An environmental risk uncertainty quantification and assessment method for aviation application scenarios

Abstract

The present invention discloses a method for quantitatively evaluating the environmental risk uncertainty in an aviation application scenario, including: determining the quantification object, defining the marginal distribution of environmental parameters, generating random samples, running the model, establishing a likelihood function, optimizing the model, establishing a risk model, and quantitatively evaluating the risk. The present invention accurately quantifies the influence of various factors such as temperature, humidity, air pressure, and vibration in the flight environment on the performance of the fuel cell, provides a scientific basis for the design optimization, performance evaluation, and risk management of the fuel cell system, and thus provides guidance for the application of the fuel cell in the aviation field.

Description

Technical Field

[0001] The present invention belongs to the field of aviation and relates to a method for quantitatively evaluating the environmental risk uncertainty in an aviation application scenario. Background Art

[0002] In the current aviation field, with the improvement of environmental protection requirements and the pursuit of sustainable energy utilization, fuel cells, as a clean energy technology, have the potential to become an important part of the aircraft power system. The efficient energy solution provided by fuel cells is of great significance for achieving low-carbon flight. However, the performance of fuel cells is directly affected by their operating environment. Especially in complex and variable flight environments, such as uncertainty factors like temperature fluctuations, humidity changes, air pressure differences, as well as vibrations and shocks during flight, all of which may have varying degrees of impact on the performance of fuel cells.

[0003] Traditionally, research on fuel cell performance has mainly focused on laboratory tests under stable conditions, lacking performance analysis and verification in actual flight environments. This has limited the engineering application of fuel cells in the aviation field. There is a lack of in-depth understanding of uncertain factors and load spectrum analysis in the flight environment, so it is difficult to formulate effective design schemes or countermeasures to ensure the stability and reliability of fuel cell power systems under complex flight conditions.

[0004] In response to the above challenges, it is necessary to conduct in-depth research on the performance impact of fuel cells under flight conditions and develop a new integrated quantification method, aiming to provide a scientific basis for the design optimization, performance evaluation, and risk management of fuel cell systems through accurate quantification of the impact of various factors such as temperature, humidity, air pressure, and vibration in the flight environment. Summary of the Invention

[0005] The object of the present invention is to provide a method for quantitatively evaluating the environmental risk uncertainty in an aviation application scenario, through accurate quantification of the impact of various factors such as temperature, humidity, air pressure, and vibration in the flight environment, to provide a scientific basis for the design optimization, performance evaluation, and risk management of fuel cell systems, thereby providing guidance for the application of fuel cells in the aviation field.

[0006] The object of the present invention is achieved through the following technical solutions:

[0007] A method for quantitatively evaluating the environmental risk uncertainty in an aviation application scenario includes the following steps:

[0008] Step 1: Determine the quantification object. Establish an environmental parameter set {E1, E2, …, E i , …, E n} under typical working conditions during aircraft operation. Each E iIt contains multiple current environmental description parameters, forming an environmental description set E;

[0009] Step 2: Define the marginal distribution of environmental parameters. Based on the n finite data in E, use the kernel density estimation method (KDE) to solve the marginal distribution of each parameter {x i , x i1 , ……, x i2 , ……, x im} in E. The specific formula is:

[0010]

[0011] where x i1 represents the sample point, n is the number of samples, h is the bandwidth, and K is the kernel function. Through kernel density estimation, the probability distribution and value range of each parameter are obtained;

[0012] Step 3: Generate random samples: Apply the Latin hypercube sampling method (LHS) to randomly sample each parameter to generate a random sample set S to ensure the multi-dimensional uniform distribution of the samples;

[0013] Step 4: Run the model. Use the random sample set S as the input condition X to run the model M to obtain the output Y, where Y includes the performance output P out and the fault points F = {F1, F2, ……, F n}, and statistically obtain the probabilities P F = {P F1 , ……, P Fn} of each fault point;

[0014] Step 5: Establish the likelihood function. Based on the maximum likelihood estimation method (MLE), establish the probability distribution function f out of P o ;

[0015] Step 6: Model optimization. Based on Bayes' theorem, further optimize the likelihood function. The specific formula is:

[0016]

[0017] After multiple iterations, obtain the posterior probability distribution f p of the parameter;

[0018] Step 7: Establish a risk model. Introduce the risk boundary B safe , and the difference P out - B safe between it and the performance output is the performance variation value Mut. From the posterior probability distribution function f out of P p , obtain the probability distribution function f M of the performance variation value, Calculate the expected value E(M), variance σ(M), and extreme value Z of Mut;

[0019] Step 8: Risk quantification assessment. Based on E(M), σ(M), and Z, evaluate the risk R of performance variation risk ; Combine the probability P of the failure point F and the severity of the failure to establish a risk assessment matrix. By means of scoring assessment, divide the risks at the current stage into different levels.

