A kind of on-line evaluation method of SOFC stack comprehensive health degree based on distiller's grains source biological methane

Abstract

The application relates to the field of SOFC stacks, and discloses an online evaluation method for the comprehensive health degree of an SOFC stack based on distiller's grains source biomethane, which comprises the following steps: synchronously acquiring electrochemical impedance spectrum data of the SOFC stack and component fluctuation data of the distiller's grains source biomethane; obtaining impedance characteristic components representing different electrochemical polarization processes; separating out substance transport polarization impedance characteristics representing the attenuation depth of carbon deposition and charge transfer polarization impedance characteristics representing the attenuation depth of sulfur poisoning; based on the substance transport polarization impedance characteristics and the charge transfer polarization impedance characteristics, constructing a multi-dimensional health degree state parameter set; calculating the Mahalanobis distance between the health degree state parameter set and a preset safety reference space and generating a health evaluation and early warning instruction for regulating the operation boundary of the SOFC stack based on the Mahalanobis distance. Through the technical scheme, the risk of stack damage caused by the aggravation of carbon deposition and sulfur poisoning is reduced.

Description

Technical Field

[0001] This application relates to the field of SOFC fuel cell stacks, and more particularly to an online method for assessing the overall health of SOFC fuel cell stacks based on biomethane from distillers grains. Background Technology

[0002] Solid oxide fuel cells (SOFCs) are characterized by their wide fuel adaptability and high power generation efficiency, and have broad application prospects in the field of biomass power generation. The brewing industry generates a large amount of by-product lees annually. Traditional methods of disposal, such as stockpiling, landfilling, or using them as low-value materials, result in low resource utilization and are prone to environmental pollution. Therefore, using biomethane produced from the anaerobic fermentation of lees as fuel for SOFCs is of great significance for achieving material resource recycling.

[0003] However, the composition of biomethane from distillers' grains fluctuates significantly, with a marked variation in the methane-to-carbon dioxide ratio, and it is often accompanied by trace amounts of impurities such as hydrogen sulfide. Under the high-temperature operating environment of SOFCs, this unstable fuel input can lead to stack degradation: carbon-rich fluctuations trigger carbon deposition at the anode, clogging the porous structure; and trace amounts of hydrogen sulfide can cause irreversible sulfur poisoning of the nickel-based anode catalyst.

[0004] Based on this, this application proposes an online comprehensive health assessment of SOFC stacks with biomethane from distillers' grains to distinguish whether the root cause of SOFC stack degradation due to biomethane from distillers' grains is anode carbon deposition or catalyst sulfur poisoning. Summary of the Invention

[0005] This application provides an online assessment method for the overall health of SOFC stacks based on biomethane from distillers' grains, in order to at least partially solve the above-mentioned technical problems.

[0006] To achieve the above objectives, this application provides an online assessment method for the comprehensive health of SOFC stacks based on biomethane from distillers' grains, including:

[0007] Simultaneously acquire electrochemical impedance spectroscopy data of SOFC stacks and component fluctuation data of biomethane from distillers grains;

[0008] The electrochemical impedance spectroscopy data is converted into relaxation time distribution curves, and the relaxation time distribution curves are analyzed by multi-peak separation to obtain impedance characteristic components characterizing different electrochemical polarization processes.

[0009] The component fluctuation data is mapped to attenuation boundary parameters, and the attenuation boundary parameters are used to perform coupling feature separation on the impedance characteristic components to separate the material transport polarization impedance characteristics characterizing the carbon deposition attenuation depth and the charge transfer polarization impedance characteristics characterizing the sulfur poisoning attenuation depth.

[0010] Based on the material transport polarization impedance characteristics and the charge transfer polarization impedance characteristics, a multi-dimensional health status parameter set is constructed.

[0011] Calculate the Mahalanobis distance between the set of health status parameters and the preset safety reference space, and generate health assessment and early warning commands for regulating the operating boundary of the SOFC stack based on the Mahalanobis distance.

[0012] In this embodiment of the application, the above technical solution can effectively eliminate signal interference caused by fuel composition fluctuations, identify the two main types of degradation and degradation depth of SOFC stacks, and reduce the risk of stack damage caused by carbon buildup and sulfur poisoning.

[0013] Other features and advantages of this application will be described in detail in the following detailed description section. Attached Figure Description

[0014] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0015] Figure 1 This is a flowchart illustrating the steps of an online assessment method for the overall health of an SOFC stack based on biomethane from distillers' grains, provided in an exemplary embodiment of this application. Detailed Implementation

[0016] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the protection scope of this application.

