresponding electrical energy is absorbed by the electronic load.

[0029] Step (2): Extract the voltage transient deviation D_V from the voltage curve. The calculation formula is Equation (1): D_V = max|V_stack(t) − V_ss| / V_ss × 100% (1), where t is located in [t0, t0+Tw], t0 is the rising edge of the step load command, Tw is the 5-second time window, and V_ss is the arithmetic mean of the voltage samples taken for 10 consecutive seconds after the step load reaches steady state; extract the actual loading time t_load (the time when the power increases from 10% to 90%) and the actual unloading time t_unload (the time when the power decreases from 90% to 10%) from the power curve. Regarding steady state determination, the following algorithm is adopted in this embodiment: when the power values ​​of 100 consecutive sampling points (based on a sampling frequency of 10 Hz, i.e., 10 consecutive seconds) all fall within ±5% of the target power, the system is determined to enter steady state. The timing starts from the first sampling point that meets the condition, and 10 consecutive seconds are collected as the steady state window.

[0030] Step (3): Read the pre-stored sea level reference values ​​t_load,SL = 24 s and t_unload,SL = 5 s from the non-volatile memory of the vehicle controller or battery management system, compare the measured loading time with the sea level reference value, and calculate the loading time change rate according to formula (2): Δt_load = (t_load − t_load,SL) / t_load,SL ×100% (2); compare the measured unloading time with the sea level reference value, and calculate the unloading time change rate according to formula (3): Δt_unload = (t_unload − t_unload,SL) / t_unload,SL × 100% (3).

[0031] The aforementioned sea-level reference values ​​and threshold parameters were pre-obtained using the following calibration method: Under standard atmospheric pressure conditions at sea level (101.3 kPa, 25 °C, 50% RH), at least three PEMFC systems of the same model underwent at least 10 standard step load tests. The loading time, unloading time, and voltage transient deviation of each test were calculated. The arithmetic mean and standard deviation of the multiple test results for each parameter were taken, and the upper limit of the normal fluctuation range (mean + 3 times the standard deviation) was taken as the basic threshold. This threshold was then multiplied by a safety factor of 1.2 to 1.5 to obtain the final threshold. In the embodiment, the first threshold Th1 = 15%, the second threshold Th2 = 8%, and the third threshold Th3 = 6.5% were obtained by calibrating a 93 kW-class fuel cell stack using this method. The above calibration method is applicable to PEMFC systems with rated power ranging from 50 kW to 150 kW. Other power levels can be recalibrated by linearly scaling the reference time and following the same statistical process. After calibration, the baseline and threshold values ​​are permanently stored in the non-volatile memory of the vehicle controller in a lookup table format. These values ​​are directly read and retrieved during actual vehicle operation, eliminating the need for repeated calibration tests. In degradation level classification applications, the second threshold is preferably calibrated within the range of 8% to 15%. When the second threshold calibration value exceeds 15%, the classification boundary is regenerated using multi-altitude calibration data to maintain consistency between the classification boundary and the degradation judgment threshold.

[0032] Step (4): Set the first threshold Th1 = 15% (allowable range 15% to 25%), the second threshold Th2 = 8% (allowable range 8% to 20%), and the third threshold Th3 = 6.5% (allowable range 6% to 10%). When Δt_load ≥ +15%, Δt_unload ≤ −8%, and D_V ≥ 6.5% are simultaneously satisfied, it is determined that the system has reversible performance degradation caused by altitude increase, and a diagnostic result is output to characterize whether the reversible performance degradation exists. In this embodiment, the measured values ​​at an altitude of 2000 m are t_load = 32 s, t_unload = 4 s, and D_V = 12%. Calculations show that Δt_load = +33.3% > Th1, |Δt_unload| = 20.0% > Th2, and D_V = 12% > Th3. All three conditions are satisfied simultaneously, therefore it is determined that the system has altitude-induced reversible degradation. In contrast, if the system experiences an internal flooding fault, it is usually characterized by a simultaneous increase in loading and unloading time and a continuous drop in voltage, without exhibiting the characteristic of "shortened unloading". If a membrane dryness fault occurs, it is usually characterized by an increase in the slope of the ohmic region and an increase in high-frequency impedance, and the change in dynamic response time does not exhibit the aforementioned bidirectional asymmetry.

