High-altitude ice crystal detection method based on genetic algorithm

Through genetic algorithms, the radar reflectivity factor and high-altitude ice water content model are optimized, and the problem of large error in ice water content estimation in the existing technology is solved, and high-accurate ice water content estimation and ice crystal hazard level quantification are achieved, which improves high-altitude ice crystals Reliability of inspection.

CN114492202BActive Publication Date: 2024-11-29CIVIL AVIATION UNIV OF CHINA
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
CN202210140853.X
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-02-16
Publication Date
2024-11-29
Estimated Expiration
2042-02-16

AI Technical Summary

Technical Problem

The existing radar reflectivity factor and high-altitude ice water content relationship model have large errors and poor robustness, making it difficult to accurately characterize complex ice water content estimation, especially under different data conditions, which leads to inaccurate ice water content estimation detected by high-altitude ice crystals. .

Method used

Genetic algorithm is used to finely optimize unknown parameters in the relationship model of radar reflectivity factor and high-altitude ice water content, establish an accurate ice water content estimation model, and preprocess and segment the training and test sets through airborne meteorological radar data. , determine the ice crystal hazard level.

Benefits of technology

Through the genetic algorithm optimization parameters, the estimation accuracy of ice water content and the quantification ability of ice crystal hazard levels are significantly improved, errors are reduced, and reliability of high-altitude ice crystal detection is improved.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN114492202B_ABST
    Figure CN114492202B_ABST
Patent Text Reader

Abstract

A high-altitude ice crystal detection method based on genetic algorithm. It includes the following steps: using the altitude and the atmospheric temperature to determine the airspace where ice crystals are distributed, pre-processing the radar reflectivity factor of the airspace where ice crystals are distributed received by the airborne meteorological radar; establishing a relationship model between the original radar reflectivity factor containing unknown parameters and the ice water content of high-altitude ice crystals; using the genetic algorithm to perform fine optimization on the unknown parameters in the model to obtain an accurate relationship model between the radar reflectivity factor and the ice water content of high-altitude ice crystals; inputting the radar reflectivity factor of the airspace where ice crystals are to be detected into the accurate relationship model between the radar reflectivity factor and the ice water content of high-altitude ice crystals to obtain an estimated value of the ice water content of the airspace, and finally determining the ice crystal hazard level according to the estimated value of the ice water content. The method of the present invention can effectively estimate the ice water content of high-altitude ice crystals and quantify the ice crystal hazard level. Thus, high-altitude ice crystal detection is effectively completed.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention belongs to the technical field of meteorological radar signal processing, and in particular relates to a high-altitude ice crystal detection method based on a genetic algorithm. Background Art

[0002] When an aircraft is flying in the cruising phase (generally above 22,500 feet), high-altitude ice crystals are a weather phenomenon that poses a greater threat to aircraft flight safety. In airspace with high ice crystal accumulation, ice crystal accumulation on the surface of aircraft engine blades can cause stall and surge, flameout and damage to the combustion chamber, etc. At the same time, ice crystals can also enter the atmospheric data sensor and cause blockage, thereby making the relevant sensor data abnormal, resulting in the aircraft losing control. In recent years, aviation accidents caused by encountering high-altitude ice crystals have occurred from time to time. The Federal Aviation Administration of the United States has conducted statistics on aircraft engine air accidents that have occurred in the past two decades, and the results show that most of these accidents are related to high-altitude ice crystals. These real events show that high-altitude ice crystals will have a serious impact on the aviation transportation industry and passenger flight safety. Therefore, it is very important to carry out research on high-altitude ice crystal detection. The estimation of the ice water content of high-altitude ice crystals is the basis of the entire detection process, which directly affects the accuracy of the subsequent hazard level determination. Therefore, the research on the estimation of the ice water content of high-altitude ice crystals has important practical significance.

[0003] Airborne weather radars can detect dangerous weather conditions in the airspace ahead of the aircraft. These conditions can be used as an important reference for flight warnings or avoiding dangerous weather. Generally, airborne weather radars detect precipitation targets based on the intensity of the radar echo, and believe that the greater the echo intensity, the more water it contains, and the greater the impact on flight safety. The detection of high-altitude ice crystals and the quantification of their danger levels can be indicated by the ice water content IWC (Ice Water Content) of the detected space, that is, the ice crystal density per unit volume of space. The ice water content IWC refers to the mass of solid water contained in a unit volume of air, expressed in g / m 3 .