[0017] This application provides an online method for comprehensive health assessment of SOFC stacks based on biomethane from distillers' grains. Please refer to [link to relevant documentation]. Figure 1 The online assessment method for the overall health of SOFC stacks based on biomethane from distillers grains provided in this application includes the following steps:

[0018] Step S101: Simultaneously acquire electrochemical impedance spectroscopy data of SOFC stack and component fluctuation data of biomethane from distillers grains.

[0019] Step S102: Convert the electrochemical impedance spectroscopy data into relaxation time distribution curves, perform multi-peak separation analysis on the relaxation time distribution curves, and obtain impedance characteristic components characterizing different electrochemical polarization processes.

[0020] Step S103: Map the component fluctuation data to attenuation boundary parameters and use the attenuation boundary parameters to perform coupling feature separation on the impedance characteristic components, separating the material transport polarization impedance characteristics characterizing the carbon deposition attenuation depth and the charge transfer polarization impedance characteristics characterizing the sulfur poisoning attenuation depth.

[0021] Step S104: Based on the material transport polarization impedance characteristics and the charge transfer polarization impedance characteristics, construct a multi-dimensional health status parameter set.

[0022] Step S105: Calculate the Mahalanobis distance between the health status parameter set and the preset safety reference space, and generate health assessment and early warning instructions for regulating the SOFC stack operating boundary based on the Mahalanobis distance.

[0023] Specifically, chemical impedance spectroscopy data refers to frequency domain impedance data obtained through impedance testing that can reflect the losses of various polarization reactions inside the fuel cell stack; biomethane from distillers' grains is a combustible gas produced by anaerobic fermentation of distillers' grains and can be used as SOFC fuel. This gas has the characteristics of unstable component ratio and contains trace impurities. Component fluctuation data is data that characterizes the changes in the concentration and ratio of various gases inside the fuel over time.

[0024] The electrochemical impedance spectroscopy data is converted into relaxation time distribution curves. Multi-peak separation analysis is performed on these relaxation time distribution curves to obtain impedance characteristic components characterizing different electrochemical polarization processes. The relaxation time distribution curves are obtained from the electrochemical impedance spectroscopy and can distinguish the response characteristics of different electrochemical processes within the SOFC stack. Electrochemical polarization is an electrochemical reaction process in SOFC stack operation that generates voltage losses due to charge transfer and mass transport. The impedance characteristic components are impedance parameters corresponding to a single polarization process after separation and can be used to assess the stack's degradation state.

[0025] The component fluctuation data is mapped to attenuation boundary parameters, and coupling feature separation is performed on the impedance characteristic components using these parameters to separate the mass transport polarization impedance characteristics characterizing the carbon deposition attenuation depth and the charge transfer polarization impedance characteristics characterizing the sulfur poisoning attenuation depth. The attenuation boundary parameters are characteristic parameters derived from fuel component fluctuation data and used to define the boundary of stack degradation. Coupling feature separation refers to a processing method that eliminates the mutual interference of multiple degradation factors and separates the impedance signal corresponding to a single degradation. Carbon deposition is a fault phenomenon where carbon-rich components of biomethane from distillers grains accumulate in the SOFC anode, clogging electrode pores. The mass transport polarization impedance characteristics correspond to the impedance characteristics generated during the diffusion and transport of fuel gas within the electrode, and their changes reflect the degree of transport degradation caused by carbon deposition. The aforementioned sulfur poisoning is an irreversible damage phenomenon caused by trace amounts of hydrogen sulfide in the fuel, resulting in a decrease in the activity of the SOFC anode catalyst. The charge transfer polarization impedance characteristics correspond to the impedance characteristics generated during the electron and ion transfer processes on the electrode surface, and their changes can assess the degradation depth of catalyst sulfur poisoning.

[0026] Based on the material transport polarization impedance characteristics and the charge transfer polarization impedance characteristics, a multi-dimensional set of health status parameters is constructed.

[0027] The Mahalanobis distance between the set of health status parameters and the preset safety reference space is calculated, and health assessment and early warning instructions for regulating the operating boundary of the SOFC stack are generated based on the Mahalanobis distance. The safety reference space is the range of normal operating parameters established by the stack under non-deterioration conditions. The Mahalanobis distance is a statistical distance used to measure the degree of deviation between the multi-dimensional parameter set and the standard reference space, which can effectively avoid the judgment error caused by the correlation between parameters.

