Fuel cell aging prediction result uncertainty quantification and decomposition method

Abstract

The invention relates to the technical field of fuel cells, and discloses a fuel cell aging prediction result uncertainty quantification and decomposition method, which comprises the following steps: acquiring and preprocessing voltage data; modeling and training of a B-MIXGU aging prediction model are carried out; aging prediction and result analysis; and uncertainty quantization and decomposition: quantizing the total uncertainty of the aging prediction result by using the variance of interval estimation, decomposing the total uncertainty into accidental uncertainty and cognitive uncertainty, and exploring the source of the uncertainty. According to the uncertainty quantification and decomposition method for the aging prediction result of the fuel cell, the predicted total uncertainty is decomposed into cognitive uncertainty and accidental uncertainty, and credibility evaluation of the aging prediction result is realized according to the uncertainty decomposition result.

Description

Technical Field

[0001] The present invention relates to the technical field of fuel cells, and in particular to a method for quantifying and decomposing uncertainty of fuel cell aging prediction results. Background Art

[0002] Proton exchange membrane fuel cells (PEMFCs) are widely used in transportation fields such as hybrid vehicles, buses, ships, and airplanes due to their advantages such as no pollution, high efficiency, and low noise. However, insufficient lifespan is one of the key issues that hinders their large-scale commercialization.

[0003] Under dynamic operating conditions, the disturbances of external environment and load changes can have a significant impact on the durability of PEMFC. Prognostic and health management (PHM) can estimate the remaining useful life (RUL) by providing degradation information of PEMFC. Therefore, accurately predicting the performance degradation trend of PEMFC is crucial to estimating its RUL.

[0004] The methods for predicting the performance degradation trend of PEMFC mainly include model-based methods, data-driven methods, and hybrid methods based on a combination of model and data-driven. Related studies have shown that PEMFC is a system with nonlinear, multi-physical parameters, and multi-scale characteristics. It is currently difficult to establish an accurate physical model to describe its degradation mechanism. The model-based method relies on complex and high-precision physical models, so it is difficult to achieve ideal prediction results. The data-driven method predicts the performance degradation by learning directly from PEMFC data. This method effectively overcomes the difficulty of establishing a high-precision physical model. The hybrid method based on a combination of model and data-driven attempts to combine the advantages of both, but it relies on a high-precision physical model, so it is quite challenging to accurately construct such a model.

[0005] In addition, with the rapid development of machine learning technology, especially the powerful generalization ability, excellent scalability and high precision of deep learning technology in processing complex data, the data-driven method based on deep learning has become an effective tool for predicting PEMFC performance degradation.

[0006] There are currently a variety of data-driven methods based on machine learning, such as support vector machines, particle filters, adaptive unscented Kalman filters, gated recurrent units, long short-term memory recurrent neural networks, etc. Although these methods have achieved many remarkable successes, they can usually only guarantee the validity of the prediction results when the distribution of the test data is similar to that of the training data. In practical applications, the distribution of test data may have a domain offset from the distribution of training data due to changes in the external environment and load, and the domain offset will have a significant impact on the prediction of PEMFC performance degradation. When faced with unknown data, deep learning models may often give overconfident and unreliable prediction results without clear warnings. Therefore, in the prediction of PEMFC performance degradation, it is necessary to pay attention not only to the accuracy of the prediction results, but also to the credibility of the prediction results.

[0007] There are two main sources of uncertainty in prediction: aleatory uncertainty and epistemic uncertainty. Although uncertainty quantification is crucial for building credible prediction models, it is often overlooked in PEMFC performance degradation prediction. Most deep learning-based methods only provide point estimate predictions, which cannot fully convey the credibility of the prediction results.

[0008] Therefore, developing models that can effectively quantify and communicate prediction uncertainties is of great significance for improving the reliability and practicality of PEMFC aging prediction. Summary of the invention

[0009] The purpose of the present invention is to address the deficiencies of the above-mentioned technologies and provide a method for quantifying and decomposing the uncertainty of fuel cell aging prediction results, decomposing the total uncertainty of the prediction into epistemic uncertainty and accidental uncertainty, and realizing the credibility assessment of the aging prediction results based on the uncertainty decomposition results.