[0004] Foreign scholars have proposed that the value of ice water content IWC can be estimated by inverting the radar reflectivity factor Z based on the statistical relationship between the radar reflectivity factor Z and the ice water content IWC, combined with the empirical formula. In 1954, Atlas of the Cambridge Research Center in the Air obtained the empirical relationship between the radar reflectivity factor Z-particle distribution-ice water content based on the spectral parameter data measured by the aircraft; in 1992, Liao et al. in the United States studied the relationship between the radar reflectivity factor Z-ice water content IWC of Ka-band radar and W-band radar; in 1995, Brown used millimeter-wave airborne radar to detect the radar reflectivity factor Z-ice water content IWC empirical formula of ice crystal cloud; in 2001, Austin and Stephens combined numerical simulation methods to A method for inverting ice water content (IWC) based on millimeter-wave radar reflectivity factor Z and visible light layer cloud microphysical parameters was proposed; in 2004, Matrosov discussed the method of estimating the water content of marine layer clouds based on radar reflectivity factor Z, and compared the changes in water content in clouds when different precipitation cloud and non-precipitation cloud intensity limits were taken; in 2015, Protat of the Atmospheric Medium Space Observation Laboratory gave the radar reflectivity factor Z-ice water content IWC relationship under 35GHz and 95GHz radars, respectively, and statistically compared the experimental errors. However, due to the fact that the empirical formula is too simple, it has great limitations, so it is difficult to accurately characterize the complex radar reflectivity factor Z-ice water content IWC relationship model, especially when using different data to estimate ice water content, there will be large errors and poor robustness. Therefore, how to obtain an accurate radar reflectivity factor Z-ice water content IWC relationship model is an urgent problem to be solved for accurate estimation of ice water content of high-altitude ice crystals.

[0005] For the radar reflectivity factor Z-ice water content IWC relationship model, which has a relatively complex mathematical model and requires many parameters to be determined, using some intelligent algorithms to perform adaptive parameter estimation to obtain an accurate model is a new solution to the problem. Genetic algorithm is an algorithm that has significant effects and strong robustness in solving various complex optimization problems. Since 1995, domestic research on genetic algorithms has also increased. As a member of heuristic search algorithms, genetic algorithms have made great progress and development in various fields such as signal processing, optimal control, and production planning in recent decades. When solving function optimization problems, compared with other optimization algorithms, the biggest feature of genetic algorithms is that the processing object is not the parameter itself, but the gene individual encoded by the parameter set, so it is not constrained by the continuity of the function. In addition, the algorithm does not require auxiliary information such as the gradient information of the problem. The optimal individual can be selected only based on the numerical value of the fitness function, and the fitness function is not constrained by continuous differentiability, so the scope of application is greatly expanded. For the problem of estimating the ice water content of high-altitude ice crystals, if genetic algorithms can be used for parameter estimation and their robustness can be brought into play, an accurate radar reflectivity factor Z-ice water content IWC relationship model can be obtained, which will have great application value. Summary of the invention

[0006] In order to solve the above problems, the object of the present invention is to provide a high-altitude ice crystal detection method based on genetic algorithm.

[0007] In order to achieve the above object, the high-altitude ice crystal detection method based on genetic algorithm provided by the present invention comprises the following steps performed in sequence:

[0008] 1) The ice crystal distribution airspace is determined by using the altitude and atmospheric temperature, and then the radar reflectivity factor of the ice crystal distribution airspace received by the airborne meteorological radar is preprocessed to obtain the preprocessed radar reflectivity factor and randomly divide it into a training set and a test set according to the proportion;

[0009] 2) Establish a relationship model between the original radar reflectivity factor with unknown parameters and the ice water content of high-altitude ice crystals;

[0010] 3) inputting the training set obtained in step 1) into the relationship model between the original radar reflectivity factor and the ice water content of high-altitude ice crystals obtained in step 2), and then using a genetic algorithm to finely optimize the unknown parameters in the relationship model between the original radar reflectivity factor and the ice water content of high-altitude ice crystals, and then inputting the test set obtained in step 1) for testing, and finally obtaining an accurate relationship model between the radar reflectivity factor and the ice water content of high-altitude ice crystals;

[0011] 4) Input the radar reflectivity factor of the airspace where ice crystals are to be measured into the above-mentioned accurate radar reflectivity factor and the ice water content relationship model of high-altitude ice crystals to obtain the estimated value of the ice water content in the airspace, and finally determine the ice crystal hazard level according to the estimated value of the ice water content.

[0012] In step 1), the ice crystal distribution airspace is set to an airspace with an altitude of more than 22,500 feet and an atmospheric temperature between -50°C and 20°C; the radar reflectivity factor is preprocessed by selecting an appropriate range as the radar reflectivity factor threshold and removing abnormal values ​​in the received radar reflectivity factor; the ratio of the training set to the test set is 7:3.

[0013] In step 2), the relationship model between the original radar reflectivity factor containing unknown parameters and the ice water content of high-altitude ice crystals is:

[0014]

[0015] Where IWC is the ice water content of high-altitude ice crystals; Z is the radar reflectivity factor; a is the parameter characterizing the composition of ice crystal particles in the ice crystal mass power law distribution function m(D); N 0 represents the parameter characterizing the size of ice crystal particles in the ice crystal particle number concentration distribution function N(D); μ represents the parameter determining the distribution shape in the ice crystal particle number concentration distribution function N(D); the last three parameters are unknown parameters.