[0028] This paper addresses the technical challenge of existing technologies where distillers' grains-derived biomethane, due to its large compositional fluctuations and hydrogen sulfide impurities, easily causes simultaneous anode carbon deposition and catalyst sulfur poisoning in SOFC stacks, and the difficulty in distinguishing the root causes of these two types of degradation signals. This solution simultaneously acquires fuel composition data and stack electrochemical impedance data, converts the impedance spectrum into a relaxation time distribution curve, and decomposes it to obtain the impedance components corresponding to each polarization process. The coupling signal is separated using attenuation boundary parameters calculated from the fuel composition, distinguishing between the mass transport polarization impedance corresponding to carbon deposition and the charge transfer polarization impedance corresponding to sulfur poisoning. The deviation between the parameter set and the safe reference space is compared using Mahalanobis distance. This solution effectively eliminates signal interference caused by fuel composition fluctuations, identifies the two main types of degradation and their depth in SOFC stacks, reduces the risk of stack damage caused by carbon deposition and sulfur poisoning, and is suitable for the unstable biomass fuel conditions of distillers' grains-derived biomethane, thus improving the operational stability of SOFC power generation systems utilizing distillers' grains resources.

[0029] In some embodiments, converting the electrochemical impedance spectroscopy data into relaxation time distribution curves includes:

[0030] Obtain the real part dataset and the imaginary part dataset from the electrochemical impedance spectroscopy data; wherein, the real part dataset is the set of values ​​representing the active component of impedance in the electrochemical impedance spectrum, and the imaginary part dataset is the set of values ​​representing the reactive component of impedance in the electrochemical impedance spectrum, and the two together constitute the impedance frequency domain information.

[0031] An equivalent circuit model is constructed, which includes a parallel resistor-capacitor sub-model and an integrator kernel. The equivalent circuit model is a simulation model that uses basic circuit elements such as resistors and capacitors to simulate various electrochemical reactions and mass transport processes inside the SOFC stack. The aforementioned integrator kernel is the core function used for integration operations, connecting the circuit model, and calculating the relaxation time distribution.

[0032] A regularization term with a penalty coefficient is introduced into the fitting residual function of the imaginary part dataset to construct the inversion objective function. This can be understood as the fitting residual function being a function that characterizes the magnitude of the deviation between the calculated value of the equivalent circuit model and the actual measured value of the imaginary part dataset. The penalty coefficient is an adjustable parameter used to control the constraint strength of the regularization term. The aforementioned regularization term is a constraint term added to suppress computational noise and ensure the smoothness of the solution results. The inversion objective function is an optimization function that integrates the residual and constraint conditions for subsequent extreme value solving.

[0033] Solve for the global minimum of the inversion objective function and output the distribution density array of impedance amplitude as a function of the logarithm of the time constant as the relaxation time distribution curve. The global minimum is the optimal solution of the inversion objective function in the entire solution interval, corresponding to the calculation result with the best fitting effect to the impedance data. The distribution density array is a set of numerical data recording the proportion of impedance distribution under different time constants. The visualization of this array is the relaxation time distribution curve.

[0034] Due to the difficulty in distinguishing different electrochemical processes within a SOFC by directly using the raw electrochemical impedance spectroscopy, the susceptibility of simple fitting calculations to measurement noise interference, and the tendency for oscillation distortion in the solution results, this step addresses the technical problems of obtaining the core data of the real and imaginary parts of the impedance spectrum. An equivalent circuit model is constructed based on the integrator kernel of the parallel resistor-capacitor sub-model that closely matches the working mechanism of the SOFC. The imaginary part data is used to construct the fitting residual function, and a regularization term with a penalty coefficient is added to form the inversion objective function to constrain the solution process and filter out noise interference. The optimal calculation result is obtained by solving for the global minimum of the objective function, and a relaxation time distribution curve is generated. This method effectively reduces the calculation error and numerical oscillation in the impedance data inversion process, improves the accuracy and smoothness of curve fitting, and can realistically reproduce the response characteristics of different electrochemical processes in a SOFC, outputting reliable relaxation time distribution data.

[0035] In some embodiments, a regularization term with a penalty coefficient is introduced into the fitting residual function of the imaginary part dataset to construct the inversion objective function, including:

[0036] The sum of squared differences between the actual measured values ​​at each frequency point in the imaginary part dataset and the predicted values ​​output by the equivalent circuit model is calculated to generate the fitting residual function. Here, each frequency point is a series of test frequency points set when collecting electrochemical impedance spectroscopy, and the sum of squared differences is the result obtained by squaring the difference between the measured value and the model predicted value at each frequency point and then summing them up. Overall, it reflects the degree of deviation between the model calculation result and the actual measurement data.

