A multi-channel in-situ electrochemical diagnostic device for AEM electrolyzers and a synergistic early warning method for membrane degradation and catalyst loss based on a single baseline.
By using a multi-channel in-situ electrochemical diagnostic device and single-chip baseline calibration technology, the problem of multi-channel measurement in alkaline environment for online diagnosis of AEM electrolyzers has been solved, enabling accurate early warning and local protection of faults, and improving the operational reliability and lifespan of the electrolyzer.
Patent Information
- Application Number
- CN202611559399.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-10-09
- Publication Date
- 2026-11-17
AI Technical Summary
Existing AEM electrolyzer diagnostic technology cannot achieve online real-time diagnosis of membrane chemical structure evolution and catalyst loss, and it is difficult to perform multi-channel in-situ electrochemical measurements in an alkaline environment, resulting in high false alarm and false alarm rates, and failing to effectively extend the electrolyzer life.
A multi-channel in-situ electrochemical diagnostic device is designed, employing an alkali-resistant isolation amplification unit and an independent reference electrode. Combined with single-cell baseline calibration and dynamic threshold discrimination, it achieves single-cell-level electrochemical parameter measurement and fault early warning. Through hardware-algorithm collaboration, the false alarm rate is reduced, and local regulation protection is implemented.
It effectively reduced the false alarm rate and false negative rate of fault diagnosis, extended the overall service life of the AEM stack, and suppressed fault development through local control strategies, thereby improving the operational reliability and durability of the electrolyzer.
Smart Images

Figure CN123108650A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of anion exchange membrane (AEM) water electrolysis hydrogen production technology, and particularly relates to an online electrochemical diagnostic hardware device suitable for AEM electrolyzer stacks and a method for early warning of synergistic faults of membrane degradation and catalyst loss based on single-chip baseline calibration. Background Technology
[0002] Alkaline ion exchange (AEM) water electrolysis technology combines the low-cost advantages of alkaline electrolysis with the high-efficiency design of proton exchange membrane electrolysis. It can use non-precious metal catalysts and is considered an important technological route for next-generation green hydrogen production. However, the durability of AEM electrolyzers is a key bottleneck restricting large-scale commercialization. AEM electrolyzers operate in an alkaline environment, and their failure mechanism has significant unique characteristics: quaternary ammonium groups in the anion exchange membrane are prone to chemical degradation under strongly alkaline and oxidizing anolytical conditions, leading to decreased hydroxide ion conductivity and increased membrane resistance; simultaneously, deterioration of membrane mechanical strength may induce micropores and cracks, causing hydrogen-oxygen transpermeation mixing; the strong oxidizing environment of the anolyte oxidation reaction causes the non-precious metal catalyst to dissolve, agglomerate, and detach, and the support corrodes, gradually leading to the decay of electrode activity.
[0003] Existing AEM electrolyzer diagnostic technologies suffer from several key shortcomings. First, online diagnostic methods for AEM membrane degradation and catalyst loss are scarce, relying mostly on offline characterization using XPS, TOF-SIMS, and SEM after shutdown and disassembly. This fails to capture real-time information on membrane chemical structure evolution and catalyst loss during normal, loaded operation of the electrolyzer. Second, existing diagnostic schemes use a uniform, fixed threshold for fault determination across cells; however, due to differences in membrane type, thickness, catalyst type and loading, and assembly conditions, the initial electrochemical parameters of different AEM stacks exhibit significant dispersion, inevitably leading to false alarms and missed alarms with a uniform threshold. Third, existing hardware solutions struggle to achieve multi-channel in-situ impedance measurement in strongly alkaline electrolyte environments. The high conductivity and corrosiveness of alkaline electrolytes, along with electrochemical crosstalk between multiple individual cells within the stack, constitute unique hardware challenges for online AEM electrolyzer diagnostics; these crosstalk and corrosion problems are not present in the design of PEM electrolyzer diagnostic devices.