[0010] To achieve the above-mentioned purpose, the uncertainty quantification and decomposition method of the fuel cell aging prediction result involved in the present invention comprises the following steps:

[0011] S1) voltage data collection and preprocessing: collecting the voltage data output under the dynamic load cycle condition of PEMFC, performing preprocessing operations to obtain a voltage data set, and dividing the voltage data set into training set data and test set data;

[0012] S2) Modeling and training of B-MIXGU aging prediction model: Replace the model parameters of the mixed gated unit model MIXGU with probability distribution to obtain the Bayesian mixed gated unit model B-MIXGU, obtain the posterior distribution of the B-MIXGU model parameters based on the Bayesian theorem, obtain the output of B-MIXGU according to the posterior distribution, and thus establish the B-MIXGU aging prediction model. Use the variational inference method to approximate the posterior distribution of the model parameters of the B-MIXGU aging prediction model, and solve the minimum KL divergence between the variational distribution and the posterior distribution, which is further converted into optimizing the maximization of the lower bound of evidence, and combine the training set data to obtain the trained B-MIXGU aging prediction model;

[0013] S3) Aging prediction and result analysis: The test set data is input into the trained B-MIXGU aging prediction model to obtain the point estimation and interval estimation aging prediction results of PEMFC, and the aging prediction results are analyzed by the error evaluation index;

[0014] S4) Uncertainty quantification and decomposition: The epistemic uncertainty measured by the randomness of model parameters and the aleatory uncertainty caused by data noise contribute to the total uncertainty together. The variance of interval estimation is used to quantify the total uncertainty of the aging prediction results, and the total uncertainty is decomposed into aleatory uncertainty and epistemic uncertainty to explore the source of uncertainty.

[0015] Preferably, in the step B), MIXGU comprises: a gated recurrent unit GRU, a minimum gated unit MGU and an adaptive weight adjustment module AWAM, and the expression is as follows:

[0016] GRU:

[0017] z k =σ(W z [h k-1 ,x k ]+b z )

[0018] r k =σ(W r [h k-1 ,x k ]+b r )

[0019]

[0020] In the formula, x k Input information for the current time k, h k-1 is the hidden state at the previous moment, h k1 is the hidden state passed from GRU to AWAM, is the candidate hidden state of GRU, z kis the update gate of GRU, r k is the reset gate of GRU, σ is the sigmoid function, tanh is the tanh function, W z Update the weight matrix of the GRU gate, b z The bias matrix of the GRU update gate, W r Reset the gate weight matrix for GRU, b r The bias matrix for the GRU reset gate, W h1 is the weight matrix of the GRU candidate hidden state, b h1 is the bias matrix of the GRU candidate hidden state;

[0021] MGU:

[0022] f k =σ(W f [h k-1 ,x k ]+b f )

[0023]

[0024] In the formula, f k is the update gate of MGU, h k2 is the hidden state passed from MGU to AWAM, is the candidate hidden state of MGU, W f Update the gate weight matrix for MGU, b f Update gate bias matrix for MGU, W h2 is the weight matrix of the candidate hidden state of MGU, b h2 is the bias matrix of the candidate hidden state of MGU;

[0025] AWAM:

[0026] MIXGU k =σ(W s [h k1 ,h k2 ]+b s )

[0027] h k =(1-MIXGU k )*h k1 +MIXGU k *h k2

[0028] Where, MIXGU k is the weight adjustment module of MIXGU at the current moment k, W s for MIXGU k The corresponding weight matrix of the weight adjustment module, b sfor MIXGU k The corresponding bias matrix of the weight adjustment module, h k is the final hidden state output, MIXGU k Decide when MIXGU k When it approaches 0, h k tend to h k1 , indicating that GRU is dominant, while MIXGU k When it approaches 1, h k tend to h k2 , indicating that MGU occupies a dominant position.