[0037] A difference operator matrix for the relaxation time distribution density is constructed, and the norm of the difference operator matrix is ​​calculated as the regularization term for constraining the smoothness of the solution. The difference operator matrix is ​​an operation matrix used to characterize the variation law of adjacent data points of the relaxation time distribution density. The norm of the matrix is ​​a measure of the overall numerical value and variation of the matrix. In this scheme, this indicator is used to limit the fluctuation of the distribution curve, thereby obtaining the regularization term used to constrain the smoothness of the solution.

[0038] The preset weight value is used as the penalty coefficient multiplied by the regularization term; the weight value is a fixed parameter set in advance and used to adjust the constraint strength of the regularization term in the overall function.

[0039] The product results are added to the fitted residual function to generate a summation polynomial as the inversion objective function. The summation polynomial is a polynomial function formed by adding the fitted residual function and the weighted regularization term, and it is also the inversion objective function on which the subsequent numerical solution is based.

[0040] This scheme first constructs a fitting residual function by summing the squared differences between the measured values ​​and the model predictions at each frequency point to measure the data fitting deviation. It then constructs a difference operator matrix and calculates its norm as a regularization term to specifically constrain the variation of the relaxation time distribution density. A preset weight value is used as a penalty coefficient to adjust the constraint strength of the regularization term. The weighted regularization term is then combined with the fitting residual function to obtain the inversion objective function. This method effectively suppresses numerical oscillations caused by measurement noise while ensuring the accuracy of impedance data fitting, ensuring the overall smoothness and continuity of the relaxation time distribution curve, and avoiding subsequent feature analysis errors caused by curve distortion.

[0041] In some embodiments, the relaxation time distribution curve is analyzed by multi-peak separation to obtain impedance characteristic components characterizing different electrochemical polarization processes, including:

[0042] The second derivative peak finding operation is performed on the relaxation time distribution curve to locate multiple extreme center frequency points. The second derivative peak finding operation is a mathematical algorithm that uses the numerical change law of the second derivative of the function to identify the peak position of the curve. The extreme center frequency point is the center position point corresponding to each peak on the relaxation time distribution curve, which corresponds to the characteristic response frequency of different electrochemical processes.

[0043] Using the extreme center frequency point as the initial expected value, a nonlinear least squares fitting is performed on the relaxation time distribution curve; wherein, the initial expected value is a preset peak position initial value before the fitting operation begins, providing the starting condition for iterative calculation, and nonlinear least squares fitting is a numerical fitting method for nonlinear curve models, which continuously iterates to reduce the deviation between the calculated value and the original curve.

[0044] The integral area corresponding to each independent Gaussian peak after fitting convergence is taken as the polarization resistance component, and the corresponding abscissa parameter is taken as the relaxation time characteristic value. The fitting convergence mentioned above refers to the state where the fitting iteration operation reaches the preset accuracy requirement and the calculation result no longer changes significantly. The Gaussian peak is an independent peak whose shape conforms to the Gaussian distribution, corresponding to a single electrochemical polarization process inside the stack. The integral area is the value obtained by integrating a single Gaussian peak within the corresponding interval. The polarization resistance component is a parameter characterizing the magnitude of the corresponding electrochemical polarization loss. The abscissa parameter is the logarithm of the time constant corresponding to the position of the Gaussian peak. The relaxation time characteristic value is a characteristic parameter characterizing the response rate of the corresponding electrochemical process.

[0045] The polarization resistance component and the relaxation time characteristic value are output as the impedance characteristic component.

[0046] This scheme uses the second derivative peak finding algorithm to accurately locate the center frequency of all peaks. The center frequency is used as the initial expectation value to carry out nonlinear least squares fitting. After the fitting converges, the integral area and abscissa parameters of each independent Gaussian peak are obtained and used as polarization resistance component and relaxation time characteristic value, respectively, and integrated into impedance characteristic component. This method effectively solves the problems of multiple overlapping curves that are difficult to separate and inaccurate peak position identification.

[0047] In some embodiments, coupling feature separation is performed on the impedance characteristic components using the attenuation boundary parameter to separate the mass transport polarization impedance characteristics characterizing the carbon deposition attenuation depth and the charge transfer polarization impedance characteristics characterizing the sulfur poisoning attenuation depth, including:

[0048] The first integral area of ​​the impedance characteristic component with a frequency lower than a preset material transport frequency threshold and the second integral area with a frequency higher than a preset charge transfer frequency threshold are read. The material transport frequency threshold is a preset frequency boundary value used to divide the low-frequency range of the corresponding material transport process. The first integral area is the integral value corresponding to the impedance characteristic component in the low-frequency range. The charge transfer frequency threshold is another preset frequency boundary value used to divide the high-frequency range of the corresponding charge transfer process. The second integral area is the integral value corresponding to the impedance characteristic component in the high-frequency range.