[0033] like Figure 3As shown, the calculation logic of the Degradation Diagnostic Index (DI) integrates three parameters—the rate of change of loading time, the rate of change of unloading time, and the voltage transient deviation—through threshold fusion to form a comprehensive judgment index. Degradation level determination is triggered only when all three conditions are simultaneously met. It is worth noting that the specific numerical ranges of the first, second, and third thresholds mentioned above are preferred embodiments for PEMFC systems with rated power ranging from 80 kW to 100 kW. For systems with different rated power levels, these ranges can be obtained through recalibration using a sea-level benchmark test adapted to that model. When removing outliers, statistical methods such as the Grubbs criterion or the Chauville criterion can be used to identify and remove test data that significantly deviates from the normal distribution.

[0034] Example 2: Quantitative assessment of degradation level and equivalent altitude

[0035] Building upon Example 1, this example further establishes a quantitative mapping relationship between degradation level and equivalent altitude. In this example, standard step load tests were repeatedly performed at simulated altitudes of 1000 m, 2000 m, and 3000 m to obtain multi-altitude calibration data, establishing the following grading criteria (Table 2). It should be noted that the dynamic response parameters under each altitude condition in the table below are obtained by rounding the arithmetic mean of five repeated tests to clearly show the grading boundaries; in actual bench tests, normal fluctuations within ±2% exist between different batches of the same type of fuel cell stack. If the loading time change rate and unloading time change rate correspond to different levels, the final degradation level can be determined according to a pre-calibrated two-dimensional lookup table rule; when no two-dimensional lookup table rule is set, the higher level is taken as the final degradation level.

[0036] Table 2. Criteria for Degradation Level and Equivalent Altitude Classification Mild degeneration ≥ +15% and < +30% ≥ −15% and ≤ −8% Approximately 1000 m to less than 2000 m Moderate degeneration ≥ +30% and < +50% >−35% and <−15% Approximately 2000 m to less than 3000 m Severe degeneration ≥ +50% ≤ −35% 3000 m and above (3000 m is the actual measurement point; above 3000 m is the trend extrapolation, which can be combined with the environmental pressure calibration of the vehicle barometer) The measured values ​​at an altitude of 2000 m, Δt_load = +33.3% and Δt_unload = −20.0%, fall within the "moderate degradation" range, corresponding to the second degradation level. This classification mapping relationship can be pre-calibrated and stored in the vehicle controller. During actual vehicle operation, the current degradation level can be quickly read through a standard step load test, and then the equivalent altitude can be obtained through table lookup mapping, providing a basis for adaptive adjustment of energy management strategies. To fully support the above classification criteria, this embodiment supplemented the standard step load test under the conditions of 1000 m (89.9 kPa) and 3000 m (70.1 kPa). The experimental results are summarized in Table 3. Table 3 Summary of Multi-Altitude Calibration Experiment Data sea ​​level 101.3 kPa 24 s 5 s 5% — — 1000 m 89.9 kPa 28 s 4.5 s 8% +16.7% −10.0% 2000 m 79.5 kPa 32 s 4 s 12% +33.3% −20.0% 3000 m 70.1 kPa 38 s 3.2 s 18% +58.3% −36.0% As shown in Table 3, the absolute values ​​of both the loading time change rate and the unloading time change rate monotonically increase with increasing altitude. The boundaries of the three degradation levels are determined by the threshold ranges in Table 2. At 1000 m, Δt_load = +16.7% falls into the mild degradation range (first degradation level), and Δt_unload = −10.0% also falls into the mild degradation range (−15% ​​to −8%). At 3000 m, Δt_load = +58.3% and Δt_unload = −36.0% fall into the severe degradation range (third degradation level), which is completely consistent with the classification criteria in Table 2.

