Cloud physical parameter prediction model application method, system, device, medium and product

Abstract

The invention relates to the field of icing prediction, in particular to a cloud physical parameter prediction model application method, system and device, a medium and a product, and the method comprises the steps: obtaining a first meteorological sample set containing a plurality of meteorological element combinations, and a corresponding median volume diameter; selecting a first test sample set from the first meteorological sample set, and training through the first test sample set by adopting a first training method and a second training method respectively to obtain a first prediction model and a second prediction model respectively; selecting a second test sample set from the second meteorological sample set, wherein the second test sample set and the first test sample set are not completely overlapped; respectively calculating the prediction accuracy of the first prediction model and the second prediction model by adopting a second test sample set; and selecting a corresponding prediction model according to the prediction accuracy recommendation. The invention provides a method for accurately predicting LWC and MVD by accurately identifying the size of the liquid drop and utilizing the characteristics of the liquid drop.

Description

Technical Field

[0001] The present invention relates to the field of icing prediction technology, and in particular to a cloud physical parameter prediction model application method, system, equipment, medium and product. Background Art

[0002] When an aircraft encounters supercooled water while passing through clouds, icing is likely to occur, which will significantly reduce the aerodynamic performance of the aircraft and may lead to serious aviation safety accidents. Accurately predicting the cloud physical parameters of the aircraft passing through the clouds and improving the accuracy of aircraft icing forecasts is one of the effective ways to ensure aviation safety. In recent years, a large number of scholars have conducted research on aircraft icing forecasts. Bernstein et al. proposed a current icing potential (CIP) algorithm, which forecasted the hourly icing potential and supercooled large droplets (SLD) in the US airspace by integrating the latest observational data and the output of the NOAA Rapid Refresh (RAP) model. Subsequently, a numerical model-based icing potential forecast (FIP) algorithm was developed based on the CIP algorithm, which laid the foundation for the establishment of an icing forecast system for aviation meteorological centers. However, with the development of aircraft anti-icing technology, the impact of trace icing on aircraft is relatively small, and the technical difficulty of icing prediction has also increased significantly.

[0003] In recent years, researchers have used advanced data assimilation techniques and improved microphysics schemes to improve forecast accuracy. Through the cloud microphysics scheme proposed by Thompson et al., the FAA Aviation Weather Research Program has strengthened the icing forecasts for aircraft and ground as well as the quantitative precipitation forecasts, which can effectively predict the cloud droplet concentration. Davis et al. conducted a simulation study on a wind power icing event in Sweden, testing different combinations of three microphysics schemes and three indeterminate boundary layer schemes, and found that the combination of the Thompson microphysics scheme and the Edelhammer scheme was relatively effective, but the false alarm rate was still high.

[0004] In addition, patent application CN111738481A discloses a method for predicting aircraft icing meteorological parameters MVD based on a BP neural network. According to the relationship between the liquid water content LWC, the average effective water droplet diameter MVD and the surrounding air temperature T listed in Appendix C of the "China Transport Aircraft Airworthiness Standards [CCAR-25-R4]", the icing conditions under different flight conditions are calculated, and an icing thickness database showing the change of ice thickness over time is established; the icing thickness database is used to train the BP neural network model, and the real-time flight conditions (flight angle of attack, flight speed), temperature, and the mapping relationship between ice thickness and ice time are used as input to predict the icing meteorological parameter MVD. Patent application CN114880947A discloses a method for predicting aircraft icing meteorological parameters MVD and LWC, including obtaining the icing thickness and icing rate at the measuring point based on the icing shape calculated by icing, and establishing a database of effective icing thickness and icing rate; using the flight speed, ambient temperature, wing angle of attack, effective icing thickness and icing rate as input parameters, and training an Elman neural network with genetic algorithm optimized initial weights and thresholds to predict the output meteorological parameters of average effective water droplet diameter MVD and liquid water content LWC.

[0005] However, the above prediction methods still have defects in prediction reliability and practicality. Summary of the invention

[0006] The purpose of the present invention is to provide a cloud physical parameter prediction model application method, system, device, medium and product, which can partially solve or alleviate the above-mentioned deficiencies in the prior art and improve the prediction accuracy of LWC and MVD in typical application scenarios such as layered clouds.