[0016] In step 3), the method of using genetic algorithm to finely optimize the unknown parameters in the relationship model between the original radar reflectivity factor and the ice water content of high-altitude ice crystals is: all unknown parameters in the relationship model between the original radar reflectivity factor and the ice water content of high-altitude ice crystals are encoded separately, then an initial population is generated, and then selection and mutation operations are performed, and then the optimal solution is selected according to the fitness function, and the optimal parameters are obtained after decoding, and finally the optimal parameters are brought into the relationship model between the original radar reflectivity factor and the ice water content of high-altitude ice crystals to obtain an accurate relationship model between the radar reflectivity factor and the ice water content of high-altitude ice crystals.

[0017] In step 4), the method for determining the ice crystal hazard level according to the estimated value of the ice water content is as follows: according to the European Union's "Feasibility Study Report on Ice Crystal Detection by Weather Radar", the ice crystal hazard level is divided into three categories according to the estimated value of the ice water content: mild, moderate and severe. 3 When the ice crystal hazard level is mild, the estimated value of ice water content is between 1g / m 3 With 3g / m 3 When the ice crystal hazard level is between 20 and 30 °C, the ice crystal hazard level is moderate, and the estimated ice water content is greater than 3 g / m 3 The ice crystal hazard level is severe.

[0018] The high-altitude ice crystal detection method based on genetic algorithm provided by the present invention is to first establish a relationship model between radar reflectivity factor and ice water content of high-altitude ice crystals, and then use genetic algorithm to finely optimize the unknown parameters in the model based on this model, and after obtaining an accurate relationship model between radar reflectivity factor and ice water content, estimate the ice water content of high-altitude ice crystals and quantify the ice crystal hazard level. Through simulation data verification, this method can effectively estimate the ice water content of high-altitude ice crystals and quantify the ice crystal hazard level. Thereby effectively completing high-altitude ice crystal detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 A flow chart of the high-altitude ice crystal detection method based on genetic algorithm provided by the present invention.

[0020] Figure 2 This is a flow chart of using genetic algorithm to estimate unknown parameters in the present invention.

[0021] Figure 3 The following are the comparison diagrams of the experimental results of ice water content estimation using this method and the classical empirical formula method for different altitude data at 14:00 on March 17, 2016. (a) is the 37.0km altitude layer, (b) is the 7.5km altitude layer, and (c) is the 8.1km altitude layer.

[0022] Figure 4 The following are comparison diagrams of the experimental results of ice water content estimation using this method and the classical empirical formula method for different altitude data at 16:00 on March 17, 2016. (a) is the 37.0km altitude layer, (b) is the 7.5km altitude layer, and (c) is the 8.1km altitude layer.

[0023] Figure 5 The following is a scatter plot of the correlation between the true value of ice water content at different altitudes at 14:00 on March 17, 2016 and the estimated value obtained using this method. (a) is the 37.0km altitude layer, (b) is the 7.5km altitude layer, and (c) is the 8.1km altitude layer.

[0024] Figure 6 The following is a scatter plot of the correlation between the true value of ice water content at different altitudes at 16:00 on March 17, 2016 and the estimated value obtained using this method. (a) is the 37.0km altitude layer, (b) is the 7.5km altitude layer, and (c) is the 8.1km altitude layer.

[0025] Figure 7The error distribution histogram of data at different altitudes when this method was used at 14:00 on March 17, 2016. (a) is the 37.0km altitude, (b) is the 7.5km altitude, and (c) is the 8.1km altitude.

[0026] Figure 8 The error distribution histogram of data at different altitudes when this method was used at 16:00 on March 17, 2016. (a) is the 37.0km altitude, (b) is the 7.5km altitude, and (c) is the 8.1km altitude.

[0027] Fig. 9 The following is a graph of ice crystal detection results based on the estimated value of ice water content at different altitudes in Oklahoma, USA at 14:00 on March 17, 2016. (a) is at the 37.0km altitude, (b) is at the 7.5km altitude, and (c) is at the 8.1km altitude.

[0028] Fig.10 The following is a graph of ice crystal detection results based on the estimated value of ice water content at different altitudes in Oklahoma, USA at 16:00 on March 17, 2016. (a) is the 37.0km altitude layer, (b) is the 7.5km altitude layer, and (c) is the 8.1km altitude layer.

[0029] Fig.11 This is the distribution map of ice crystal hazard level at 14:00 on March 17, 2016 when this method was used. (a) is the 37.0km altitude layer, (b) is the 7.5km altitude layer, and (c) is the 8.1km altitude layer.