[0037] The basis for determining the endpoint values ​​of the grading ranges in Table 2 is explained as follows: The three boundary values ​​for the loading time change rate (+15%, +30%, +50%) and the three boundary values ​​for the unloading time change rate (−8%, −15%, −35%) in Table 2 were determined by linear interpolation and empirical division of the measured data at the three calibration points of 1000 m, 2000 m, and 3000 m. Specifically, the measured loading time change rate of +16.7% at 1000 m is rounded down to +15% as the lower boundary for mild degradation (with a safety margin); the measured loading time change rate of +33.3% at 2000 m is taken as +30% according to the engineering margin as the lower boundary for moderate degradation; and the measured loading time change rate of +58.3% at 3000 m is taken as +50% according to the engineering margin as the lower boundary for severe degradation. The boundary values ​​for the unloading time change rate are determined according to the same principle. For data near boundary values ​​(such as loading time change rate in the range of +28% to +32% or unloading time change rate in the range of -17% to -13%), since they may be in the boundary area between two degradation levels, the environmental pressure directly measured by the vehicle barometer can be used for auxiliary judgment, or the information from both the loading time and unloading time dimensions can be cross-confirmed by combining the two dimensions of loading time and unloading time through two-dimensional lookup table rules.

[0038] To further clarify, Tables 2 and 3 show the range of severe degradation corresponding to 3000 m and above. 3000 m is the actual measured calibration point, while the range above 3000 m is the trend extrapolation result based on the three calibration points of 1000 m, 2000 m, and 3000 m. Due to experimental limitations in this application, no actual bench tests were conducted above 3000 m. The equivalent altitude mapping result for this range was cross-validated and corrected using environmental pressure directly measured by a vehicle-mounted barometer. In this embodiment, the second threshold is set to 8% (the lower limit of the allowable range of 8% to 20%). At an altitude of 1000 m, |Δt_unload| = 10% strictly meets the ≥8% criterion, ensuring that mild degradation can be correctly detected.

[0039] The loading and unloading times in Table 3 for each altitude condition are the arithmetic mean of five repeated tests. The standard deviation of loading time is ≤1.5 s, the standard deviation of unloading time is ≤0.3 s, and the standard deviation of voltage transient deviation is ≤1.2%. The rate of change values ​​in the table are rounded to one decimal place. In actual testing, there may be fluctuations within ±2% for different individual fuel cell stacks and environmental conditions.

[0040] The following details the two-dimensional lookup table rules when the loading time change rate and unloading time change rate correspond to different degradation levels. Table 4 provides a specific example of a two-dimensional decision matrix, which has rows for the loading time change rate dimension and columns for the unloading time change rate dimension, with the matrix elements representing the final degradation level.

[0041] Table 4 Example of a two-dimensional degradation level determination matrix Matrix rows Loading time change rate dimension: <+15% is below the threshold; ≥+15% and <+30% is mild; ≥+30% and <+50% is moderate; ≥+50% is severe. Matrix column The absolute value of the rate of change in unloading time is as follows: <8% is below the threshold; 8% to 15% is mild; >15% and <35% is moderate; ≥35% is severe. Matrix element rules When two dimensions fall into the same level, that level is taken directly; when they fall into different levels, the higher level is taken based on the rate of change over loading time; when only one dimension reaches the threshold, reversible degradation is not directly determined, and repeated testing or confirmation by environmental pressure measured by an on-board barometer is required. The calibration method for the above two-dimensional judgment matrix is ​​as follows: Standard step load tests are performed at three calibration points: 1000 m, 2000 m, and 3000 m, to obtain (Δt_load, Δt_unload) measured data pairs. Based on the distribution of the measured data pairs on the two-dimensional plane, and using the single-dimensional threshold intervals in Table 2 as a basis, the two-dimensional boundaries of each degradation level are delineated. Under the premise of satisfying the three-parameter judgment conditions described in claim 1, when two dimensions fall into the same level interval, that level is directly taken as the final judgment result; when two dimensions fall into different level intervals, the loading time change rate dimension is used as the main basis (because it has a stronger monotonic correlation with altitude), while the unloading time change rate dimension is referenced for correction, and the higher of the two corresponding levels is taken as the final judgment result. If only one dimension of the loading time change rate or unloading time change rate reaches the threshold, it is not directly judged as reversible degradation, and repeated testing or auxiliary confirmation using information such as environmental pressure measured by a vehicle-mounted barometer is required; if the three-parameter judgment conditions of claim 1 are not met, it is judged as altitude-inducible reversible degradation.