[0007] In order to solve the technical problems mentioned above, the present invention specifically adopts the following technical solutions: a cloud physical parameter prediction model application method, comprising the steps of: S401, obtaining a first meteorological sample set containing a combination of multiple meteorological elements, and a corresponding median volume diameter; S402, selecting a first test sample set from the first meteorological sample set, and respectively training the first test sample set using a first training method and a second training method to obtain a first prediction model and a second prediction model respectively; wherein the step of obtaining the first prediction model by training using the first training method comprises: inputting the first test sample set into a neural convolutional network, at which time the output of the neural convolutional network is a probability value of the median volume diameter, and the corresponding probability value indicates the probability that the corresponding sample belongs to a large droplet or a small droplet; according to the probability value, the predicted The samples measured as small droplets are input into the neural convolution network again to update the neural convolution network, and the first prediction model is output after the training converges; at this time, the output of the neural convolution network is the second liquid water content and the median volume diameter; the step of training the second prediction model using the second training method includes: inputting the first test sample set into the neural convolution network, the output of the neural convolution network is the second liquid water content and the median volume diameter, and the second prediction model is output after the training converges; S403, selecting a second test sample set from the second meteorological sample set, the second test sample set does not completely overlap with the first test sample set; S404, using the second test sample set to respectively calculate the prediction accuracy of the first prediction model and the second prediction model; S405, selecting the corresponding prediction model according to the prediction accuracy recommendation.

[0008] In some embodiments, S403 includes the steps of: selecting a combination of meteorological elements belonging to a set vertical motion interval and a set relative humidity interval from the second meteorological sample set, and forming a second test sample set according to the combination of meteorological elements. In some embodiments, S403 includes the steps of: selecting a combination of meteorological elements belonging to a set altitude interval from the second meteorological sample set, and forming a second test sample set according to the combination of meteorological elements. In some embodiments, the meteorological sample group includes: temperature, air pressure, relative humidity, cloud water mass mixing ratio, rain water mass mixing ratio, vertical speed, ice water mass mixing ratio, ice water content, raindrop number concentration, ice number concentration, and first liquid water content.

[0009] The present invention also provides a cloud physical parameter prediction model application system, including: a first sample acquisition module, used to obtain a first meteorological sample set containing a combination of multiple meteorological elements, and a corresponding median volume diameter; a training module, used to select a first test sample set from the first meteorological sample set, and respectively use a first training method and a second training method to train the first test sample set to obtain a first prediction model and a second prediction model; wherein the step of using the first training method to train the first prediction model includes: inputting the first test sample set into a neural convolutional network, at which time the output of the neural convolutional network is a probability value of the median volume diameter, and the corresponding probability value indicates the probability that the corresponding sample belongs to a large droplet or a small droplet; according to the probability value, the sample predicted to be a small droplet is input into the neural network again; A convolutional network is used to update the neural convolutional network, and the first prediction model is output after the training converges; at this time, the output of the neural convolutional network is the second liquid water content and the median volume diameter; the step of training the second prediction model using the second training method includes: inputting the first test sample set into the neural convolutional network, the output of the neural convolutional network is the liquid water content and the median volume diameter, and the second prediction model is output after the training converges; a second sample acquisition module is used to select a second test sample set from the second meteorological sample set, and the second test sample set does not completely overlap with the first test sample set; a prediction accuracy calculation module is used to use the second test sample set to respectively calculate the prediction accuracy of the first prediction model and the second prediction model; a recommendation module is used to recommend the selection of the corresponding prediction model according to the prediction accuracy.

[0010] In some embodiments, the second sample acquisition module is used to select a combination of meteorological elements belonging to a set vertical movement interval and a set relative humidity interval from the second meteorological sample set, and form a second test sample set according to the combination of meteorological elements. In some embodiments, the second sample acquisition module is used to select a combination of meteorological elements belonging to a set altitude interval from the second meteorological sample set, and form a second test sample set according to the combination of meteorological elements.

[0011] In some embodiments, the meteorological sample group includes: temperature, air pressure, relative humidity, cloud water mass mixing ratio, rain water mass mixing ratio, vertical velocity, ice water mass mixing ratio, ice water content, raindrop number concentration, ice number concentration and first liquid water content.