[0029] Preferably, the step B) comprises the following steps:

[0030] S201) The model parameters of MIXGU are replaced with probability distribution to obtain B-MIXGU, and the posterior distribution of the B-MIXGU model parameter ω is obtained based on Bayes' theorem:

[0031]

[0032] In the formula, p(ω|D) is the posterior distribution of parameter ω, D is the training set data, p(ω) is the prior distribution of parameter ω, p(D|ω) is the likelihood function of parameter ω, p(D) is the prior distribution of training set data, and parameter ω includes W in the GRU expression z , W r , b z , b r , W h1 , b h1 , W in the MGU expression f , b f , W h2 , b h2 , W in the AWAM expression s , b s ;

[0033] S202) Obtain the output of B-MIXGU according to the posterior distribution, and establish a B-MIXGU aging prediction model:

[0034] p(y|x,D)=∫p(y|x,ω)p(ω,D)dω

[0035] Where x is the input data, y is the model output, p(y|x,D) is the output distribution given the training data D and the input data x, and p(y|x,ω) is the output distribution given the parameter ω and the input data x.

[0036] S203) using a variational inference method to approximate the posterior distribution of the model parameters of the B-MIXGU aging prediction model, and creating a variational distribution q with a parameter θ θ(ω) to approximate the true posterior distribution p(ω|D);

[0037] S204) By solving the minimum KL divergence between the variational distribution and the posterior distribution, it is further transformed into optimizing the maximization of the evidence lower bound ELBO:

[0038] argminKL[q θ (ω)||p(ω|D)]=argmaxELBO;

[0039] S205) Combining the training set data to obtain a trained B-MIXGU aging prediction model.

[0040] Preferably, the step S4) comprises the following steps:

[0041] S401) Epistemic uncertainty measured by the randomness of model parameters and aleatory uncertainty caused by data noise contribute jointly to the total uncertainty;

[0042] p(y|x,D)=∫p(y|x,ω)q θ (ω)dω

[0043] Among them, the randomness of the model parameters is given by the variational distribution q θ (ω) represents, quantifying epistemic uncertainty, and data noise is represented by the probability distribution p(y|x,ω), quantifying aleatoric uncertainty;

[0044] S402) using the variance of interval estimation to quantify the total uncertainty of aging prediction results, decomposing the total uncertainty into aleatory uncertainty and epistemic uncertainty, and exploring the sources of uncertainty;

[0045] D(y)=E(y 2 )-E(y) 2

[0046] =E[E(y 2 |x)]-E[E(y|x)] 2

[0047] =E[D(y|x)+E(y|x) 2 ]-E[E(y|x)] 2

[0048] =E[E(y|x)] 2 -E[E(y|x) 2 ]+E[D(y|x)]

[0049] =D[E(y|x)]+E[D(y|x)]

[0050] Where D(y) represents the total prediction uncertainty, D[E(y|x)] represents the fluctuation of expected E(y|x) due to the variability of model parameters, measuring epistemic uncertainty, and E[D(y|x)] represents the calculation of the average value of D(y|x), eliminating the effects caused by the variability of model parameters, thereby extracting the accidental uncertainty caused by random noise.

[0051] Compared with the prior art, the present invention has the following advantages:

[0052] 1. The B-MIXGU model has excellent PEMFC aging prediction accuracy. The dynamic load cycle durability test data set verifies that the point estimation prediction accuracy of B-MIXGU is far superior to other advanced recurrent neural network models, and it can accurately predict PEMFC aging based on very little training data, making it more suitable for practical applications.

[0053] 2. An uncertainty perception method for PEMFC aging prediction under the probabilistic Bayesian framework is proposed. The variance of the interval prediction results under the probability distribution of model parameters can be used to measure the total uncertainty. The uncertainty information can be used to distinguish the invisible domain, realize the credibility assessment of the aging prediction results, and improve the transparency of the aging prediction model.