[0042] The visual mapping relationship of the above-mentioned hierarchical criteria is as follows: Figure 4 As shown. Figure 4 Only three measured calibration points at 1000 m, 2000 m, and 3000 m are plotted. For areas above 3000 m, they are only used as trend extrapolation intervals. In engineering applications, the environmental pressure can be directly measured by a vehicle-mounted barometer for cross-validation.

[0043] It is important to note that this diagnostic method is applicable to PEMFC systems within the normal aging range. When the system exhibits severe irreversible aging (such as severe catalyst sintering, severe proton exchange membrane degradation, bipolar plate corrosion and perforation), these aging conditions may couple with the dynamic response characteristics of altitude degradation, thus affecting diagnostic accuracy. For example, severe catalyst aging itself can lead to prolonged loading time, which may overlap with the characteristics of altitude degradation. Therefore, it is recommended to update the sea-level baseline value regularly during the system's service life (e.g., every 5000 km or every 6 months) to track the system's normal aging trend and eliminate interference from aging factors on altitude degradation diagnosis. For systems that have exceeded their design life or have known serious faults, it is recommended to repair or replace the system before performing this diagnostic method.

[0044] Example 3: On-board in-situ diagnostics and air compressor pre-compensation control

[0045] This embodiment illustrates a specific implementation of the present invention in an on-vehicle environment. Each time the fuel cell vehicle is powered on, the vehicle controller automatically triggers a standard step load self-test: when the vehicle is stationary or idling, the vehicle controller sends a virtual load request from 10% to 90% of the rated power to the fuel cell stack. This virtual load request is not output to the drive motor to generate actual traction torque; the corresponding electrical energy is absorbed by the power battery, DC / DC converter, or on-board energy consumption unit. In bench testing scenarios, it is absorbed by the electronic load of the test bench. The vehicle controller continuously collects fuel cell stack voltage and power data, calculates the loading time, unloading time, and voltage transient deviation, and compares them with a pre-stored sea-level reference value.

[0046] When the degradation is determined to be moderate (e.g., equivalent altitude of 2000 m, corresponding to the second degradation level), the vehicle controller outputs a signal carrying degradation level information to trigger air compressor pre-boost control. Upon receiving this signal, the air compressor controller increases the engine speed to 85% of the target steady-state speed 3 to 5 seconds before the next actual power request, compensating for the extra time required for the air compressor to build pressure in high-altitude environments. The target speed ratio for pre-boost is determined according to the degradation level: 80% for mild degradation (first degradation level), 85% to 90% for moderate degradation (second degradation level), and 90% to 95% for severe degradation (third degradation level). This pre-compensation strategy can reduce the system loading time at 2000 m altitude from 32 seconds to approximately 26 to 28 seconds, significantly improving the vehicle's high-altitude dynamic response.

[0047] This embodiment fully illustrates the engineering value of the diagnostic-control closed loop of the present invention: the diagnostic process requires no additional sensors and can be achieved using only the existing battery management system (BMS) and vehicle control unit (VCU); control compensation triggers the existing air compressor actuator through the output signal, requiring no hardware modification. The overall implementation cost of the solution is low, but it can significantly improve the operating quality and safety of fuel cell vehicles in high-altitude areas.