[0012] The present invention also provides an electronic device, comprising: a memory for storing a computer program; a processor for implementing the steps of the cloud physical parameter prediction model application method of any embodiment when executing the computer program. The present invention also provides a computer-readable storage medium, in which a computer program is stored, and when the computer program is executed by the processor, the steps of the cloud physical parameter prediction model application method of any embodiment are implemented. The present invention also provides a computer program product, including a computer program / instruction, which, when executed by the processor, implements the steps of the cloud physical parameter prediction model application method described in any embodiment.

[0013] Beneficial technical effects: It should be noted that traditional LWC or MVD prediction algorithms are usually extremely dependent on aircraft flight data, such as flight speed, wing angle of attack, icing time, icing thickness and other measured requirements (or non-meteorological data), and the applicant has found that this will make the application of icing prediction algorithms very limited.

[0014] On the contrary, the present invention proposes a model training and application method for predicting LWC and MVD based on meteorological parameters. This scheme of directly predicting based on pure meteorological parameters can expand the application scenarios of the prediction algorithm. For example, before the aircraft performs a flight mission, such as in the design of the aircraft's related deicing system or the flight mission planning stage, the prediction of the degree of icing is also very important. The prediction algorithm proposed in the present invention can use the meteorological parameters obtained by meteorological forecasts to predict the degree of icing in the future long-term process (such as the next day, the next month, etc.), reducing dependence on other non-meteorological parameters.

[0015] Moreover, the first type of prediction method (or the first prediction model) proposed in the present invention has significant prediction advantages in typical flight areas such as layered clouds. Layered clouds refer to uniform (referring to uniform thickness, grayscale and light transmittance) clouds covering the entire sky or part of the sky. For example, the macroscopic structure of large-scale precipitation layered clouds associated with low-value weather systems is layered, sometimes in two layers and sometimes in three layers. There is a cloudless area between the two layers of clouds. The high-level clouds produce ice crystal particles and fall to the bottom-level clouds, which are catalytic clouds; the ice crystal particles entering the low-level clouds continue to grow, and the moisture and environment required for the growth of ice crystals in the bottom-level clouds are supply clouds.

[0016] During the flight, flying through layered clouds is one of the most typical flight scenes. However, due to the complexity of the cloud structure and meteorological characteristics inside the layered clouds, the difficulty of predicting the relevant parameters of the layered cloud area has also increased sharply. In this regard, the secondary training method based on the probability interval classification of large and small droplets proposed in the present invention can effectively improve the prediction model's recognition sensitivity for the droplet size inside the cloud layer, thereby significantly improving the model's prediction accuracy for LWC and MVD.

[0017] Furthermore, experiments have shown that the prediction model proposed in the present invention has shown extremely high reliability in predicting the parameters of small droplets. This also enables the present invention to significantly reduce the false negative rate when the freezing potential is relatively low (such as when there are more small droplets). In other words, the prediction method proposed in the present invention can alleviate the problem of low prediction accuracy for small droplets in the prior art.

[0018] In addition, the secondary training method based on large and small droplet probability interval classification adopted by the present invention can also reduce the dependence on the number of training samples, and it also shows extremely high reliability in the case of small sample training. Furthermore, since aircraft icing forecast is crucial to flight safety, the present invention also provides two training methods for users to choose from.

[0019] It is worth noting that the icing mechanism of aircraft is very complex and may vary greatly under different flight scenarios, climate conditions and aircraft types. In addition, it will cost a lot to create targeted modeling for ice accumulation prediction in different scenarios.

[0020] In this regard, the applicant chose to approach the issue from the perspective of droplet size prediction accuracy and provided two model training methods. One is to improve the droplet classification accuracy (and improve the effective learning depth for small droplets) in special scenarios (for example, the amount of sample data is relatively limited, or when flying to typical complex climate areas such as layered clouds) through a secondary classification training method (i.e., the first training method); the other is to directly use neural networks for regression training (corresponding to the second training method) to quickly model another type of flight scenario (such as a relatively small amount of sample data, or a relatively simple climate).