[0020] The beneficial effects of the present invention are:

[0021] By accurately quantifying the effects of various factors in the flight environment, such as temperature, humidity, air pressure, and vibration, on the performance of fuel cells, the present invention provides a scientific basis for the design optimization, performance evaluation, and risk management of fuel cell systems, thereby providing guidance for the application of fuel cells in the aviation field. Brief Description of the Drawings

[0022] Figure 1 is a flowchart of the present invention. Detailed Embodiment

[0023] Taking the quantification of risk uncertainty in the climbing condition of an aircraft by a fuel cell as an example, the present invention: a method for quantifying environmental risk uncertainty in an aviation application scenario will be further described.

[0024] An environmental risk uncertainty quantification assessment method for an aviation application scenario includes the following steps:

[0025] Step 1: Determine the quantification object

[0026] Collect the environmental conditions that the aircraft may encounter during the climbing condition, including: temperature T, pressure P, humidity H, vibration V, inclination angle A, and define the environmental parameters at the i-th time point as E i ={T i , P i , H i , V i , A i}, collect data at n time points to form an environmental description set E = {E1, E2,..., E n};

[0027] Step 2: Define the marginal distribution of parameters

[0028] For each set of parameters T, P, H, V, A, use the kernel density estimation method to determine its marginal distribution f = {f T , f P , f H , f V , f A}, and the specific formula of the kernel density estimation method is:

[0029]

[0030] where x i represents a sample point, n is the number of samples, h is the bandwidth that determines the smoothness of the kernel density estimation, and K is the kernel function. K can be selected from Gaussian kernel function, rectangular kernel function, triangular kernel function, and polynomial kernel function. According to the computational complexity requirements and data characteristics, K and h are comprehensively selected. Or through the method of cross-validation, compare their estimation effects and select the most suitable kernel function. Here, the Gaussian kernel function is selected, that is:

[0031]

[0032] At this time, the formula of the kernel density estimation method is:

[0033]

[0034] Step 3: Generate random samples

[0035] 3.1 Define the marginal probability distribution f of all parameters as the parameter space;

[0036] 3.2 For each item in f, that is, {f T , f P , f H , f V , f A}, divide their probability density functions into m equally probable intervals, which are respectively defined as interval 1 to interval m, where m is equal to the number of samples to be generated. This ensures that the same probability density is covered within each interval;

[0037] 3.3 For a certain interval of a certain parameter, randomly select a value from this interval, so as to uniformly sample within the entire probability distribution range of the parameter;

[0038] 3.4 Repeat step 3.3 for the 5 parameters. Each iteration selects an interval and randomly selects a value from it, ensuring that each parameter only selects one value within its corresponding interval. Finally, a 5-dimensional sample point is generated by combining the selected values of the 5 parameters.

[0039] 3.5 Repeat steps 3.3 and 3.4 to generate m such sample points, that is, the random sample set S.

[0040] Step 4: Run the model

[0041] Using each sample point in S as the input condition X, run the fuel cell performance simulation model M to obtain the output Y, and Y includes the performance output P out and the fault points F = {F 膜干 , F 水淹 , F fc中毒}, the probabilities P of each fault point are statistically obtained F ={P F膜干 , P F水淹 , P F中毒};

[0042] Step 5: Establish the likelihood function

[0043] Based on the maximum likelihood estimation method (MLE), establish the probability likelihood function f out of P O . Taking the likelihood function of the normal distribution as an example, assume that the data follows a normal distribution with a mean of μ and a variance of σ 2 . Then the probability density function of a single observation (x i ) is:

[0044]

[0045] The likelihood function is the product of the probability density functions of all observations, that is:

[0046]

[0047] For simplicity of calculation, take the logarithm of the likelihood function to obtain the log-likelihood function:

[0048]

[0049] Maximize the likelihood function to obtain the maximum likelihood estimates of μ and σ, and obtain the probability distribution function f.

[0050] Step 6: Model optimization

[0051] Based on Bayes' theorem, further optimize the likelihood function. The specific formula is:

[0052]

[0053] After multiple iterations, obtain the posterior probability distribution f P ;

[0054] Step 7: Establish the risk model

[0055] Take B safe = 70%P 预期 as the safety margin. The difference P out - B safe between it and the performance output is the performance variation value Mut. From the posterior probability distribution function f out of P p , obtain the probability distribution function f M of the performance variation value. Calculate the expectation E(M), variance σ(M), and extreme value Z of Mut. The specific formulas are:

[0056]

[0057] σ(M) = E[(M - E(M)) 2 (11)

[0058] Z = |M max - M min | (12)

[0059] Step 8: Risk quantification

[0060] Based on E(M), σ(M), and Z, construct a risk index R in a weighted manner risk (1 ≤ R risk ≤ 8), that is

[0061] R risk = A·E(M) + B·σ(M) + C·Z (13)

[0062] Using R risk and three types of faults to establish a risk assessment matrix. Refer to the fault severity table, and based on the probability P F of each fault point, through a scoring and evaluation method, use R fault to quantify the risk value of the fault point. Taking R fault + R risk as the judgment criterion, divide the risks in the current stage into different levels, such as low risk, general risk, medium risk, and major risk. Table 1 is the table of fault types and their severities, and Table 2 is the risk assessment matrix table.