[0004] Search results indicate that existing AEM diagnostic patents primarily focus on diagnostic methods or algorithms. For example, methods for early anomaly diagnosis of AEM electrolysis systems based on fault map learning, and health diagnostic schemes for AEM electrolyzers that construct feature vectors from multi-source operating data to output degradation patterns, both emphasize methods or backend data analysis. However, existing technologies lack clear guidance on how to achieve stable and reliable multi-channel in-situ electrochemical measurement hardware integration in the alkaline electrolyte environment of an AEM electrolyzer. Summary of the Invention
[0005] To address the shortcomings of existing technologies, the first objective of this invention is to provide a multi-channel in-situ electrochemical diagnostic hardware device suitable for the alkaline operating environment of AEM electrolyzers, solving the technical problems of alkaline electrolyte corrosion protection and multi-cell crosstalk isolation. The second objective is to provide a synergistic early warning method for membrane degradation and catalyst loss based on single-cell baseline calibration. Through hardware-algorithm collaboration, the false alarm and false negative rates of fault diagnosis are reduced, and local regulation and protection at the single-cell level are achieved, thereby improving the overall service life of the AEM stack.
[0006] To achieve the above-mentioned objectives, the first aspect of this invention provides a multi-channel in-situ electrochemical diagnostic device for an AEM electrolyzer, comprising: a multi-channel isolated acquisition front-end, including N independent measurement channels, where N equals the number of single cells in the stack; each channel includes: - an alkali-resistant isolation amplification unit, employing an alkali-resistant epoxy-encapsulated differential amplification circuit, with an input stage using a high-impedance instrumentation amplifier and a common-mode rejection ratio of not less than 100dB, used to acquire the single-cell voltage response in an alkaline electrolyte environment; - a channel selection switch matrix, used to connect selected single cells to the excitation acquisition circuit in diagnostic mode, and to electrically isolate all channels from the stack in normal mode; - an independent reference electrode interface, each channel configured with an independent quasi-reference electrode connection terminal, used for single-cell level electrochemical potential measurement; an alkaline-compatible in-situ excitation unit, connected to the multi-channel isolated acquisition front-end, used to selectively output an electrochemical impedance spectroscopy (EIS) excitation signal or an open-circuit voltage (OCV) test excitation signal to selected single cells; an alkali-resistant electrical interlock switch is provided between the excitation output terminal of the in-situ excitation unit and the main rectifier power supply unit of the stack; A single-cell voltage monitoring unit is connected to the voltage detection terminal of each individual cell inside the stack, used to collect the operating voltage of each individual cell in real time; a host computer fault analysis and early warning unit has a built-in decoupling discrimination model based on single-cell baseline calibration, and receives impedance spectrum data and OCV attenuation data from the multi-channel isolated acquisition front end, single-cell voltage timing data from the single-cell voltage monitoring unit, and operating parameters from the operating condition acquisition submodule; the host computer also establishes signal interaction with the electrolyzer main control system.
[0007] Furthermore, the alkali-resistant isolation amplification unit is powered by an isolation power supply with an isolation voltage of not less than 2.5kV; the two sides of the isolation barrier are referenced by the potential of the battery pack electrolyte and the ground of the host computer, respectively, to eliminate the common-mode interference of the alkaline electrolyte circuit on the measurement circuit.
[0008] Furthermore, the alkali-resistant electrical interlock switch employs a sealed relay with nickel-plated or platinum-plated contacts, the contact portion of which is encapsulated in alkali-resistant epoxy resin, providing a mechanical life of not less than 10 years. 5Second; used to momentarily disconnect the main rectifier output in diagnostic mode and immediately restore power after the OCV test is completed.
[0009] Furthermore, the multi-channel isolation acquisition front-end is equipped with an inter-channel crosstalk suppression circuit between the channels; including a common-mode choke connected in series at the differential input terminal of each channel, and a shielding isolation layer between adjacent channels, so that the inter-channel crosstalk attenuation is not less than 60dB.
[0010] Furthermore, the host computer fault analysis and early warning unit is configured to execute a cell-by-cell baseline calibration process: after the electrolyzer is put into operation for the first time or after maintenance is completed, EIS test and OCV decay test are performed on each cell of the stack in sequence; the initial charge transfer resistance Rct0, initial membrane ion conduction resistance R_ion0, and initial open circuit voltage decay rate dV / dt0 of the cell are recorded; and the above parameters are bound to the cell number and calibration conditions and stored in the historical database.
[0011] Furthermore, the host computer is configured to independently calculate the normalized change for each single cell in subsequent online diagnostics: ΔRctᵢ=(Rctᵢ−Rct0ᵢ) / Rct0ᵢ; ΔR_ionᵢ=(R_ionᵢ−R_ion0ᵢ) / R_ion0ᵢ; Δ(dV / dt)ᵢ=[(dV / dt)ᵢ−(dV / dt0)ᵢ] / (dV / dt0)ᵢ; where the subscript i represents the i-th single cell.