[0021] Therefore, when performing ice accumulation prediction for a new flight mission, the user can use a small number of samples to complete preliminary test training through the first and second training methods in advance, and then select the recommended training method based on the preliminary prediction results of the two prediction models, so as to select the recommended training method to complete the final model training. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. In all the drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, each element or part is not necessarily drawn according to the actual scale. Obviously, the drawings described below are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can also be obtained based on these drawings without paying creative labor.

[0023] Figure 1is a schematic diagram of a method flow in an exemplary embodiment of the present invention; Figure 2 A schematic diagram of a WRF mode domain configuration in an exemplary embodiment; Figure 3 The figure is a schematic diagram of the NAC0012 airfoil and a curve diagram of the change of the ice type at the leading edge of the wing with the exposure time; wherein (a) is a schematic diagram of the NAC0012 airfoil, and (b) is a curve diagram of the change of the ice type at the leading edge of the wing with the exposure time under a given state; Figure 4 ROC curve diagram for large droplet classification prediction based on CNN-Attention0 model; Figure 5 TSS curve diagram for large droplet classification prediction based on CNN-Attention0 model; Figure 6 The following are the prediction results of microphysical meteorological parameters by the CNN-Attention0 model; (a) shows the relationship between the model prediction accuracy and vertical motion and relative humidity, (b) shows the relationship between the model prediction accuracy and pressure and liquid water content, (c) shows the relationship between the model prediction accuracy and ice water content and temperature, and (d) shows the relationship between the model prediction accuracy and raindrop number concentration and ice number concentration; Figure 7 This is a scatter plot of the model prediction and actual observation density of LWC based on CNN-Attention1; Figure 8 This is a scatter plot of the model prediction and actual observation density of LWC based on CNN-Attention0; Fig. 9 This is a scatter plot of the model prediction and actual observation density of MVD based on CNN-Attention1; Fig.10 This is a scatter plot of the model prediction and actual observation density of MVD based on CNN-Attention0. DETAILED DESCRIPTION

[0024] In order to make the purpose, technical solution and advantages of the embodiment of the present invention clearer, the technical solution in the embodiment of the present invention will be clearly and completely described in conjunction with the drawings in the embodiment of the present invention. Obviously, the described embodiment is a part of the embodiment of the present invention, not all of the embodiments. Based on the embodiment of the present invention, all other embodiments obtained by ordinary technicians in the field without creative work are within the scope of protection of the present invention. In this article, the use of suffixes such as "module", "component" or "unit" used to represent elements is only to facilitate the description of the present invention, and has no specific meaning in itself. Therefore, "module", "component" or "unit" can be used in a mixed manner. In addition, the terms "first" and "second" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance. In this article, unless otherwise clearly specified and limited, the term "connection" and the like should be understood in a broad sense, for example, "connection" can be directly connected, or indirectly connected through an intermediate medium, or it can be the internal connection of two elements. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood in specific circumstances. "And / or" herein includes any and all combinations of one or more listed related items. In this paper, "multiple" means two or more than two, that is, it includes two, three, four, five, etc. The term "approximately" used in this specification is typically expressed as + / -5% of the value, more typically + / -4% of the value, more typically + / -3% of the value, more typically + / -2% of the value, even more typically + / -1% of the value, and even more typically + / -0.5% of the value. In this specification, some embodiments may be disclosed in a format that is in a certain range. It should be understood that this description of "being in a certain range" is only for convenience and simplicity, and should not be interpreted as a rigid limitation to the disclosed range. Therefore, the description of the range should be considered to have specifically disclosed all possible sub-ranges and independent numerical values ​​within this range. For example, the description of the range 1-6 should be considered to have specifically disclosed sub-ranges such as from 1 to 3, from 1 to 4, from 1 to 5, from 2 to 4, from 2 to 6, from 3 to 6, etc., and individual numbers within this range, such as 1,2,3,4,5 and 6. The above rules apply regardless of the breadth of the scope.