[0048] Example 4: Verification of differentiation from internal faults

[0049] To verify the ability of this invention to distinguish between altitude degradation and internal faults, this embodiment simulated three typical operating conditions at an altitude of 2000 m: (a) Normal altitude degradation (no internal faults): The environmental chamber was set to 79.5 kPa, 25 °C, and 50% RH, and the system operated normally; (b) Slight cathode flooding: Based on (a), the opening frequency of the cathode drain valve was reduced to 50% of the normal value, causing liquid water to accumulate in the cathode flow channel, simulating the industry-standard flooding condition; (c) Membrane drying: Based on (a), the humidifier outlet temperature was reduced from 75 °C to 50 °C, causing the membrane water content to decrease, simulating the industry-standard membrane drying condition. The simulation conditions for (b) and (c) were set according to the recommended method for fault simulation in GB / T 24554-2022. The dynamic response characteristics under the three operating conditions are compared in Table 5.

[0050] Table 5. Verification of the distinction between altitude degradation and internal faults (a) Normal altitude degradation +33.3% −20.0% 12% Reversible degradation (moderate / secondary) (b) Slight flooding of the cathode +18.0% +25.0% 3% The judgment conditions are not met. (c) Membrane drying +8.0% +5.0% 6% The judgment conditions are not met. As shown in Table 5, both flooding and membrane drying internal faults exhibited a simultaneous increase in loading and unloading times (both with positive rates of change), and the voltage transient deviation did not significantly increase. Therefore, they did not meet the three-parameter reverse combination judgment conditions of this invention (loading time change rate ≥ first threshold, unloading time change rate ≤ negative second threshold, voltage transient deviation ≥ third threshold), and were thus correctly excluded. This fully demonstrates the specific identification capability of the diagnostic logic of this invention for altitude degradation.

[0051] Example 5: Computer Program Implementation

[0052] In one software implementation, the above diagnostic method can be implemented in a controller or processor via a computer program. When the computer program is executed by the processor, it performs the following functions, and each functional module corresponds to the method content in claim 1: (1) reading the dynamic response parameters of the sea level reference (load time, unload time reference values) and the first threshold, second threshold, and third threshold obtained by calibration, corresponding to the reference reading and rate of change calculation in claim 1; (2) receiving the controlled load change command triggered by the controller and collecting real-time data of the total voltage of the fuel cell stack and the load power, corresponding to the data acquisition in claim 1; (3) calling the feature extraction module to calculate the voltage transient deviation, the actual load time, and the actual unload time, corresponding to the feature extraction module in claim 1. (4) Call the comparison calculation module to calculate the loading time change rate and unloading time change rate, corresponding to the change rate calculation in claim 1; (5) Call the degradation judgment module and the diagnostic result output module to perform the three-parameter threshold combination condition judgment and output the diagnostic result used to characterize whether reversible performance degradation exists; when a graded evaluation is required, further output the degradation level and equivalent altitude, corresponding to the degradation judgment in claim 1 and the grade judgment in claim 6; (6) If it is determined to be reversible degradation, output a signal carrying degradation level information to trigger the air compressor pre-pressurization control, corresponding to the pre-pressurization control in claim 7.

[0053] The processor can be a general-purpose processor (such as a central processing unit CPU), a digital signal processor (DSP), an application-specific integrated circuit (ASIC), or a field-programmable gate array (FPGA). The computer program can be implemented using programming languages ​​such as Python, MATLAB, C / C++, or Simulink, and can be deployed in the form of an on-board controller software upgrade package, firmware package, or installation package. As long as the computer program can implement the diagnostic method described in this invention when executed by the processor, it falls within the protection scope of this invention.