[0049] Obtain the extreme value of carbon-oxygen ratio fluctuation and the cumulative flux of hydrogen sulfide in the attenuation boundary parameters; wherein the extreme value of carbon-oxygen ratio fluctuation is the maximum value of the carbon-oxygen ratio in biomethane from distillers grains deviating from the standard ratio during the evaluation period, and the cumulative flux of hydrogen sulfide is the cumulative value of the total amount of hydrogen sulfide in the fuel during the evaluation period, which is used to reflect the degree of toxic accumulation.

[0050] The extreme values ​​of the carbon-oxygen ratio fluctuation are converted into the corresponding carbon deposition conversion rate and a first compensation factor is generated. The first compensation factor is used to perform a coefficient correction operation on the first integral area to remove the low-frequency impedance change introduced by the fuel calorific value fluctuation and output the material transport polarization impedance characteristics. The carbon deposition conversion rate is a parameter that characterizes the anode carbon deposition generation rate, calculated based on the fuel carbon-oxygen ratio fluctuation. The first compensation factor is a correction coefficient used to offset fuel interference. The coefficient correction operation is a calculation process that completes data correction by multiplying the coefficients. The fuel calorific value fluctuation is the fluctuation of fuel energy density caused by the change of biomethane composition in distillers' grains. The low-frequency impedance change is the interference impedance component superimposed on the material transport polarization signal caused by the fuel calorific value fluctuation.

[0051] The cumulative hydrogen sulfide flux is converted into the corresponding catalyst poisoning inhibition coefficient and a second compensation factor is generated. The second compensation factor is used to perform coefficient correction calculation on the second integral area to remove the high-frequency impedance change introduced by the trace toxic gas fluctuation and output the charge transfer polarization impedance characteristics. The catalyst poisoning inhibition coefficient is a parameter that characterizes the degree of inhibition of anode catalyst activity, which is calculated based on the cumulative hydrogen sulfide flux. The second compensation factor is a correction coefficient specifically used to offset the interference of hydrogen sulfide. The trace toxic gas fluctuation is an interference factor caused by the continuous change of hydrogen sulfide concentration in the fuel. The high-frequency impedance change is the interference impedance component superimposed on the charge transfer polarization signal caused by the hydrogen sulfide fluctuation.

[0052] This scheme obtains the integral areas corresponding to high and low frequencies from the impedance characteristic components based on a preset frequency threshold. It uses the extreme value of carbon-oxygen ratio fluctuation and hydrogen sulfide cumulative flux in the attenuation boundary parameters to calculate the carbon deposition conversion rate and catalyst poisoning inhibition coefficient, and generates two sets of compensation factors accordingly. The coefficients of the high and low frequency integral areas are corrected respectively, and the interference impedance components caused by fuel calorific value and trace toxic gases are eliminated one by one, so as to separate the pure mass transport polarization impedance characteristics and charge transfer polarization impedance characteristics. This method effectively solves the problem of coupling and superposition of multiple types of degradation signals and fuel interference signals.

[0053] In some embodiments, a multi-dimensional set of health state parameters is constructed based on the material transport polarization impedance characteristics and the charge transfer polarization impedance characteristics, including:

[0054] Calculate the first degradation rate of the mass transport polarization impedance characteristic relative to the initial calibration state; wherein, the initial calibration state refers to the nominal operating state of the SOFC stack under standard operating conditions and without degradation; the first degradation rate is used to characterize the rate of degradation of the mass transport performance of the stack due to carbon deposition over time.

[0055] Calculate the second degradation rate of the charge transfer polarization impedance characteristic relative to the initial calibration state; the second degradation rate is used to characterize the rate of degradation of the charge transfer performance of the fuel cell due to sulfur poisoning over time.

[0056] The high-frequency intercept resistance in the electrochemical impedance spectroscopy data is obtained, and its third degradation rate, which characterizes ohmic degradation, is calculated. The high-frequency intercept resistance is the resistance value corresponding to the intersection of the electrochemical impedance spectroscopy with the horizontal axis in the high-frequency range. It mainly reflects the ohmic losses generated by the electrodes, connecting parts, electrolytes, etc. inside the stack. Ohmic degradation is the phenomenon of continuous increase in ohmic losses caused by the aging of the conductive structure of the stack and the decline in contact performance. The third degradation rate is used to characterize the degradation rate of the ohmic losses of the stack.