[0025] Example 1: See Figure 1 As shown, the present invention provides a method for predicting cloud physical parameters, comprising the steps of: training a model (equivalent to providing a first training method), which comprises the steps of: S100, obtaining a first meteorological sample set, the first meteorological sample set comprising: meteorological sample groups at multiple moments, the meteorological sample groups comprising: temperature ( T )、Air pressure( P )、Relative humidity( RH ), cloud water mass mixing ratio ( Q cloud )、Rainwater mass mixing ratio( Q rain )、Vertical speed( W )、Ice-water mass mixing ratio( Q ice )、Ice water content( IWC )、Raindrop number concentration( N rain )、ice number concentration( N ice )、First liquid water content( LWC ); S101, inputting the first meteorological sample set into a convolutional neural network for iterative calculation, wherein the input end of the convolutional neural network is: temperature, air pressure, relative humidity, cloud water mass mixing ratio, rain water mass mixing ratio, vertical velocity, ice water mass mixing ratio, ice water content, raindrop number concentration, ice number concentration and first liquid water content; the output end of the convolutional neural network includes: probability value of median volume diameter (MVD) belonging to large droplets; Preferably, the activation function of the convolutional neural network adopts a normalized exponential function, which is used to calculate the probability value corresponding to the meteorological sample group, and the loss function of the convolutional neural network adopts a binary cross-point function, which is used to calculate the convergence state of the current probability value; That is to say, the S101 process is equivalent to the classification step. Specifically, in order to improve the accuracy of large droplet classification, the present invention uses a normalized exponential function to convert the unnormalized MVD value into a probability value to improve the iterative convergence speed of the model and the accuracy of model training; and uses a binary crossover function to calculate the convergence during the iteration process. Subsequently, the classification of different sample data is completed with the help of probability values; in other words, the rapid classification of samples is completed through probability values.

[0026] Preferably, in the classification training process of S101 of this embodiment, an initial model for predicting the actual liquid water content (equivalent to the second liquid water content) and the median volume diameter can also be obtained according to the temperature, air pressure, relative humidity, cloud water mass mixing ratio, rain water mass mixing ratio, vertical velocity, ice water mass mixing ratio, ice water content, raindrop number concentration, ice number concentration, and predicted liquid water content (equivalent to the first liquid water content) output by the WRF model; specifically, in this process, the output end of the convolutional neural network includes: the second liquid water content and the median volume diameter.

[0027] Specifically, the probability value in this embodiment can be used to indicate the probability that the droplet is a large droplet.

[0028] In some embodiments, the input layer in the convolutional neural network inputs the meteorological sample group and the MVD value (such as the average MVD (μm)); the activation function can calculate the probability values ​​corresponding to different meteorological sample groups through the association between the meteorological sample group and the MVD value (such as the average MVD (μm)).

[0029] In some embodiments, the loss function is used to calculate the difference between the probability value of the current iteration process and the probability value of the previous iteration process. When the difference is small, it is considered that the two have converged and the iteration is terminated, otherwise the iteration continues.

[0030] S102, the meteorological samples corresponding to the probability values ​​belonging to the set probability interval and the temperature being greater than the set temperature threshold are combined to form a second meteorological sample set; preferably, when the probability value belongs to the set probability interval and the temperature is greater than the set temperature threshold, it is considered that the corresponding sample droplets are small droplets with a high probability. That is to say, in this embodiment, the second meteorological sample set is a set of samples that are small droplets with a high probability.

[0031] S103, inputting the second meteorological sample set into the convolutional neural network again for updating and iterative calculation, and finally obtaining a corresponding first prediction model (also referred to as CNN-Attention0 in this article), the output of the first prediction model is: second liquid water content and median volume diameter; In this embodiment, the first liquid water content may be a sample value output according to the WRF mode; and the second liquid water content refers to a predicted value output according to a prediction mode (such as the first prediction model). Preferably, the activation function adopts a linear rectification function, which is used to calculate the second liquid water content and the median volume diameter, and the loss function adopts a mean square error function, which is used to calculate the convergence state of the second liquid water content and the median volume diameter. In other words, this process can update the initial model obtained in S101.

[0032] It is worth noting that a model training method based on sample interval classification for secondary training is proposed in this embodiment. Specifically, the present invention roughly screens out sample data belonging to small droplets by the probability value of the droplets, and inputs the screened small droplet sample data into the convolutional neural network again for intensive training, so as to enhance the recognition ability of the first prediction model for droplets of different particle sizes and improve the prediction accuracy of LWC and MVD.