[0012] Furthermore, the host computer is configured to: perform 3-5 verification diagnostics on each single cell after the baseline calibration is completed; collect the natural fluctuation data of ΔRctᵢ, ΔR_ionᵢ, and Δ(dV / dt)ᵢ of the single cell within the normal operating range; calculate the standard deviation of the natural fluctuation of each parameter σ_Rctᵢ, σ_Rionᵢ, and σ_dVdtᵢ; and set the dynamic threshold specific to the single cell accordingly.
[0013] Furthermore, the host computer is configured to output cell-by-cell fault identification results based on the combination mode of normalized change: (1) When ΔR_ionᵢ≥3σ_Rionᵢ and ΔRctᵢ<3σ_Rctᵢ, the cell is determined to be dominated by membrane degradation; (2) When ΔRctᵢ≥3σ_Rctᵢ and ΔR_ionᵢ<3σ_Rionᵢ and |Δ(dV / dt)ᵢ|<2σ_dVdtᵢ, the cell is determined to be dominated by catalyst loss; (3) When Δ(dV / dt)ᵢ≥3σ_dVdtᵢ and ΔR_ionᵢ≥3σ_Rionᵢ and |ΔRctᵢ|<2σ_Rctᵢ, the cell is determined to be dominated by mechanical damage and transpermeation. (4) When ΔR_ionᵢ≥3σ_Rionᵢ and ΔRctᵢ≥3σ_Rctᵢ are both true, the single cell coupling of the chip is determined to be degraded.
[0014] Furthermore, the host computer is configured to: when the fault diagnosis result of a single cell triggers an early warning, perform targeted control actions only on that single cell or the local battery pack containing that single cell, rather than uniformly reducing the load on the entire stack.
[0015] In a second aspect, this invention provides a method for synergistic early warning of membrane degradation and catalyst loss in an AEM electrolyzer based on the above-mentioned device, comprising the following steps: S1, baseline calibration per cell: After the electrolyzer is put into operation for the first time or after maintenance, EIS testing and OCV decay testing are performed on each single cell of the stack sequentially, and the Rct0ᵢ, R_ion0ᵢ, and dV / dt0ᵢ of the single cell are recorded; S2, natural fluctuation calibration per cell: 3-5 verification diagnostics are performed on each single cell; natural fluctuation data of ΔRctᵢ, ΔR_ionᵢ, and Δ(dV / dt)ᵢ within the normal operating range are collected; the standard deviations σ_Rctᵢ, σ_Rionᵢ, and σ_dVdtᵢ of each cell are calculated; S3, periodic multi-channel diagnostic triggering: During the normal hydrogen production operation of the electrolyzer, online diagnostics are triggered at preset time intervals; each single cell is sequentially connected to the excitation acquisition circuit through a channel selection switch matrix; at 10mHz-10kHz... EIS data is collected within the frequency range, and Rctᵢ and R_ionᵢ are extracted; OCV decay curves are collected within a 10-150s time window after the rectifier power supply is cut off, and dV / dtᵢ is calculated; S4, cell-by-cell normalization calculation and decoupling discrimination: the normalized change is calculated for each cell, and the fault type of the cell is determined according to the aforementioned discrimination logic; S5, cell-by-cell graded early warning and local control linkage: when membrane degradation is warned: the electrolyte temperature in the area where the cell is located is reduced by 2-5℃ for local temperature adjustment; when catalyst loss is warned: the current change rate is limited for the cell; when membrane mechanical damage and transpermeation are warned: the anode-cathode pressure difference is reduced and the diagnosis cycle of the cell is shortened; when coupling degradation is warned: the load reduction protection of the cell or the entire stack is interlocked and shut down is implemented; S6. Individual Cell Historical Data Tracking and Remaining Life Prediction: Store the diagnostic data of each cell in the historical database; plot the degradation trend curve with the individual baseline of the cell as a reference; predict the remaining service life of each cell; and issue a targeted maintenance prompt when the predicted remaining life of any cell is lower than the preset value. Detailed Implementation