[0054] 3. An uncertainty decomposition method is proposed to decompose the total prediction uncertainty into epistemic uncertainty and accidental uncertainty. In terms of the uncertainty caused by changes in invisible load current conditions and invisible noise conditions, the proposed uncertainty decomposition and source tracing method is effective and can help experts analyze whether the uncertainty comes from insufficient understanding of the measurement data or noise interference. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 A schematic flow chart of the uncertainty quantification and decomposition method of the fuel cell aging prediction result of the present invention;

[0056] Figure 2 It is a schematic diagram of the process of uncertainty quantification and decomposition in the present invention. DETAILED DESCRIPTION

[0057] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0058] like Figure 1 and Figure 2 As shown, a method for quantifying and decomposing uncertainty of fuel cell aging prediction results includes the following steps:

[0059] S1) voltage data collection and preprocessing: collecting the voltage data output under the dynamic load cycle condition of PEMFC, performing preprocessing operations to obtain a voltage data set, and dividing the voltage data set into training set data and test set data;

[0060] S2) Modeling and training of B-MIXGU aging prediction model: Replace the model parameters of the mixed gated unit model MIXGU with probability distribution to obtain the Bayesian mixed gated unit model B-MIXGU, obtain the posterior distribution of the B-MIXGU model parameters based on the Bayesian theorem, obtain the output of B-MIXGU according to the posterior distribution, and thus establish the B-MIXGU aging prediction model. Use the variational inference method to approximate the posterior distribution of the model parameters of the B-MIXGU aging prediction model. By solving the minimum KL (Kullback-Leibler) divergence between the variational distribution and the posterior distribution, it is further converted into optimizing the maximum evidence lower bound, and the trained B-MIXGU aging prediction model is obtained by combining the training set data;

[0061] S3) Aging prediction and result analysis: The test set data is input into the trained B-MIXGU aging prediction model to obtain the point estimation and interval estimation aging prediction results of PEMFC, and the aging prediction results are analyzed by the error evaluation index;

[0062] S4) Uncertainty quantification and decomposition: The epistemic uncertainty measured by the randomness of model parameters and the aleatory uncertainty caused by data noise contribute to the total uncertainty together. The variance of interval estimation is used to quantify the total uncertainty of the aging prediction results, and the total uncertainty is decomposed into aleatory uncertainty and epistemic uncertainty to explore the source of uncertainty.

[0063] Wherein, in step B), MIXGU includes: a gated recurrent unit GRU, a minimum gated unit MGU and an adaptive weight adjustment module AWAM, and the expression is as follows:

[0064] GRU:

[0065] z k =σ(W z [h k-1 ,x k ]+b z )

[0066] r k =σ(W r [h k-1 ,x k ]+b r )

[0067]

[0068] In the formula, xk Input information for the current time k, h k-1 is the hidden state at the previous moment, h k1 is the hidden state passed from GRU to AWAM, is the candidate hidden state of GRU, z k is the update gate of GRU, r k is the reset gate of GRU, σ is the sigmoid function, tanh is the tanh function, W z Update the weight matrix of the GRU gate, b z The bias matrix of the GRU update gate, W r Reset the gate weight matrix for GRU, b r The bias matrix for the GRU reset gate, W h1 is the weight matrix of the GRU candidate hidden state, b h1 is the bias matrix of the GRU candidate hidden state;

[0069] MGU:

[0070] f k =σ(W f [h k-1 ,x k ]+b f )

[0071]

[0072] In the formula, f k is the update gate of MGU, h k2 is the hidden state passed from MGU to AWAM, is the candidate hidden state of MGU, W f Update the gate weight matrix for MGU, b f Update gate bias matrix for MGU, W h2 is the weight matrix of the candidate hidden state of MGU, b h2 is the bias matrix of the candidate hidden state of MGU;

[0073] AWAM:

[0074] MIXGU k =σ(W s [h k1 ,h k2 ]+b s )

[0075] h k =(1-MIXGU k )*h k1 +MIXGU k *h k2

[0076] Where, MIXGU k is the weight adjustment module of MIXGU at the current moment k, W s for MIXGU k The corresponding weight matrix of the weight adjustment module, b s for MIXGU k The corresponding bias matrix of the weight adjustment module, h k is the final hidden state output, MIXGU k Decision, when MIXGU k When it approaches 0, h k tend to h k1 , indicating that GRU is dominant, while MIXGU k When it approaches 1, h k tend to h k2 , indicating that MGU occupies a dominant position.