[0057] The first degradation growth rate, the second degradation growth rate, and the third degradation growth rate are used as independent feature terms, and the Pearson correlation coefficient between each growth rate is used as a cross-correlation feature term. These are then fused to generate the health status parameter set. The aforementioned independent feature terms are basic parameters that individually characterize a certain type of degradation degree. The Pearson correlation coefficient is a statistical indicator used to measure the degree of linear correlation between two sets of data. The cross-correlation feature term is used to reflect the mutual influence relationship between different degradation processes. By fusing the two types of features, a complete multi-dimensional health status parameter set can be obtained.

[0058] This scheme calculates the growth rates of three types of degradation—mass transport, charge transfer, and ohmic loss—as independent features, introduces the Pearson correlation coefficient to characterize the correlation between each degradation rate, and integrates them to form a multi-dimensional health status parameter set. This approach comprehensively covers the main degradation types of SOFC stacks, achieves independent characterization of each degradation rate, captures the correlation patterns of different degradation processes, and solves the problems of traditional assessments having a single dimension and ignoring the interaction between degradations.

[0059] In some embodiments, acquiring the component fluctuation data of the biomethane from the distiller's grains includes:

[0060] Within a continuous sampling time window, data on methane concentration, carbon dioxide concentration, and trace hydrogen sulfide concentration were collected in the fuel input manifold. The fuel input manifold is the main pipeline for supplying biomethane from distillers grains and precipitates to the SOFC stack. The methane and carbon dioxide concentration data represent the real-time content values ​​of the two main components in the fuel, respectively. The trace hydrogen sulfide concentration data represents the real-time content value of hydrogen sulfide, a harmful impurity in the fuel.

[0061] The carbon-oxygen ratio is determined based on the methane concentration data and carbon dioxide concentration data at adjacent sampling times, and carbon-oxygen ratio data is generated. The carbon-oxygen ratio is the ratio of carbon to oxygen elements in the fuel, calculated based on the methane and carbon dioxide content. The carbon-oxygen ratio data is a time-series data set composed of carbon-oxygen ratios at different times.

[0062] The concentration change rate characteristic of the trace hydrogen sulfide concentration data is determined to generate poison impact data; wherein, the concentration change rate characteristic is an indicator characterizing how fast the hydrogen sulfide concentration changes over time, and the poison impact data is used to reflect the instantaneous and continuous impact of hydrogen sulfide impurities on the anode catalyst of the fuel cell stack.

[0063] The carbon-oxygen ratio data and the poison impact data are aligned and stitched together according to the timestamp to generate the component fluctuation data.

[0064] This scheme simultaneously collects the concentration information of methane, carbon dioxide, and hydrogen sulfide in fuel within a continuous sampling time window. The carbon-oxygen ratio data is obtained by calculating the component concentration, and the poison shock data is obtained based on the hydrogen sulfide concentration change rate. The two types of data are aligned and stitched together to form complete component fluctuation data.

[0065] In some embodiments, the step of obtaining the preset security reference space includes:

[0066] Obtain reference electrochemical impedance spectroscopy (EIS) data under a calibration environment. The reference EIS data is collected from the SOFC stack under a pure standard fuel and non-polarization decay state. The calibration environment is a standard and stable test operating condition set by the user. The pure standard fuel is a nominal fuel with a fixed component ratio and free of harmful impurities such as hydrogen sulfide. The non-polarization decay state refers to the nominal state in which the various electrochemical processes of the SOFC stack have no performance degradation and are in brand-new healthy operation. The reference EIS data is the nominal impedance spectrum data collected under this nominal state.

[0067] The reference electrochemical impedance spectroscopy data is subjected to regularized inversion and multi-peak separation to obtain the reference impedance characteristic components. Regularized inversion is a calculation process that introduces regularization constraints to perform inverse solution of impedance data and obtain relaxation time distribution curves. The reference impedance characteristic components are standard impedance parameters obtained by decomposing under nominal operating conditions and corresponding to various electrochemical polarization processes.

[0068] A reference initial parameter set is constructed based on the reference impedance characteristic components, and the boundary of the reference initial parameter set is expanded based on a preset safety margin coefficient to generate an envelope space as the safety reference space. The aforementioned reference initial parameter set is a set of basic parameters representing the optimal health state of the fuel cell stack, composed of nominal impedance characteristic components. The safety margin coefficient is a pre-set proportional coefficient used to reserve a margin for safe operation. Boundary expansion is the operation of expanding the range of parameter set values ​​according to the safety margin coefficient. The envelope space is a multi-dimensional parameter interval that can encompass the normal fluctuation range of the fuel cell stack.

