CN-121982141-A - Flight safety assessment method and device based on icing condition inversion and uncertainty quantification

CN121982141ACN 121982141 ACN121982141 ACN 121982141ACN-121982141-A

Abstract

The application relates to a flight safety assessment method and device based on icing condition inversion and uncertainty quantification. The method comprises the steps of obtaining icing observation data of the surface of an aircraft, generating icing type prediction data based on a forward prediction model, defining an observation difference index through the icing observation data and the icing type prediction data, modeling an icing condition inversion problem as a Bayesian inversion problem for solving posterior probability distribution of the flight environment parameters, performing adaptive iterative sampling in a parameter space through an adaptive Monte Carlo sampling algorithm to generate posterior probability distribution of the flight environment parameters, constructing an uncertainty quantization result of the icing conditions based on the posterior probability distribution, and generating a flight safety assessment result through the uncertainty quantization result. The method can obviously improve the credibility and the robustness of the icing simulation and prediction of the aircraft, and has wide application value in the design of aircrafts, the aviation analysis and icing test planning.

Inventors

ZHANG XIAOQUN
DING QIAOQIAO
WANG YUXIN
XU JINGBO
JIN SHI
ZHANG BIN
ZHANG MEIHONG
HAN ZHIRONG

Assignees

上海交通大学

Dates

Publication Date: 20260505
Application Date: 20251126

Claims (10)

1. The flight safety assessment method based on icing condition inversion and uncertainty quantification is characterized by comprising the following steps of: Acquiring icing observation data of the surface of the aircraft; generating ice-type prediction data based on a forward prediction model that generates the ice-type prediction data by denoising diffusion; defining an observation difference index by the icing observation data and the ice type prediction data; modeling an icing condition inversion problem as a Bayesian inversion problem for solving posterior probability distribution of the flight environment parameters, wherein the observation difference index is used as a likelihood function of the Bayesian inversion problem; Performing adaptive iterative sampling in a parameter space by an adaptive Monte Carlo sampling algorithm to generate a posterior probability distribution of the flight environment parameter; constructing an uncertainty quantification result of icing conditions based on the posterior probability distribution; and generating a flight safety assessment result through the uncertainty quantification result.
2. The method as recited in claim 1, further comprising: Constructing the forward prediction model based on a physical driving conditional diffusion generation model; the input of the forward prediction model is a flight environment parameter and a wing reference profile, the flight environment parameter comprises a flight height, an attack angle, an airflow speed, a droplet diameter, a droplet water content and an environment temperature, and the output of the forward prediction model is ice type prediction data.
3. The method of claim 1, wherein generating ice-type prediction data based on a forward prediction model comprises: generating an initial sampling value of a flight environment parameter; and inputting the initial sampling value into a forward prediction model to generate the ice type prediction data.
4. The method of claim 1, wherein defining an observation difference index from the icing observation data and the ice type prediction data comprises: Defining an image-level observation difference index by the icing observation data and the ice type prediction data, wherein the observation difference index comprises RMSE, SSIM, DICE coefficients.
5. The method of claim 1, wherein modeling an icing condition inversion problem as a bayesian inversion problem solving the flight environment parameter posterior probability distribution comprises: The objective function defining the Bayesian inversion problem is: P(θ∣D)∝P(D∣θ)P(θ) Wherein θ represents the flight environment parameters to be inverted, D is icing observation data, P (d|θ) is a likelihood function, and P (θ) represents a priori distribution of the flight environment parameters.
6. The method of claim 1, wherein the adaptive iterative sampling in parameter space by an adaptive monte carlo sampling algorithm to generate the posterior probability distribution of the flight environment parameter comprises: iterative sampling in parameter space by an adaptive Monte Carlo sampling algorithm; The mean value and variance of the proposal distribution are dynamically updated through the historical sample distribution, so that the dynamic self-adaptive adjustment of the sampling area is realized; And when the adaptive Monte Carlo sampling process meets the convergence condition, obtaining posterior probability distribution of the flight environment parameters.
7. The method of claim 6, wherein obtaining the posterior probability distribution of the flight environment parameter when the adaptive monte carlo sampling process satisfies a convergence condition comprises: each sampling point calculates weight according to the likelihood function and performs normalization processing; Resampling the high confidence region samples according to the sample weights to gradually approximate the true posterior probability distribution.
8. The method of claim 1, wherein constructing an uncertainty quantization of icing conditions based on the posterior probability distribution comprises: Based on the posterior probability distribution, the mean value, variance and confidence interval of each parameter in the flight environment parameters are counted, and an uncertainty quantification result of the aircraft icing condition is constructed.
9. The method of claim 1, wherein generating a flight safety assessment result from the uncertainty quantization result comprises: Generating a routing graph from the uncertainty quantization result, and/or Generating a weather avoidance maneuver by the uncertainty quantization result, and/or Generating anti-icing structural optimization data from the uncertainty quantization result, and/or And generating an icing condition judgment index through the uncertainty quantification result.
10. A flight safety assessment device based on icing condition inversion and uncertainty quantification, comprising: the observation module is used for acquiring icing observation data of the surface of the aircraft; The prediction module is used for generating ice type prediction data based on a forward prediction model, and the forward prediction model generates the ice type prediction data through denoising diffusion; the index module is used for defining an observation difference index through the icing observation data and the icing prediction data; The modeling module is used for modeling an icing condition inversion problem as a Bayesian inversion problem for solving posterior probability distribution of the flight environment parameters, wherein the observation difference index is used as a likelihood function of the Bayesian inversion problem; The iteration module is used for carrying out self-adaptive iterative sampling in a parameter space through a self-adaptive Monte Carlo sampling algorithm so as to generate posterior probability distribution of the flight environment parameter; the quantization module is used for constructing an uncertainty quantization result of the icing condition based on the posterior probability distribution; And the evaluation module is used for generating a flight safety evaluation result through the uncertainty quantification result.