[0030] Fig.12 This is the distribution map of ice crystal hazard level at 16:00 on March 17, 2016 when this method was used. (a) is the 37.0km altitude layer, (b) is the 7.5km altitude layer, and (c) is the 8.1km altitude layer. DETAILED DESCRIPTION

[0031] The high-altitude ice crystal detection method based on genetic algorithm provided by the present invention is described in detail below with reference to the accompanying drawings and specific examples.

[0032] like Figure 1 As shown, the high-altitude ice crystal detection method based on genetic algorithm provided by the present invention comprises the following steps performed in sequence:

[0033] 1) The ice crystal distribution airspace is determined by using the altitude and atmospheric temperature, and then the radar reflectivity factor of the ice crystal distribution airspace received by the airborne meteorological radar is preprocessed to obtain the preprocessed radar reflectivity factor and randomly divide it into a training set and a test set according to the proportion;

[0034] The specific method is as follows:

[0035] Atmospheric environments with high concentrations of ice crystals usually exist in deep convective storms in the tropics and subtropics. The distribution height of ice crystals in the high altitude is mainly above 22,500 feet, and the atmospheric temperature is between -50℃ and 20℃, where the supercooled water will decrease and the specific gravity of ice crystal particles will increase. Studies have shown that convective updrafts associated with large convective environments can usually bring high concentrations of supercooled liquid water to high altitudes and freeze them into very small ice crystals. Therefore, weather conditions at altitudes above 30,000 feet are generally considered to contain a large number of completely frozen solid ice crystal particles, which contain almost no supercooled liquid water. Therefore, the present invention sets the airspace above 22,500 feet above sea level and between -50℃ and 20℃ as the ice crystal distribution airspace.

[0036] The altitude and the atmospheric temperature around the aircraft can be detected by the altitude and temperature sensors on the aircraft. The atmospheric temperature changes with the altitude, so the atmospheric temperature can also be inverted from the altitude. The specific operation process of inverting the atmospheric temperature using the altitude is to use the vertical temperature decrease rate and the stepwise iteration method of the inverse distance weight to revise the atmospheric temperature at a point in the airspace to the altitude at the point, thereby inverting the atmospheric temperature at the point.

[0037] In meteorology, the formula for the change of atmospheric temperature with altitude is based on sea level. The atmospheric temperature decreases by a certain amount for every 1,000 meters increase in altitude. The specific formula for the change of atmospheric temperature T with altitude is: Among them, T 0 is the average atmospheric temperature around the aircraft, in °C; δ is the vertical temperature gradient; h is the height difference between the required airspace and the aircraft altitude, in m.

[0038] Most of the ice crystals in the high altitude exist in the troposphere. The atmospheric temperature in the troposphere decreases with the increase of altitude. Taking the troposphere as an example for analysis, under standard atmospheric pressure, the vertical temperature lapse rate δ is usually about 6.5℃ / 1000m. According to the vertical temperature lapse rate δ, the average atmospheric temperature at sea level in the equatorial region is used as the benchmark, and the vertical temperature lapse rate δ is used. The stepwise iteration method of inverse distance weighting can be used to calculate the atmospheric temperature using the altitude. However, since the vertical temperature lapse rate δ will change with the change of geographical regions, seasons and atmospheric altitude layers, and is not a fixed value, in practical applications, the selection of the vertical temperature lapse rate δ should be determined according to the specific situation.

[0039] In addition, in practical applications, the airspace where ice crystals are distributed can also be determined by setting an altitude threshold and a temperature threshold; if the altitude of a certain airspace is not higher than the set altitude threshold or is not lower than the set temperature threshold, then ice crystals will not exist in the airspace. In the present invention, the altitude threshold is set to 22,500 feet and the temperature threshold is set to -40°C.

[0040] Airborne weather radar can provide radar reflectivity signal intensity echo information. The radar reflectivity factor Z from ice crystals is generally lower than 30dBZ. However, the weather target information reflected by the radar reflectivity factor Z that is too low is of no practical significance, and the radar echo is easily affected by the non-meteorological echo in the atmosphere. Therefore, the radar reflectivity factor Z actually received often contains abnormal values ​​caused by interference such as noise. Therefore, it is necessary to select an appropriate range as the radar reflectivity factor threshold to remove the abnormal values ​​in the received radar reflectivity factor Z, that is, to pre-process the received radar reflectivity factor Z. In the present invention, the radar reflectivity factor threshold is set to -10dBZ-30dBZ.

[0041] Finally, the preprocessed radar reflectivity factor Z is randomly divided into training set and test set in a ratio of 7:3.