[0054] Example 6: Specificity verification of diagnostic criteria

[0055] To verify the specificity of the three-parameter reverse combination judgment condition proposed in this invention for reversible degradation caused by altitude increase, this embodiment systematically completed a control experiment on various common interference factors under standard atmospheric pressure at sea level. Specifically, while keeping the ambient pressure constant at 101.3 kPa, the following operating parameters were artificially changed: (i) the ambient temperature was stepped within a range of 15 °C to 35 °C in increments of 5 °C; (ii) the intake relative humidity was adjusted within a range of 30% to 80% in increments of 10%; (iii) a slight gas starvation condition was simulated by slightly limiting the cathode intake flow rate; and (iv) a random power disturbance of ±5% was applied during the load step process to simulate load fluctuations in actual road conditions. Each of the above interference conditions was repeated at least 5 times according to the same standard step load procedure as in Example 1, and the loading time change rate, unloading time change rate, and voltage transient deviation were calculated.

[0056] Experimental results show that under all the above-mentioned disturbance conditions, no situation occurred where the three-parameter inverse combination condition was simultaneously satisfied. Specifically: temperature disturbances and humidity changes mainly affect steady-state performance, with an impact on transient loading / unloading time not exceeding ±5%, and the unloading time change rate is always positive or close to zero; slight gas starvation leads to a prolonged loading time, but also a prolonged unloading time, failing to satisfy the inverse characteristic of unloading shortening; random power disturbances cause each parameter to fluctuate slightly around the baseline value, but the fluctuation amplitude does not reach any threshold. The above negative control results indicate that the three-parameter inverse combination condition of prolonged loading + shortened unloading + increased transient deviation proposed in this invention has good environmental specificity within the test range of this embodiment, and can be used to identify reversible degradation of low pressure caused by altitude increase, while common disturbance factors such as temperature, humidity, gas starvation, and load fluctuations are unlikely to form the same response as this diagnostic feature combination.

[0057] The above-described specificity verification experiments were conducted on the same 93 kW PEMFC system as in Example 1. Studies have shown that this specificity principle also applies to other models or power levels of systems, but the specific threshold parameters need to be recalibrated using a sea-level benchmark test appropriate for that model.

[0058] This invention achieves rapid, in-situ diagnosis of reversible performance degradation in high-altitude environments through standard step load testing. Experimental data verifies the reliability and specificity of the altitude-related degradation characteristic combination of prolonged loading, shortened unloading, and increased transient deviation. Adaptive adjustments to the threshold range, time window width, and pre-boost ratio, without departing from the technical principles of this invention, all fall within the scope of protection of this invention; the scope of protection of this invention is defined by the appended claims.

Claims

1. A method for diagnosing reversible performance degradation of a proton exchange membrane fuel cell system in a high-altitude environment, characterized in that, include: One or more controlled load change commands are applied to the proton exchange membrane fuel cell system to induce identifiable loading and unloading processes, and system voltage response signals and load power response signals are acquired. The voltage transient deviation is obtained from the voltage response signal, and the actual loading time and actual unloading time are obtained from the load power response signal. Preset reference loading time and preset reference unloading time are read, and the loading time change rate and unloading time change rate are calculated respectively. When the loading time change rate is greater than or equal to a preset first threshold, the unloading time change rate is less than or equal to a negative preset second threshold, and the voltage transient deviation is greater than or equal to a preset third threshold, it is determined that the proton exchange membrane fuel cell system exhibits reversible performance degradation caused by altitude increase, and a diagnostic result is output. The preset first threshold, preset second threshold, and preset third threshold are pre-calibrated based on benchmark test data of the same model of proton exchange membrane fuel cell system.

2. The diagnostic method according to claim 1, characterized in that, The one or more controlled load change commands include a standard step load command, which is a load change command that increases from a first preset power to a second preset power, and then falls back to the first preset power after reaching a steady state, wherein the first preset power is lower than the second preset power; wherein the first preset power is 10% of the rated power, the second preset power is 90% of the rated power, and the sampling frequency of the voltage response signal and the load power response signal is not lower than 10 Hz.