[0016] In this embodiment, a 20-channel AEM electrolytic cell in-situ electrochemical diagnostic prototype was built; the AEM stack under test consisted of 20 single cells connected in series; the MEA used a non-precious metal NiFe-LDH anode catalyst; the anion exchange membrane used a quaternized polyaromatic AEM membrane; and the electrolyte was a 1 mol / L KOH alkaline solution. Hardware prototype parameters: The multi-channel isolated acquisition front-end of the diagnostic device uses an AD8253 high-impedance instrumentation amplifier for each channel's alkali-resistant isolation amplification unit; the entire circuit is encapsulated with alkali-resistant epoxy resin; the measured single-channel common-mode rejection ratio is ≥112dB; the isolation voltage of the isolation power supply is 2.7kV; the channel selection switch matrix uses rhodium-plated reed relays; common-mode chokes and metal shielding isolation layers are added between channels; COMSOL multiphysics simulation results show that under the condition of 1mol / L KOH high-conductivity electrolyte, the electrical crosstalk between channels can be attenuated by 67dB after adding the shielding isolation layer; compared with the traditional acquisition front-end without the added shielding layer (crosstalk attenuation is only 38dB), the parasitic crosstalk current caused by the alkaline electrolyte is reduced by about two orders of magnitude; simulation confirms that this hardware architecture can effectively suppress parasitic shunting interference formed between adjacent batteries through the electrolyte manifold. The in-situ excitation unit output frequency setting range is 10mHz~10kHz; AC disturbance amplitude is 8-40mV; the alkali-resistant electrical interlock relay is epoxy-encapsulated, and the measured mechanical life is greater than 1.2×10⁻⁶. 5 Second-rate.
[0017] Baseline calibration conditions were set as follows: current density 1.0 A·cm⁻²; electrolyte temperature 60℃; anode-cathode operating pressure difference 0.25 MPa; electrolyte 1M KOH. EIS and OCV baselines were sequentially acquired for 20 individual cells. The measured original baseline parameter distribution ranges for the 20 cells were: Rct0 = 0.27~0.44 Ω·cm²; R_ion0 = 0.085~0.155 Ω·cm²; dV / dt0 = −0.36~−0.54 mV / s. The baseline data shows that there is significant initial dispersion among different individual cells within the same stack; directly using a uniform fixed threshold for the entire stack inevitably leads to the risk of misjudgment. After calibration, four confirmatory diagnostic tests were performed to obtain the standard deviations σ_Rctᵢ, σ_Rionᵢ, and σ_dVdtᵢ for each cell. Taking cells No. 7, No. 12, and No. 18 as examples, the verification calibration results are as follows: Table 1. Standard deviation obtained from single-cell verification calibration (example prototype) sheet Long-term endurance operation test: The fuel cell stack was continuously operated for 820 hours; during the operation, a complete multi-channel online diagnostic was automatically triggered every 12 hours; at the same time, a comparative test group was set up: the comparative scheme adopted the traditional whole-stack unified fixed threshold early warning strategy; the fault diagnosis performance indicators of the two schemes were statistically analyzed. During the durability test, the membrane of cell #7 gradually underwent chemical degradation with increasing operating time. Online diagnostic data for cell #7 yielded the following parameters: R_ion = 0.170 Ω·cm², Rct = 0.302 Ω·cm², dV / dt0 = -0.41 mV / s. Calculations showed ΔR_ionᵢ = 0.092 (greater than 3σ_Rionᵢ = 0.072) and ΔRctᵢ = 0.029 (less than 3σ_Rctᵢ = 0.093). The system determined the degradation to be membrane degradation-driven. A local control command was issued to lower the electrolyte temperature in the flow channel region containing this cell from 60℃ to 56℃. The remaining 19 cells maintained their original operating conditions. After this localized cooling intervention, subsequent 200 hours of monitoring showed a significant slowdown in the rate of increase of the membrane's ion resistance, indicating that degradation was somewhat suppressed. During the later stages of durability testing, the NiFe catalyst at the anode of cell number 12 gradually dissolved and was lost. Diagnostic measurements showed that ΔRctᵢ=0.122>3σ_Rctᵢ; ΔR_ionᵢ=0.047<3σ_Rionᵢ; |Δ(dV / dt)ᵢ|=0.088<2σ_dVdtᵢ. This was determined to be a catalyst loss-driven degradation fault. The system was activated by limiting the current rise and fall rate of this single cell to prevent large step fluctuations in current and slow down further electrode failure. After 712 hours of operation, hydrogen-oxygen transperfusion occurred in cell number 18 due to microcracks in the membrane; Δ(dV / dt)ᵢ increased significantly; the judgment result indicated that the membrane was mechanically damaged and transperfused; the system immediately reduced the anode-cathode voltage difference and automatically shortened the diagnostic cycle of this cell to once every 4 hours for continuous and close monitoring; when two subsequent consecutive diagnoses confirmed that the membrane resistance and charge transfer resistance of this cell increased simultaneously, coupling degradation was determined, and the system implemented local load reduction protection to avoid direct emergency shutdown of the entire stack.