[0077] Specifically, step B) comprises the following steps:

[0078] S201) The model parameters of MIXGU are replaced with probability distribution to obtain B-MIXGU, and the posterior distribution of the B-MIXGU model parameter ω is obtained based on Bayes' theorem:

[0079]

[0080] In the formula, p(ω|D) is the posterior distribution of parameter ω, D is the training set data, p(ω) is the prior distribution of parameter ω, p(D|ω) is the likelihood function of parameter ω, p(D) is the prior distribution of training set data, and parameter ω includes W in the GRU expression z , W r , b z , b r , W h1 , b h1 , W in the MGU expression f , b f , W h2 , b h2 , W in the AWAM expression s , b s ;

[0081] S202) Obtain the output of B-MIXGU according to the posterior distribution, and establish a B-MIXGU aging prediction model:

[0082] p(y|x,D)=∫p(y|x,ω)p(ω,D)dω

[0083] Where x is the input data, y is the model output, p(y|x,D) is the output distribution given the training data D and the input data x, and p(y|x,ω) is the output distribution given the parameter ω and the input data x.

[0084] S203) using a variational inference method to approximate the posterior distribution of the model parameters of the B-MIXGU aging prediction model, and creating a variational distribution q with a parameter θ θ (ω) to approximate the true posterior distribution p(ω|D);

[0085] S204) By solving the minimum KL divergence between the variational distribution and the posterior distribution, it is further transformed into optimizing the maximization of the evidence lower bound ELBO:

[0086] argminKL[q θ (ω)||p(ω|D)]=argmaxELBO;

[0087] S205) Combining the training set data to obtain a trained B-MIXGU aging prediction model.

[0088] In addition, step S4) includes the following steps:

[0089] S401) Epistemic uncertainty measured by the randomness of model parameters and aleatory uncertainty caused by data noise contribute jointly to the total uncertainty;

[0090] p(y|x,D)=∫p(y|x,ω)q θ (ω)dω

[0091] Among them, the randomness of the model parameters is given by the variational distribution q θ (ω) represents, quantifying epistemic uncertainty, and data noise is represented by the probability distribution p(y|x,ω), quantifying aleatoric uncertainty;

[0092] S402) using the variance of interval estimation to quantify the total uncertainty of aging prediction results, decomposing the total uncertainty into aleatory uncertainty and epistemic uncertainty, and exploring the sources of uncertainty;

[0093] D(y)=E(y 2 )-E(y) 2

[0094] =E[E(y 2 |x)]-E[E(y|x)] 2

[0095] =E[D(y|x)+E(y|x) 2 ]-E[E(y|x)] 2

[0096] =E[E(y|x)] 2 -E[E(y|x) 2 ]+E[D(y|x)]

[0097] =D[E(y|x)]+E[D(y|x)]

[0098] Where D(y) represents the total prediction uncertainty, D[E(y|x)] represents the fluctuation of expected E(y|x) due to the variability of model parameters, measuring epistemic uncertainty, and E[D(y|x)] represents the calculation of the average value of D(y|x), eliminating the effects caused by the variability of model parameters, thereby extracting the accidental uncertainty caused by random noise.

[0099] In the process of PEMFC aging prediction, when faced with unknown data, the data-driven method based on machine learning may often give overconfident and unreliable prediction results without clear warnings. Therefore, although uncertainty quantification is crucial to building a credible prediction model, it is often overlooked in the prediction of PEMFC performance degradation. In order to fill the research gap in quantifying and decomposing the uncertainty in the process of PEMFC aging prediction, the present invention proposes a novel Bayesian mixed gating unit model (B-MIXGU) under the framework of probabilistic Bayesian deep learning, which is used for PEMFC performance degradation prediction and uncertainty quantification and decomposition, and provides the credibility of aging prediction results. The aging prediction model based on B-MIXGU combines probabilistic Bayesian theory with MIXGU, uses random probability distribution to replace MIXGU model parameters, and uses variational inference methods to derive the optimal distribution of model parameters, which greatly improves the prediction accuracy, especially when the proportion of training data sets is very low. The uncertainty perception method of PEMFC aging prediction under the probabilistic Bayesian framework uses the variance of interval estimation results to quantify total uncertainty, uses the posterior distribution of model parameters to perceive cognitive uncertainty, and uses the probability distribution of model output to perceive accidental uncertainty, which realizes the credibility assessment of aging prediction results and improves the transparency of aging prediction models. And through the changes in uncertainty components under invisible working conditions and invisible noise conditions, the source of uncertainty in aging prediction results can be traced back, helping experts analyze whether the uncertainty comes from insufficient understanding of measurement data or noise interference.