[0069] In some embodiments, mapping the component fluctuation data to attenuation boundary parameters includes:

[0070] The carbon-oxygen ratio data and the poison impact data of the synchronous time series are extracted from the component fluctuation data; wherein the synchronous time series refers to a continuous data group with one-to-one time stamp and completely consistent time sequence, ensuring that the two types of data are matched in time.

[0071] Within a preset evaluation time window, the maximum absolute difference between the carbon-oxygen ratio data and the theoretical safe ratio is calculated to generate the extreme value of carbon-oxygen ratio fluctuation. The theoretical safe ratio is the standard carbon-oxygen ratio of fuel that ensures stable operation of SOFC stacks and is not prone to carbon deposition. The maximum absolute difference is the maximum value between the actual carbon-oxygen ratio and the theoretical safe ratio, which is used to reflect the degree of deviation of the fuel carbon-oxygen ratio.

[0072] Within the evaluation time window, the poison impact data is integrated and accumulated to generate the hydrogen sulfide cumulative flux, which characterizes the depth of irreversible poisoning of the catalyst. The integration and accumulation is a processing method that integrates and sums the time-series data within a specified time range. The hydrogen sulfide cumulative flux is the total effect of hydrogen sulfide on the anode catalyst within the statistical period, reflecting the degree of irreversible damage to the catalyst by the poison.

[0073] The extreme value of the carbon-oxygen ratio fluctuation and the cumulative flux of hydrogen sulfide are combined to form the attenuation boundary parameter; this parameter set comprehensively reflects the two types of core degradation effects caused by fuel composition fluctuations.

[0074] The above are merely preferred embodiments of this application and are not intended to limit this application in any way. Any simple modifications, equivalent changes, and alterations made to the above embodiments based on the technical essence of this application without departing from the scope of the technical solution of this application shall still fall within the scope of the technical solution of this application.

Claims

1. A method for online assessment of the overall health of SOFC stacks based on biomethane from distillers' grains, characterized in that, include: Simultaneously acquire electrochemical impedance spectroscopy data of SOFC stacks and component fluctuation data of biomethane from distillers grains; The electrochemical impedance spectroscopy data is converted into relaxation time distribution curves, and the relaxation time distribution curves are analyzed by multi-peak separation to obtain impedance characteristic components characterizing different electrochemical polarization processes. The component fluctuation data is mapped to attenuation boundary parameters, and the attenuation boundary parameters are used to perform coupling feature separation on the impedance characteristic components to separate the material transport polarization impedance characteristics characterizing the carbon deposition attenuation depth and the charge transfer polarization impedance characteristics characterizing the sulfur poisoning attenuation depth. Based on the material transport polarization impedance characteristics and the charge transfer polarization impedance characteristics, a multi-dimensional health status parameter set is constructed. Calculate the Mahalanobis distance between the set of health status parameters and the preset safety reference space, and generate health assessment and early warning commands for regulating the operating boundary of the SOFC stack based on the Mahalanobis distance.

2. The method according to claim 1, characterized in that, The electrochemical impedance spectroscopy data is converted into relaxation time distribution curves, including: Obtain the real part and imaginary part datasets from the electrochemical impedance spectroscopy data; Construct an equivalent circuit model that includes an integrator kernel for a parallel resistor-capacitor sub-model; A regularization term with a penalty coefficient is introduced into the fitting residual function of the imaginary part dataset to construct the inversion objective function; The global minimum of the inversion objective function is solved, and the distribution density array of the output impedance amplitude as a function of the logarithm of the time constant is used as the relaxation time distribution curve.

3. The method according to claim 2, characterized in that, A regularization term with a penalty coefficient is introduced into the fitting residual function of the imaginary part dataset to construct the inversion objective function, including: Calculate the sum of squares of the differences between the actual measured values ​​at each frequency point in the imaginary part dataset and the predicted values ​​output by the equivalent circuit model, and generate the fitting residual function; Construct a difference operator matrix for the relaxation time distribution density and calculate the norm of the difference operator matrix as the regularization term for constraining the smoothness solution; The preset weight value is used as the penalty coefficient and multiplied by the regularization term; The product results are added to the fitted residual function to generate a summation polynomial as the inversion objective function.