[0018] Diagnostic performance comparison test results: The dynamic threshold early warning scheme based on a single-chip baseline of this invention was compared with the traditional unified fixed threshold scheme for the entire stack. A total of 3 real single-chip fault events occurred within the 820-hour durability cycle; simultaneously, 14 instantaneous parameter disturbances occurred due to operating condition fluctuations (load increases / decreases, temperature disturbances). Table 2 compares the diagnostic effects of the two fault early warning strategies. sheet As can be seen from the test results in the table above: the traditional unified threshold scheme had 7 false alarms and 2 missed alarms; while the scheme of this invention had only 1 false alarm and no missed alarms; the overall false alarm rate decreased from 33.3% to 4.8%; proving that this scheme can significantly improve the reliability of fault warning for the discreteness problem of single cell in AEM stack; and through the local control protection of single cell, the actual endurance service life of the prototype stack is improved by about 17% compared with the reference control prototype.
[0019] Simulation life prediction verification: Based on the degradation trend curves of R_ion and Rct obtained from previous diagnostics, the host computer uses linear extrapolation to predict the remaining service life of a single cell; when the predicted remaining life is less than 250 hours, a maintenance prompt is output; in this 820-hour endurance test, the system outputs a maintenance prompt for the damaged No. 18 battery 272 hours in advance; this is consistent with the failure state analysis results after offline disassembly, confirming that the life prediction module can provide early maintenance warning prompts. Attached Figure Description Figure 1 is a schematic diagram of the connection architecture between the multi-channel isolated acquisition front-end and the AEM stack of the present invention; Figure 2 is a flowchart of the overall process of baseline calibration, verification calibration, and periodic online diagnosis of this invention. Figure 3 is a schematic diagram comparing the fault location capabilities of the chip-by-chip dynamic threshold diagnosis mode of the present invention and the traditional whole-stack unified threshold diagnosis mode. Figure 4 is a schematic diagram comparing the EIS Nyquist impedance spectra of the AEM electrolyzer of the present invention under three typical degradation states. Figure 5 is a schematic diagram comparing the open-circuit voltage decay curves of a single cell under different degradation states according to the present invention.
[0020] Beneficial effects: First, the hardware of this invention is specifically designed for the strongly alkaline KOH electrolyte conditions of AEM, employing an alkali-resistant epoxy-encapsulated isolation amplification unit, a common-mode choke between channels, and a shielding isolation layer. Simulation verification suppresses channel parasitic crosstalk caused by the electrolyte manifold, solving the hardware challenge of multi-channel in-situ EIS-OCV synchronous acquisition in alkaline systems; overcoming the limitation that existing PEM diagnostic equipment cannot be directly ported to AEM stacks. Second, it uses independent baseline calibration for each single cell + statistical analysis of the cell's historical operating fluctuations to obtain a dedicated dynamic threshold, replacing the traditional unified fixed threshold for the entire stack; effectively adapting to the discrete characteristics of initial electrochemical parameters between individual cells within an AEM stack; prototype durability tests confirm a significant reduction in false alarm and false negative rates. Third, a four-mode fault decoupling discrimination logic is established for the unique failure mechanism of AEM, which can distinguish four types of failure: membrane chemical degradation, anode catalyst loss, membrane micropore transpermeation, and membrane-electrode coupling degradation. Moreover, after a fault warning, it no longer simply shuts down the entire stack and reduces the load directly; it implements differentiated closed-loop protection strategies such as local temperature control, current change rate limiting, and differential pressure regulation for the faulty cell. While maintaining the continued operation of the remaining intact cells as much as possible, it suppresses the further development of the fault, which is conducive to extending the service life of the entire AEM stack. At the same time, it has built-in single-cell-level degradation trend tracking and remaining life prediction, providing quantitative reference for operation and maintenance.