[0042] 2) Establish a relationship model between the original radar reflectivity factor with unknown parameters and the ice water content of high-altitude ice crystals (Z-IWC relationship model);

[0043] The specific method is as follows:

[0044] According to the research of Heymsfield et al., the expressions of the equivalent radar reflectivity factor Z and ice water content IWC of ice crystal particle group are shown in formula (1) and formula (2) respectively:

[0045]

[0046]

[0047] in, is the reflectivity-weighted average density of total ice crystal particles; ρ 0 is the density of liquid water, and it is usually considered that ρ 0 =1kg / m 3 ; The coefficient 1.09 is the ratio of the density of liquid water to the density of solid water; K i / K w is the conversion factor, which converts the radar reflectivity relative to solid ice into the equivalent ice crystal radar reflectivity. It is generally considered that (K i / K w ) 2=0.174 / 0.93; a and b are two parameters in the ice crystal mass power law distribution function m(D) that characterize the composition of ice crystal particles. When the ice crystal particles are approximately spherical, parameter b=3; N 0 is the parameter characterizing the size of ice crystal particles in the ice crystal particle number concentration distribution function N(D); λ is the slope parameter, μ is the parameter determining the distribution shape in the ice crystal particle number concentration distribution function N(D); ρ e is the effective density of ice crystal particles; is the overall average mass-weighted effective density of ice crystal particles; D is the maximum ice crystal particle diameter.

[0048] However, Heymsfield et al. did not directly give a model for the relationship between the radar reflectivity factor and the ice water content of high-altitude ice crystals. Since both formula (1) and formula (2) contain the slope parameter λ, the two formulas can be combined using the common slope parameter λ to obtain a model for the relationship between the radar reflectivity factor and the ice water content of high-altitude ice crystals. By transforming formula (1) and proposing the slope parameter λ, we can obtain:

[0049]

[0050] By transforming formula (2) and proposing the slope parameter λ, we can obtain:

[0051]

[0052] Combining formula (3) with formula (4) yields:

[0053]

[0054] By transforming formula (5), we can get:

[0055]

[0056] The parameters b, liquid water density ρ, and ice crystal particle composition in the determined ice crystal mass power law distribution function m(D) are 0 , Conversion coefficient K i / K w Substituting into formula (6), we can obtain the relationship model between the original radar reflectivity factor and the ice water content of high-altitude ice crystals shown in formula (7):

[0057]

[0058] The relationship model contains the parameter a in the ice crystal mass power law distribution function m(D) that represents the composition of the ice crystal particles, and the parameter N in the ice crystal particle number concentration distribution function N(D) that represents the size of the ice crystal particles. 0The three unknown parameters, μ, which determines the distribution shape in the ice crystal particle number concentration distribution function N(D), can be used to obtain an accurate relationship model between the radar reflectivity factor and the ice water content of high-altitude ice crystals.

[0059] 3) inputting the training set obtained in step 1) into the relationship model between the original radar reflectivity factor and the ice water content of high-altitude ice crystals obtained in step 2), and then using a genetic algorithm to finely optimize the unknown parameters in the relationship model between the original radar reflectivity factor and the ice water content of high-altitude ice crystals, and then inputting the test set obtained in step 1) for testing, and finally obtaining an accurate relationship model between the radar reflectivity factor and the ice water content of high-altitude ice crystals;

[0060] Genetic algorithms are suitable for solving parameter estimation problems due to their good parallelism, strong versatility and robustness, and a wide range of applications. When genetic algorithms are used for parameter estimation problems, their most obvious advantages include being able to handle extremely complex optimization function solution problems without being constrained by function continuity; being able to search for multiple solutions at the same time, with parallelism; at the same time, genetic algorithms do not require auxiliary information, and can obtain the optimal solution based solely on the numerical size of the fitness function, thus greatly expanding their scope of application.

[0061] like Figure 2 As shown in the figure, the specific method of using genetic algorithm to finely optimize the unknown parameters in the relationship model between the original radar reflectivity factor and the ice water content of high-altitude ice crystals is as follows:

[0062] ① Coding and population initialization: Parameter estimation is performed using a genetic algorithm. First, the solution to the problem to be optimized is encoded. The purpose of encoding is mainly to change the form of expression of the solution to the optimization problem to facilitate genetic operations. The present invention uses binary coding to encode the parameter a that characterizes the composition of ice crystal particles in the ice crystal mass power law distribution function m(D) in the relationship model between the original radar reflectivity factor and the ice water content of high-altitude ice crystals, and the parameter N that characterizes the size of ice crystal particles in the ice crystal particle number concentration distribution function N(D). 0 The three unknown parameters, namely, the parameter μ that determines the distribution shape in the ice crystal particle number concentration distribution function N(D), are encoded separately, and each chromosome (solution) is encoded as a genotype. After the encoding operation is completed, the initial population needs to be generated. For each unknown parameter in the relationship model between the original radar reflectivity factor and the ice water content of high-altitude ice crystals, the present invention generates an initial population in the form of random generation. In the present invention, 200 individuals with uniform distribution are randomly generated as the initial population;