4. The method according to claim 3, characterized in that, The relaxation time distribution curve was analyzed by multi-peak separation to obtain impedance characteristic components characterizing different electrochemical polarization processes, including: Perform second-order derivative peak finding operation on the relaxation time distribution curve to locate multiple extreme value center frequency points; Using the extreme center frequency point as the initial expected value, perform nonlinear least squares fitting on the relaxation time distribution curve; The integral area corresponding to each independent Gaussian peak after the fitting convergence is taken as the polarization resistance component, and the corresponding abscissa parameter is taken as the relaxation time characteristic value. The polarization resistance component and the relaxation time characteristic value are output as the impedance characteristic component.

5. The method according to claim 4, characterized in that, Using the attenuation boundary parameters, coupling feature separation is performed on the impedance characteristic components to separate the mass transport polarization impedance characteristics characterizing the carbon deposition attenuation depth and the charge transfer polarization impedance characteristics characterizing the sulfur poisoning attenuation depth, including: Read the first integral area of ​​the impedance characteristic component with a frequency lower than a preset material transport frequency threshold and the second integral area with a frequency higher than a preset charge transfer frequency threshold. Obtain the extreme value of carbon-oxygen ratio fluctuation and the cumulative flux of hydrogen sulfide in the attenuation boundary parameters; The extreme value of the carbon-oxygen ratio fluctuation is converted into the corresponding carbon deposition conversion rate and a first compensation factor is generated. The first compensation factor is used to perform coefficient correction calculation on the first integral area, and the low-frequency impedance change introduced by the fuel calorific value fluctuation is removed, and the material transport polarization impedance characteristics are output. The cumulative flux of hydrogen sulfide is converted into the corresponding catalyst poisoning retardation coefficient and a second compensation factor is generated. The second compensation factor is used to perform coefficient correction calculation on the second integral area to remove the high-frequency impedance change introduced by the fluctuation of trace toxic gas and output the charge transfer polarization impedance characteristics.

6. The method according to claim 5, characterized in that, Based on the aforementioned material transport polarization impedance characteristics and charge transfer polarization impedance characteristics, a multi-dimensional set of health state parameters is constructed, including: Calculate the first degradation rate of the material transport polarization impedance characteristic relative to the initial calibration state; Calculate the second degradation rate of the charge-transfer polarization impedance characteristic relative to the initial calibration state; Obtain the high-frequency intercept resistance from the electrochemical impedance spectroscopy data and calculate its third degradation rate, which characterizes ohmic degradation. The first degradation growth rate, the second degradation growth rate, and the third degradation growth rate are used as independent features, and the Pearson correlation coefficient between the growth rates is used as a cross-correlation feature to generate the health status parameter set.

7. The method according to claim 6, characterized in that, Obtaining the component fluctuation data of the biomethane from the distiller's grains includes: Within a continuous sampling time window, data on methane concentration, carbon dioxide concentration, and trace hydrogen sulfide concentration in the fuel input manifold were collected. The carbon-oxygen ratio is determined based on the methane concentration data and carbon dioxide concentration data at adjacent sampling times, and carbon-oxygen ratio data is generated. The concentration change rate characteristics of the hydrogen sulfide trace concentration data are determined to generate toxic substance impact data. The carbon-oxygen ratio data and the poison impact data are aligned and stitched together according to the timestamp to generate the component fluctuation data.

8. The method according to claim 7, characterized in that, The steps for obtaining the preset security reference space include: Obtain reference electrochemical impedance spectroscopy data under calibration conditions; the reference electrochemical impedance spectroscopy data is obtained by collecting SOFC stack data under pure standard fuel and non-polarization decay conditions; Regularized inversion and multi-peak separation are performed on the reference electrochemical impedance spectroscopy data to obtain the reference impedance characteristic components; A reference initial parameter set is constructed based on the reference impedance characteristic components, and the boundary of the reference initial parameter set is expanded based on a preset safety margin coefficient to generate an envelope space as the safety reference space.

9. The method according to claim 8, characterized in that, Mapping the component fluctuation data to attenuation boundary parameters includes: The carbon-oxygen ratio data and the poisoning agent impact data of the synchronous time series are extracted from the component fluctuation data; Within a preset evaluation time window, the maximum absolute difference between the carbon-oxygen ratio data and the theoretical safe ratio is calculated, generating the extreme value of carbon-oxygen ratio fluctuation. Within the evaluation time window, the poison impact data are integrated and accumulated to generate the cumulative hydrogen sulfide flux characterizing the depth of irreversible poisoning of the catalyst. The extreme value of the carbon-oxygen ratio fluctuation is combined with the cumulative flux of hydrogen sulfide to form the attenuation boundary parameter.

Images