Claims
1. A multi-channel in-situ electrochemical diagnostic device for an AEM electrolyzer, comprising an AEM electrolyzer stack and a main rectifier power supply unit, characterized in that, It also includes: a multi-channel isolated acquisition front-end, containing N independent measurement channels, where N equals the number of individual cells in the stack; each channel includes: - an alkali-resistant isolation amplification unit, employing an alkali-resistant epoxy-encapsulated differential amplification circuit, with a high-impedance instrumentation amplifier at the input stage and a common-mode rejection ratio of not less than 100dB; - a channel selection switch matrix, used to connect selected individual cells to the excitation acquisition circuit in diagnostic mode, and to electrically isolate all channels from the stack in normal mode; - an independent reference electrode interface, with each channel configured with an independent quasi-reference electrode connection terminal for single-cell-level electrochemical potential measurement; an alkaline-compatible in-situ excitation unit, connected to the multi-channel isolated acquisition front-end, used to selectively output electrochemical impedance spectroscopy (EIS) excitation signals or open-circuit voltage (OCV) test excitation signals to selected individual cells; an alkali-resistant electrical interlock switch is provided between the excitation output terminal of the in-situ excitation unit and the main rectifier power supply unit of the stack; and a single-cell voltage monitoring unit, connected to the voltage detection terminal of each individual cell inside the stack. The host computer fault analysis and early warning unit has a built-in decoupling discrimination model based on single-cell baseline calibration. The host computer fault analysis and early warning unit receives impedance spectrum data and OCV attenuation data from the multi-channel isolated acquisition front end, single-cell voltage timing data from the single-cell voltage inspection unit, and operating parameters from the operating condition acquisition submodule. The host computer also establishes signal interaction with the electrolyzer main control system. The decoupling discrimination model is configured to perform the following processing: (a) Perform baseline calibration process independently for each single cell of the stack: Under the preset calibration conditions, perform EIS test and OCV attenuation test on each single cell in sequence, and record the initial charge transfer resistance Rct0_i, initial membrane ion conduction resistance R_ion0_i, and initial open circuit voltage attenuation rate dV / dt0_i of the i-th single cell; (b) Perform 3-5 verification diagnostics on each single cell, and collect ΔRct_i, ΔR_ion_i, and Δ(dV / dt)_i of the single cell within the normal operating range. (c) Calculate the standard deviation of natural fluctuation data for each cell, σ_Rct_i, σ_Rion_i, and σ_dVdt_i; (d) Calculate the normalized change amount for each cell independently during the subsequent online diagnosis process: ΔRct_i=(Rct_i−Rct0_i) / Rct0_i; ΔR_ion_i=(R_ion_i−R_ion0_i) / R_ion0_i; Δ(dV / dt)_i=[(dV / dt)_i−(dV / dt0)_i] / (dV / dt0)_i; (d) Output the single cell level fault judgment result based on the combination mode of the normalized change amount for each cell: (1) When ΔR_ion_i≥3σ_Rion_i and ΔRct_i<3σ_Rct_i, the i-th cell is judged to be degraded mainly by membrane degradation;(2) When ΔRct_i≥3σ_Rct_i and ΔR_ion_i<3σ_Rion_i and |Δ(dV / dt)_i|<2σ_dVdt_i, the i-th single cell is determined to be dominated by catalyst loss degradation; (3) When Δ(dV / dt)_i≥3σ_dVdt_i and ΔR_ion_i≥3σ_Rion_i and |ΔRct_i|<2σ_Rct_i, the i-th single cell is determined to be transpermeation due to membrane mechanical damage; (4) When ΔR_ion_i≥3σ_Rion_i and ΔRct_i≥3σ_Rct_i are both true, the i-th single cell is determined to be coupled degradation; The host computer fault analysis and early warning unit is further configured to: when the fault judgment result of a single cell triggers an early warning, perform targeted control actions only on that single cell or the local battery pack containing that single cell, rather than uniformly reducing the load on the entire pack.
2. The AEM electrolytic cell multichannel in-situ electrochemical diagnostic device according to claim 1, characterized in that, The alkali-resistant isolation amplification unit is powered by an isolation power supply with an isolation voltage of not less than 2.5kV. The isolation barrier is referenced on both sides by the potential of the battery pack electrolyte and the ground of the host computer, respectively, to eliminate common-mode interference introduced by the alkaline electrolyte circuit.
3. The AEM electrolytic cell multichannel in-situ electrochemical diagnostic device according to claim 1, characterized in that, The alkali-resistant electrical interlock switch is a sealed relay with nickel-plated or platinum-plated contacts; the contact area is entirely encapsulated in alkali-resistant epoxy resin, with a mechanical life of not less than 10 years. 5 Next; in diagnostic mode, the main rectifier output is momentarily disconnected, and the main circuit power supply is restored after the OCV test is completed.