[0063] ② Selection of fitness function: The convergence speed of the genetic algorithm and whether it can find the optimal solution directly depend on the selection of the fitness function. The fitness function is basically determined by the objective function of the optimization problem. Once the fitness function is determined, the law of natural selection determines which chromosomes are suitable for survival and which are eliminated by the probability distribution determined by the fitness value. The surviving individuals form a new population and form a group that can reproduce the next generation. In order to make the estimated value of the ice water content closer to the actual value of the ice water content, the present invention constructs an objective function G (a, N) that meets the following optimization criteria: 0 , μ) as the fitness function:

[0064]

[0065] In the formula, IWC(a, N 0 , μ) is the ice water content calculated using parameters including air density and water vapor mixing ratio of meteorological scene data, and is used as the true value of ice water content. 0 , μ) is the estimated value of ice water content obtained by using the relationship model between the original radar reflectivity factor and the ice water content of high-altitude ice crystals shown in formula (7), and m is the number of samples used.

[0066] ③Genetic selection: The first thing that individuals in a group must undergo is the selection operation. The purpose of the selection operation is to select excellent individuals from the group and pass them on to the next generation in order to retain high-quality genes and eliminate inferior individuals. Through the selection operation, it can be determined which individuals can be retained and passed on to the next generation. The role of the selection operation is to improve the convergence of the algorithm and improve computational efficiency, because genetic selection can retain useful information from the parent generation. The quality of the genetic algorithm selection operation depends on the selection of genetic operators. If the genetic operator is not properly selected, it will increase the similarity of individuals in the group, so that the offspring will be similar to the parent individuals, which will eventually affect the direction of group development and cause the genetic algorithm process to stagnate. Therefore, the genetic algorithm will lose diversity, and then the problem of premature maturity will occur. The genetic selection operation in the present invention is based on the fitness evaluation of individuals in the group, and adopts a probability-based (G i The offspring are selected by the roulette wheel method with k = 1, 2, L, N, N being the total number of individuals (where L is the individual fitness value and k = 1, 2, L, N, N being the total number of individuals).

[0067] ④Genetic crossover and mutation operations: The convergence of genetic algorithms mainly depends on the two core operations of mutation and crossover. The essence of the mutation operation is to explore the diversity of individuals in the group and improve the local random search ability of the algorithm. The crossover operation determines the speed of generating new individuals and plays a key role in realizing the global search performance of the algorithm. Therefore, the combination of mutation and crossover jointly completes the local search and global search of the search space, enabling the genetic algorithm to complete the optimization process of the optimization problem with good search performance. In order to better complete the crossover and mutation operations and thereby obtain the optimal solution as much as possible, the present invention sets the crossover probability to 0.6 and the mutation probability to 0.08.

[0068] ⑤ Iteration termination: The present invention sets the evolutionary algebra counter t = 0, sets the maximum evolutionary algebra T = 100, and sets the operation termination condition to be if the target value of the objective function of the optimization criterion satisfies G(x) ≤ G 0 =0.01, or the number of iterations is T = 100, the genetic evolution process is stopped, and the best individual of the last generation is selected as the optimal solution of the unknown parameters, and finally an accurate relationship model between the radar reflectivity factor and the ice water content of high-altitude ice crystals is obtained.

[0069] 4) Input the radar reflectivity factor of the airspace where ice crystals are to be measured into the above-mentioned accurate radar reflectivity factor and the ice water content relationship model of high-altitude ice crystals to obtain the estimated value of the ice water content in the airspace, and finally determine the ice crystal hazard level according to the estimated value of the ice water content.

[0070] The estimated value of ice water content, which is a quantitative indicator of ice crystals, can represent the degree of danger in the airspace where a certain ice crystal is located. Generally, the higher the ice water content, the faster the aircraft engine power loss, the lower the safety, and the greater the danger. Therefore, according to the European Union's "Feasibility Study Report on Ice Crystal Detection by Weather Radar", the ice crystal hazard level is divided into three categories: mild, moderate and severe according to the estimated value of ice water content: when the estimated value of ice water content is less than 1g / m 3 When the ice crystal hazard level is mild, the estimated value of ice water content is between 1g / m 3 With 3g / m 3 When the ice crystal hazard level is between 20 and 30 °C, the ice crystal hazard level is moderate, and the estimated ice water content is greater than 3 g / m 3 The ice crystal hazard level is severe.