[0100] In the Bayesian framework, aleatory uncertainty and epistemic uncertainty are integrated into the probability distribution of the prediction results, and then the variance of the interval estimate is used to characterize the total uncertainty, and the contribution of aleatory uncertainty and epistemic uncertainty to the prediction variance is analyzed. This method can effectively reflect the credibility of the prediction results in the prediction of PEMFC performance degradation. High total uncertainty means low credibility of the prediction results, and low total uncertainty means high credibility of the prediction results. Higher epistemic uncertainty indicates a larger distribution difference between the test set and the training set. Higher aleatory uncertainty indicates that there are more unknown quantities in the test set data. In short, by quantifying uncertainty, the prediction results of the model can be better understood, and the prediction credibility analysis results can be combined with human expert judgment to accurately estimate the RUL of PEMFC.

Claims

1. A method for quantifying and decomposing uncertainty of fuel cell aging prediction results, characterized by: The steps include: S1) voltage data collection and preprocessing: collecting the voltage data output under the dynamic load cycle condition of PEMFC, performing preprocessing operations to obtain a voltage data set, and dividing the voltage data set into training set data and test set data; S2) Modeling and training of B-MIXGU aging prediction model: Replace the model parameters of the mixed gated unit model MIXGU with probability distribution to obtain the Bayesian mixed gated unit model B-MIXGU, obtain the posterior distribution of the B-MIXGU model parameters based on the Bayesian theorem, obtain the output of B-MIXGU according to the posterior distribution, and thus establish the B-MIXGU aging prediction model. Use the variational inference method to approximate the posterior distribution of the model parameters of the B-MIXGU aging prediction model, and solve the minimum KL divergence between the variational distribution and the posterior distribution, which is further converted into optimizing the maximization of the lower bound of evidence, and combine the training set data to obtain the trained B-MIXGU aging prediction model; S3) Aging prediction and result analysis: The test set data is input into the trained B-MIXGU aging prediction model to obtain the point estimation and interval estimation aging prediction results of PEMFC, and the aging prediction results are analyzed by the error evaluation index; S4) Uncertainty quantification and decomposition: The epistemic uncertainty measured by the randomness of model parameters and the aleatory uncertainty caused by data noise contribute to the total uncertainty together. The variance of interval estimation is used to quantify the total uncertainty of the aging prediction results, and the total uncertainty is decomposed into aleatory uncertainty and epistemic uncertainty to explore the source of uncertainty.

2. The uncertainty quantification and decomposition method of fuel cell aging prediction results according to claim 1, characterized in that: In the step B), MIXGU includes: a gated recurrent unit GRU, a minimum gated unit MGU and an adaptive weight adjustment module AWAM, and the expression is as follows: GRU: z k =σ(W z [h k-1 ,x k ]+b z ) r k =σ(W r [h k-1 ,x k ]+b r ) In the formula, x k Input information for the current time k, h k-1 is the hidden state at the previous moment, h k1 is the hidden state passed from GRU to AWAM, is the candidate hidden state of GRU, z k is the update gate of GRU, r k is the reset gate of GRU, σ is the sigmoid function, tanh is the tanh function, W z Update the weight matrix of the GRU gate, b z The bias matrix of the GRU update gate, W r Reset the gate weight matrix for GRU, b r The bias matrix for the GRU reset gate, W h1 is the weight matrix of the GRU candidate hidden state, b h1 is the bias matrix of the GRU candidate hidden state; MGU: f k =σ(W f [h k-1 ,x k ]+b f ) In the formula, f k is the update gate of MGU, h k2 is the hidden state passed from MGU to AWAM, is the candidate hidden state of MGU, W f Update the weight matrix of the MGU gate, b f Update gate bias matrix for MGU, W h2 is the weight matrix of the candidate hidden state of MGU, b h2 is the bias matrix of the candidate hidden state of MGU; AWAM: MIXGU k =σ(W s [h k1 ,h k2 ]+b s ) h k =(1-MIXGU k )*h k1 +MIXGU k *h k2 Where, MIXGU k is the weight adjustment module of MIXGU at the current moment k, W s for MIXGU k The corresponding weight matrix of the weight adjustment module, b s for MIXGU k The corresponding bias matrix of the weight adjustment module, h k is the final hidden state output, MIXGU k Decision, when MIXGU k When it approaches 0, h k tend to h k1 , indicating that GRU is dominant, while MIXGU k When it approaches 1, h k tend to h k2 , indicating that MGU occupies a dominant position.