4. The AEM electrolytic cell multichannel in-situ electrochemical diagnostic device according to claim 1, characterized in that, The multi-channel isolation acquisition front end is equipped with an inter-channel crosstalk suppression circuit between each channel; a common-mode choke is connected in series at the differential input terminal of each channel, and a metal shielding isolation layer is set between adjacent channels; the inter-channel crosstalk attenuation is not less than 60dB, which is used to suppress parasitic electrical crosstalk formed through the alkaline electrolyte manifold.
5. The multi-channel in-situ electrochemical diagnostic device for AEM electrolytic cells according to claim 1, characterized in that, The host computer fault analysis and early warning unit also has a built-in remaining life prediction submodule; extrapolation is performed using the R_ion and Rct degradation trend curves obtained from each individual battery's previous diagnoses to obtain the predicted remaining service life of each individual battery; when the predicted remaining life of any individual battery is lower than a preset threshold, a targeted maintenance prompt is output.
6. A method for synergistic early warning of membrane degradation and catalyst loss in an AEM electrolyzer based on the apparatus described in any one of claims 1-5, characterized in that, The process includes the following steps: S1. Baseline calibration per cell: After the electrolyzer is first put into operation or maintenance is completed, EIS and OCV decay tests are performed on each cell in the stack sequentially, and the baseline parameters Rct0_i, R_ion0_i, and dV / dt0_i of each cell are recorded; S2. Natural fluctuation calibration per cell: 3-5 verification diagnostics are performed on each cell, and parameter fluctuations under normal operating conditions are collected to calculate the standard deviation of natural fluctuation σ_Rct_i, σ_Rion_i, and σ_dVdt_i; S3. Periodic multi-channel diagnostic triggering: During normal hydrogen production operation of the AEM electrolyzer, online diagnostics are initiated at preset time intervals; the cell to be tested is connected to the excitation acquisition circuit by sequentially polling through the channel selection switch matrix; EIS data is collected in the frequency range of 10mHz-10kHz, and Rct_i and R_ion_i are extracted; OCV is collected in the time window of 10-150s after the main rectified power supply is cut off. S4. Cell-by-cell normalization calculation and fault decoupling judgment: Calculate the normalized change for each cell independently and determine the fault type according to the judgment logic described in claim 1; S5. Cell-by-cell graded early warning and local closed-loop control linkage: When membrane degradation is warned: Locally cool the electrolyte in the area where the cell is located, reducing the temperature by 2-5℃; When catalyst loss is warned: Limit the rate of change of current for the cell, prohibiting current step jumps; When membrane mechanical damage and transpermeation are warned: Reduce the anode-cathode voltage difference of the electrolyzer and shorten the online diagnosis cycle of the cell; When coupling degradation is warned: Implement load reduction protection for the faulty cell or interlock shutdown of the entire stack; S6. Cell-by-cell degradation data archiving and remaining life prediction: Save the diagnostic data of each cell, plot the degradation trend curve based on the cell baseline, and predict the remaining service life of each cell; When the predicted remaining life is lower than the set threshold, output a device maintenance prompt.
7. The AEM electrolyzer membrane degradation-catalyst loss synergistic early warning method according to claim 6, characterized in that, In step S3, the single-cell diagnostic time window is controlled within 10-30 seconds; the channel selection switch matrix is used to sequentially poll and scan; for a stack containing 20 single cells, the complete cycle of diagnostics for all single cells is controlled within 60-300 seconds.
8. The AEM electrolyzer membrane degradation-catalyst loss synergistic early warning method according to claim 6, characterized in that, In step S5, if the same single cell is diagnosed as having a coupling degradation fault in two consecutive diagnostic tests, the local battery pack load reduction protection will be triggered first; if more than a preset proportion of single cells in the same area trigger the coupling degradation warning at the same time, the entire stack will be shut down.
9. The AEM electrolyzer membrane degradation-catalyst loss synergistic early warning method according to claim 6, characterized in that, The calibration conditions are as follows: current density 0.8-1.2 A·cm⁻², electrolyte temperature 55-65℃, electrolyte is 0.8-1.2 mol / L KOH solution, and anode-cathode pressure difference 0.2-0.3 MPa.