[0071] Low concentrations of ice crystals do not pose a significant threat to flight safety. An analysis of the EU engine event database and data from flight test activities conducted so far shows that ice water concentrations of 1 g / m 3This may be the representative threshold of moderate ice crystals that defines most aircraft engine events that occur in flight service. Flying in ice crystals below this content does not pose a significant threat to flight safety, but flying in ice crystals above this content is prone to the risk of ice accumulation in aircraft engines. Therefore, ice crystal content below this threshold is considered light ice crystals. In addition, the FAA's analysis of data generated from the Airbus detection event database and flight test activities shows that ice water content of 3g / m 3 This is probably the representative threshold for heavy ice crystals that defines most detection events that occur during flight test campaigns. High levels of ice crystals can not only affect aircraft engines, but can also cause detector failures. In addition, ice water concentrations between 1 g / m 3 With 3g / m 3 The ice crystals in between are moderate ice crystals, which will pose a certain threat to aircraft engines.

[0072] In order to verify the effect of the present method, the inventor used the output data of the numerical weather forecast model of a real meteorological scene in the range of 95° to 100° west longitude and 30° to 34° north latitude in Oklahoma, the United States on March 17, 2016 as samples for detection and analysis, and selected the data at two time points, 14:00 on March 17, 2016 and 16:00 on March 17, 2016 in Oklahoma for the experiment.

[0073] According to the analysis of the altitude of ice crystals and atmospheric temperature, the data of 7.0km, 7.5km and 8.1km altitude layers were selected for experiments. In order to verify the effectiveness and accuracy of this method, Figure 3 and Figure 4 The experimental results of ice water content estimation using this method and the classical empirical formula method are compared for the data of different altitude layers at two different time points. The empirical formula IWC=0.064Z proposed by Atlas et al. in 1954 is used. 0.58 , In 1997, Aydin et al. proposed the empirical formula IWC = 0.104Z 0.483 And the empirical formula IWC=0.104Z proposed by Mace et al. in 2002 0.516 For comparison, Figure 3 and Figure 4 It can be seen from the experimental results that the estimated values ​​of ice water content using the empirical formula method have different degrees of deviation due to the influence of empirical values, and the error is large. However, the ice water content estimation using the genetic algorithm has a smaller error and has a better estimation performance.

[0074] From the perspective of the correlation between the true value and the estimated value, the inventors provide a scatter plot of the correlation between the true value of the ice water content at different altitudes in Oklahoma at 14:00 on March 17, 2016 and 16:00 on March 17, 2016 and the estimated value obtained by the method, as shown in Figure 2. Figure 5 and Figure 6 The horizontal axis in the figure is the estimated value of the ice water content after taking the logarithm obtained by the genetic algorithm, and the vertical axis is the true value of the ice water content after taking the logarithm. When the estimated value is completely equal to the true value, the data points should all be distributed on the diagonal line. The closer to the diagonal line, the more accurate the estimated value is. Figure 5 and Figure 6 It can be seen that the data points are all near the diagonal line, which proves the accuracy of this method in estimating the unknown parameters in the relationship model between the radar reflectivity factor and the ice water content of high-altitude ice crystals.

[0075] While providing a correlation scatter plot of the true value and estimated value of ice water content, the inventors have statistically analyzed the error size and probability of ice water content estimated by this method, and provided the error distribution histograms of data at different altitudes at 14:00 on March 17, 2016 and 16:00 on March 17, 2016, respectively. Figure 7 and Figure 8 As shown, Figure 7 and Figure 8 The error in is the absolute error between the estimated value of ice water content and the true value of ice water content, given by Figure 7 and Figure 8 It can be seen that the error between the true value and the estimated value of the data of different altitude layers at two time points is likely to be near 0, accounting for about 80%, which verifies the accuracy of this method.

[0076] In addition, the correlation coefficient and root mean square error (RMSE) of ice water content estimation using this method and the empirical formula method are statistically analyzed. The statistical results of the data at 14:00 on March 17, 2016 are shown in Tables 1 and 2, and the statistical results of the data at 16:00 on March 17, 2016 are shown in Tables 3 and 4. As can be seen from the table, compared with the empirical formula method, the correlation coefficient of this method is closer to 1, indicating that it has a strong correlation. At the same time, the RMSE of this method is less than 0.1, indicating that the deviation between the estimated value and the true value is small, while the RMSE deviation of the empirical formula method is large. The comparison shows that this method has higher accuracy.

[0077] After the estimation of ice water content is completed, the ice crystal detection results of different altitude data in Oklahoma, USA at 14:00 on March 17, 2016 and 16:00 on March 17, 2016 are given according to the estimated value of ice water content, such as Fig. 9 and Fig.10As shown. Fig. 9 and Fig.10 It can be seen that as time goes by, affected by the wind, the entire ice crystal distribution airspace gradually moves to the right.

[0078] Fig.11 and Fig.12 The distribution diagrams of ice crystal hazard levels at 14:00 on March 17, 2016 and 16:00 on March 17, 2016 when this method was used are given respectively. Fig.11 and Fig.12 By comparison, it can be seen that the ice crystal cloud in Oklahoma during this period gradually moved to the right, and the cloud that produced the ice crystals gradually became smaller. The changing characteristics are consistent with the characteristics of the generation, development, maturity and dissipation process of ice crystal meteorological conditions, indicating that the ice crystals in Oklahoma during this period are gradually dissipating.