[0063] Table 1

[0064]

[0065]

[0066] Table 2

[0067]

[0068] Through in-depth research on the performance impact of fuel cells under flight conditions, the present invention proposes a brand-new integrated quantification method. By accurately quantifying the impact of various factors such as temperature, humidity, air pressure, and vibration in the flight environment on the performance of fuel cells, it provides a scientific basis for the design optimization, performance evaluation, and risk management of fuel cell systems.

Claims

1. A quantitative assessment method for environmental risk uncertainty in aviation application scenarios, characterized in that: The following steps are involved: Step 1: Determine the quantification object; establish the environmental parameter set {E1, E2, …, E i ,…,E n }, each E i Contains a set of current environment description parameters, forming an environment description set E; Step 2: Define the marginal distribution of environmental parameters; Based on the n finite data in E, use the kernel density estimation method to solve E i The parameters {x i1 ,x i2 ,……,x im The marginal distribution of} is as follows: where x i1 represents the sample point, n is the number of samples, h is the bandwidth, K is the kernel function, and the probability distribution and value range of each parameter are obtained through kernel density estimation; Step 3: Generate random samples; use Latin hypercube sampling to randomly sample each parameter and generate a random sample set S to ensure multi-dimensional uniform distribution of samples; Step 4: Run the model; use the random sample set S as the input condition X, run the fuel cell performance simulation model Mod, and obtain the output Y, which includes the performance output P out With the fault point F={F1,F2,……,F n }, get the probability of each fault point P by statistics F = {P F1 ,……,P Fn }; Step 5: Establish the likelihood function; Based on the maximum likelihood estimation method, establish P out The probability distribution function f o ; Step 6: Model optimization: Based on Bayesian theorem, further optimize the likelihood function. The specific formula is: After multiple iterations, the parameter posterior probability distribution f is obtained p ; Step 7: Establish risk model; introduce risk boundary B safe , the difference between it and the performance output P out -B safe That is the performance variation value M, which is given by P out The posterior probability distribution function f p , obtain the performance variation probability distribution function f(M), calculate the expectation E(M), variance σ(M), and extreme value Z of M; Step 8: Risk quantification assessment; based on E(M), σ(M), and Z, assess the risk R of performance variation risk ; Comprehensive failure point probability P F Based on the severity of the fault, a risk assessment matrix is ​​established, and the risks at the current stage are divided into different levels through scoring and evaluation.

2. The method for quantitatively assessing environmental risk uncertainty in aviation application scenarios according to claim 1 is characterized in that: In step 2, K selects a Gaussian kernel function, a rectangular kernel function, a triangular kernel function, or a polynomial kernel function.

3. The method for quantitatively assessing environmental risk uncertainty in aviation application scenarios according to claim 1 is characterized in that: In step 3, the details are as follows: 3.1) Define the marginal probability distribution f of all parameters as parameter space; 3.2) For each item in f, that is, {f T , f P , f H , f V , f A }, divide their probability density functions into m intervals of equal probability, defined as interval 1 to interval m, where m is equal to the number of samples to be generated, to ensure that the same probability density is covered in each interval; 3.3) For a certain interval of a parameter, randomly select a value from the interval, so as to sample uniformly in the entire probability distribution range of the parameter; 3.4) Repeat step 3.3) for the five parameters, select an interval and randomly select a value from it in each iteration, ensuring that each parameter only selects a value once in its corresponding interval. Finally, a 5-dimensional sample point is generated by combining the selected values ​​of the five parameters; 3.5) Repeat steps 3.3) and 3.4) to generate m such sample points, namely, the random sample set S.

4. The method for quantitatively assessing environmental risk uncertainty in aviation application scenarios according to claim 1 is characterized in that: In step 7, B safe =70%P 预期 As a safety margin, the difference between it and the performance output is P out -B safe That is the performance variation value M, which is given by P out The posterior probability distribution function f p , get the performance variation probability distribution function f(M), calculate the expectation E(M), variance σ(M), and extreme value Z of M. The specific formula is: σ(M)=E[(ME(M)) 2 ] Z=|M max -M min |。 5. The method for quantitatively assessing environmental risk uncertainty in aviation application scenarios according to claim 1 is characterized in that: In step 8, based on E(M), σ(M), and Z, a risk index R is constructed in a weighted manner. risk (1≤R risk ≤8), that is R risk =A·E(M)+B·σ(M)+C·Z R risk And three types of failures to establish a risk assessment matrix, based on the probability of each failure point P F , through scoring evaluation, using R fault Quantify the risk value of the fault point, expressed as R fault +R risk As the evaluation criteria, the risks at the current stage are divided into different levels: low risk, general risk, medium risk and major risk.

Images