3. The uncertainty quantification and decomposition method of fuel cell aging prediction results according to claim 2, characterized in that: The step B) comprises the following steps: S201) The model parameters of MIXGU are replaced with probability distribution to obtain B-MIXGU, and the posterior distribution of the B-MIXGU model parameter ω is obtained based on Bayes' theorem: In the formula, p(ω|D) is the posterior distribution of parameter ω, D is the training set data, p(ω) is the prior distribution of parameter ω, p(D|ω) is the likelihood function of parameter ω, p(D) is the prior distribution of training set data, and parameter ω includes W in the GRU expression z , W r , b z , b r , W h1 , b h1 , W in the MGU expression f , b f , W h2 , b h2 , W in the AWAM expression s , b s ; S202) Obtain the output of B-MIXGU according to the posterior distribution, and establish a B-MIXGU aging prediction model: p(y|x,D)=∫p(y|x,ω)p(ω,D)dω Where x is the input data, y is the model output, p(y|x,D) is the output distribution given the training data D and the input data x, and p(y|x,ω) is the output distribution given the parameter ω and the input data x. S203) using a variational inference method to approximate the posterior distribution of the model parameters of the B-MIXGU aging prediction model, and creating a variational distribution q with a parameter θ θ (ω) to approximate the true posterior distribution p(ω|D); S204) By solving the minimum KL divergence between the variational distribution and the posterior distribution, it is further transformed into optimizing the maximization of the evidence lower bound ELBO: argminKL[q θ (ω)||p(ω|D)]6argmaxELBO. S205) Combining the training set data to obtain a trained B-MIXGU aging prediction model.

4. The method for quantifying and decomposing uncertainty of fuel cell aging prediction results according to claim 3, characterized in that: The step S4) comprises the following steps: S401) Epistemic uncertainty measured by the randomness of model parameters and aleatory uncertainty caused by data noise contribute jointly to the total uncertainty; p(y|x,D)=∫p(y|x,ω)q θ (ω)dω Among them, the randomness of the model parameters is given by the variational distribution q θ (ω) represents, quantifying epistemic uncertainty, and data noise is represented by the probability distribution p(y|x,ω), quantifying aleatoric uncertainty; S402) using the variance of interval estimation to quantify the total uncertainty of aging prediction results, decomposing the total uncertainty into aleatory uncertainty and epistemic uncertainty, and exploring the sources of uncertainty; D(y)=E(y 2 )-Hey) 2 =E[E(y 2 |x)]-E[E(y|x)] 2 =E[D(y|x)+E(y|x) 2 ]-E[E(y|x)] 2 =E[E(y|x)] 2 -E[E(y|x) 2 ]+E[D(y|x)] =D[E(y|x)]+E[D(y|x)] Where D(y) represents the total prediction uncertainty, D[E(y|x)] represents the fluctuation of expected E(y|x) due to the variability of model parameters, measuring epistemic uncertainty, and E[D(y|x)] represents the calculation of the average value of D(y|x), eliminating the effects caused by the variability of model parameters, thereby extracting the accidental uncertainty caused by random noise.

Images