3. The diagnostic method according to claim 1 or 2, characterized in that, The voltage transient deviation D_V is calculated within a preset time window starting from the start time of the controlled load change command. The preset time window is 3 to 10 seconds. The voltage transient deviation D_V is the ratio of the maximum absolute deviation of the voltage sample value relative to the steady-state voltage average value within the preset time window to the steady-state voltage average value. Furthermore, the voltage sample value is subjected to a moving average filter, and the filter window width is the number of data points corresponding to 0.5 to 1.0 seconds.

4. The diagnostic method according to claim 1 or 2, characterized in that, The actual loading time is the time required from the start of power increase until the power first reaches the allowable deviation range of the target load power and remains in a steady state; the actual unloading time is the time required from the start of power decrease until the power first reaches the allowable deviation range of the target unloading power and remains in a steady state; the allowable deviation range is ±2% to ±5%, and the steady state determination time is 1 second to 10 seconds.

5. The diagnostic method according to claim 1, characterized in that, The preset first threshold is 15% to 25%, the preset second threshold is 8% to 20%, and the preset third threshold is 6% to 10%; wherein, the loading time change rate is greater than or equal to the preset first threshold to characterize the extension of loading time, the unloading time change rate is less than or equal to the negative preset second threshold to characterize the shortening of unloading time, and the voltage transient deviation is greater than or equal to the preset third threshold to characterize the increase of transient deviation.

6. The diagnostic method according to claim 5, characterized in that, Based on the simultaneous fulfillment of the following conditions: the loading time change rate is greater than or equal to a preset first threshold, the unloading time change rate is less than or equal to a negative preset second threshold, and the voltage transient deviation is greater than or equal to a preset third threshold, the reversible performance degradation is classified into mild, moderate, or severe according to the threshold range in which the loading time change rate and the unloading time change rate are located; if the degradation levels corresponding to the loading time change rate and the unloading time change rate are inconsistent, the final degradation level is determined according to the pre-defined two-dimensional lookup table rules, or the higher degradation level of the two corresponding levels is taken as the final degradation level.

7. The diagnostic method according to claim 6, characterized in that, When applied to vehicles, the controlled load change command is automatically triggered by the vehicle controller during power-on self-test and executed when the vehicle is stationary or idling. The controlled load change command is a virtual load request and does not drive the motor to output actual traction torque. The electrical energy generated is absorbed by the power battery, DC / DC converter, or on-board energy consumption unit. When reversible performance degradation is determined, a pre-boost control signal carrying degradation level information is output, so that the air compressor is boosted to 80% to 95% of the target steady-state speed before the actual power request arrives.

8. A diagnostic system for reversible performance degradation of a proton exchange membrane fuel cell system in a high-altitude environment, characterized in that, include: The data acquisition module is used to apply one or more controlled load change commands to the proton exchange membrane fuel cell system to cause the proton exchange membrane fuel cell system to produce identifiable loading and unloading processes, and to acquire system voltage response signals and load power response signals; The feature extraction module is used to obtain the voltage transient deviation, actual loading time, and actual unloading time. The benchmark storage module is used to store the preset benchmark loading time, preset benchmark unloading time, preset first threshold, preset second threshold, and preset third threshold. The comparison calculation module is used to calculate the rate of change of loading time and the rate of change of unloading time. The degradation determination module is used to determine that there is reversible performance degradation caused by altitude increase when the loading time change rate is greater than or equal to a preset first threshold, the unloading time change rate is less than or equal to a negative preset second threshold, and the voltage transient deviation is greater than or equal to a preset third threshold. The diagnostic results output module is used to output diagnostic results.

9. The diagnostic system according to claim 8, characterized in that, It also includes a level determination module and a pre-boost control signal output module; the level determination module is used to determine the mild, moderate or severe degradation level based on the threshold range of the loading time change rate and the unloading time change rate; the pre-boost control signal output module is used to output a pre-boost control signal carrying degradation level information when the performance is determined to be reversible, so that the air compressor can be boosted to 80% to 95% of the target steady-state speed before the actual power request arrives.

Images