[0079] Table 1. Correlation coefficient between estimated and true values ​​of ice water content (data at 14:00 on March 17, 2016)

[0080]

[0081] Table 2. RMSE between the estimated and true values ​​of ice water content (data at 14:00 on March 17, 2016)

[0082]

[0083] Table 3. Correlation coefficient between estimated and true values ​​of ice water content (data at 16:00 on March 17, 2016)

[0084]

[0085] Table 4. RMSE between the estimated and true values ​​of ice water content (data at 16:00 on March 17, 2016)

[0086]

Claims

1. A high-altitude ice crystal detection method based on genetic algorithm, characterized in that: The method comprises the following steps performed in sequence: 1) The ice crystal distribution airspace is determined by using the altitude and atmospheric temperature, and then the radar reflectivity factor of the ice crystal distribution airspace received by the airborne meteorological radar is preprocessed to obtain the preprocessed radar reflectivity factor and randomly divide it into a training set and a test set according to the proportion; 2) Establish a relationship model between the original radar reflectivity factor with unknown parameters and the ice water content of high-altitude ice crystals; 3) inputting the training set obtained in step 1) into the relationship model between the original radar reflectivity factor and the ice water content of high-altitude ice crystals obtained in step 2), and then using a genetic algorithm to finely optimize the unknown parameters in the relationship model between the original radar reflectivity factor and the ice water content of high-altitude ice crystals, and then inputting the test set obtained in step 1) for testing, and finally obtaining an accurate relationship model between the radar reflectivity factor and the ice water content of high-altitude ice crystals; 4) Input the radar reflectivity factor of the airspace where ice crystals are to be measured into the above-mentioned accurate radar reflectivity factor and the ice water content relationship model of high-altitude ice crystals to obtain the estimated value of the ice water content in the airspace, and finally determine the ice crystal hazard level according to the estimated value of the ice water content.

2. The high-altitude ice crystal detection method based on genetic algorithm according to claim 1 is characterized in that: In step 1), the ice crystal distribution airspace is set to an airspace above 22,500 feet above sea level and with an atmospheric temperature between -50°C and 20°C; The method for preprocessing the radar reflectivity factor is to select an appropriate range as the radar reflectivity factor threshold and remove the outliers in the received radar reflectivity factor; the ratio of the training set to the test set is 7:

3.

3. The high-altitude ice crystal detection method based on genetic algorithm according to claim 1, characterized in that: In step 2), the relationship model between the original radar reflectivity factor containing unknown parameters and the ice water content of high-altitude ice crystals is: Where IWC represents the ice water content of high-altitude ice crystals; Z represents the radar reflectivity factor; a represents the parameter characterizing the composition of ice crystal particles in the ice crystal mass power law distribution function m(D); N0 represents the parameter characterizing the size of ice crystal particles in the ice crystal particle number concentration distribution function N(D); μ represents the parameter determining the distribution shape in the ice crystal particle number concentration distribution function N(D); the last three parameters are unknown parameters.

4. The high-altitude ice crystal detection method based on genetic algorithm according to claim 1, characterized in that: In step 3), the method of using genetic algorithm to finely optimize the unknown parameters in the relationship model between the original radar reflectivity factor and the ice-water content of high-altitude ice crystals is: all unknown parameters in the relationship model between the original radar reflectivity factor and the ice-water content of high-altitude ice crystals are encoded separately, then an initial population is generated, and then selection and mutation operations are performed, and then the optimal solution is selected according to the fitness function, and the optimal parameters are obtained after decoding, and finally the optimal parameters are brought into the relationship model between the original radar reflectivity factor and the ice-water content of high-altitude ice crystals to obtain an accurate relationship model between the radar reflectivity factor and the ice-water content of high-altitude ice crystals.

5. The high-altitude ice crystal detection method based on genetic algorithm according to claim 1, characterized in that: In step 4), the method for determining the ice crystal hazard level according to the estimated value of the ice water content is as follows: according to the European Union's "Feasibility Study Report on Ice Crystal Detection by Weather Radar", the ice crystal hazard level is divided into three categories according to the estimated value of the ice water content: mild, moderate and severe. 3 When the ice crystal hazard level is mild, the estimated value of ice water content is between 1g / m 3 With 3g / m 3 When the ice crystal hazard level is between 20 and 30 °C, the ice crystal hazard level is moderate, and the estimated ice water content is greater than 3 g / m 3 The ice crystal hazard level is severe.

Citation Information

Patent Citations

  • Dual-polarization meteorological radar precipitation particle classification method based on discrete attribute BNT

    CN110852245A

  • Prediction of ice crystal presence in a volume of airspace

    EP3070501A1