Description

Flight safety assessment method and device based on icing condition inversion and uncertainty quantification Technical Field The application relates to the field of computer information processing, in particular to a flight safety assessment method and device based on icing condition inversion and uncertainty quantification. Background In the flight process of an aircraft, when the outside air temperature is low and supercooled water drops are contained, the positions of wings, tail wings, engine air inlets and the like are easy to generate icing. Icing not only can change the aerodynamic profile of the wing, disrupt airflow adhesion, reduce lift and increase drag, but can also lead to engine failure, instrument errors, and even flight accidents. Therefore, the method accurately predicts and analyzes the icing process and the uncertainty thereof of the aircraft, and has important significance for flight safety guarantee and anti-icing and deicing system design. Currently, aircraft icing research mainly comprises two major directions, namely a positive problem and a negative problem. The positive problem is to predict icing morphology and thickness distribution based on known environmental conditions (such as droplet diameter, water content, angle of attack, velocity, temperature, etc.), using physical models or numerical simulation methods. Common methods include icing simulation calculations based on thermal equilibrium equations, droplet impact models, and freeze fraction models. The inverse problem is to infer the environmental parameters that may lead to the icing morphology in the reverse direction based on known icing results (experimental observations of ice types or icing images). The process involves solving a high-dimensional nonlinear equation, and has the problems of multiple solutions, noise sensitivity, uncertainty propagation and the like. The existing anti-problem research often depends on a deterministic optimization method (such as least square inversion, gradient descent, evolutionary algorithm and the like), the result of the method usually depends on an initial value, is easy to sink into local optimum, probability distribution information of input parameters is difficult to obtain, and uncertainty sources and propagation paths of icing conditions cannot be comprehensively reflected. In addition, the traditional icing model shows remarkable prediction volatility under the disturbance of environmental parameters, and the lack of robustness and credibility assessment mechanism leads to difficulty in quantitative analysis of the stability of the model in the model verification process. In recent years, a Bayesian inference-based uncertainty quantization method has been introduced into complex engineering systems. However, the scheme related to the inversion of the icing condition and the uncertainty analysis of the airplane in the prior art still cannot be solved, and the defects of insufficient precision, low model stability, difficulty in quantifying the uncertainty and the like exist. The above information disclosed in the background section is only for enhancement of understanding of the background of the application and therefore it may contain information that does not form the prior art that is already known to a person of ordinary skill in the art. Disclosure of Invention In view of the above, the application provides a flight safety assessment method and a flight safety assessment device based on icing condition inversion and uncertainty quantification, which can remarkably improve the reliability and robustness of aircraft icing simulation and prediction, and have wide application value in aircraft design, airworthiness pre-analysis and icing test planning. Other features and advantages of the application will be apparent from the following detailed description, or may be learned by the practice of the application. According to one aspect of the application, a flight safety assessment method based on icing condition inversion and uncertainty quantification is provided, and the method comprises the steps of obtaining icing observation data of an aircraft surface, generating icing prediction data based on a forward prediction model, generating the icing prediction data through noise elimination diffusion, defining an observation difference index through the icing observation data and the icing prediction data, modeling an icing condition inversion problem as a Bayesian inversion problem for solving posterior probability distribution of flight environment parameters, wherein the observation difference index is used as a likelihood function of the Bayesian inversion problem, performing adaptive iterative sampling in a parameter space through an adaptive Monte Carlo sampling algorithm to generate posterior probability distribution of the flight environment parameters, constructing an uncertainty quantification result of the icing conditions based on the posterior probability distribut