[0033] For example, in this embodiment, the weather research and forecasting model (WRF) is used to obtain the T , RH , P , LWC , Qcloud , Q rain , IWC , W , Q ice , N ice and N rain , etc., and are transmitted to the convolutional layer through the input layer. The convolutional layer includes 4 one-dimensional convolutions. The first convolution reads the input sequence and projects the result onto the feature map. The second convolution performs the same operation on the feature map created by the first layer, and performs maximum pooling after each convolution to amplify its significant features. In the first training stage, for the large droplet classification algorithm, the activation function is the normalized exponential function, namely the Softmax function, which is used to calculate the probability value of the corresponding droplet belonging to a large droplet, and the loss function is the binary intersection function. Then enter the maximum pooling layer, take the maximum value in the sample as the sample value after sampling, use the maximum pooling layer to simplify the feature map, and then flatten the feature map into a long vector for decoding. Finally, enter the fully connected layer, and use the LWC and MVD values ​​matched in time and space as the model output. Droplets with MVD less than 100μm are identified as small droplets, and the sample data identified as small droplets are input into the convolutional neural network again. At this time, the activation function uses the RELU function and the loss function uses the mean square error function.

[0034] The loss function is calculated after each iteration cycle of the training set and the test set. The model is terminated after 500 iteration training cycles, and the corresponding first prediction model is obtained.

[0035] In some embodiments, before S101, the method further includes the steps of: obtaining a grid model of the area to be predicted, the grid model having a plurality of sample grid points; performing vertical linear interpolation on the first meteorological sample set to match corresponding meteorological sample values ​​for the plurality of sample grid points, wherein the plurality of meteorological sample values ​​form a meteorological sample group corresponding to one sample grid point. In some embodiments, the first meteorological sample set is derived from aircraft detection data and a WRF model.

[0036] In some embodiments, before S102, the steps include: setting a probability interval; the setting of the probability interval includes the steps of: selecting multiple discrimination thresholds between [a, b], the discrimination thresholds are used to judge large droplets and small droplets, a and b are set probability end values ​​(set by the user); setting a first verification index, the first verification index includes: a first ratio R1 of the number of samples correctly classified as large droplets to the actual number of large droplet samples, and a second ratio R2 of the number of samples classified as small droplets to the actual number of small droplet samples; wherein, when the probability value is greater than or equal to the discrimination threshold, it is considered to be a large droplet, and when the probability value is less than the discrimination threshold, it is considered to be a small droplet; calculating multiple receiver operating characteristic curves (Receiver Operating Characteristic curve, ROC) corresponding to the multiple discrimination thresholds according to the first ratio R1 and the second ratio R2; calculating the area under the receiver operating characteristic curve (Area UnderCurve, AUC); select at least one discrimination threshold corresponding to the largest area size as the discrimination threshold c in the model training process, and generate the corresponding probability interval based on the discrimination threshold, such as the probability interval can be [a, c]. Among them, the receiver operating characteristic curve refers to a curve graph with the false alarm probability (the second ratio R2) as the horizontal axis and the hit probability (the first ratio R1) as the vertical axis, and the area under the curve can characterize the model's discrimination ability. For example, when the AUC value is 0.5, it means there is no discrimination ability; the closer the AUC value is to 1, the better the prediction performance. The following experimental verification shows that the present invention can achieve an AUC value of 0.979 for the entire sample set by combining the above model training with the probability setting scheme, indicating that the model training method has high reliability.

[0037] In some embodiments, the first verification indicator may also include: a third ratio R3 of the number of samples correctly classified as small droplets to the actual number of small droplet samples; a fourth ratio R4 of the number of samples classified as large droplets as small droplets to the actual number of large droplet samples; accuracy, that is, the ratio of the number of correctly classified droplet samples (including large droplets and small droplets) to the total number of samples.

[0038] Furthermore, in some embodiments, in order to improve the accuracy of model training, the steps may also be included: calculating the true skill statistic (TSS) under each discrimination threshold according to the first ratio R1 and the second ratio R2; when the true skill statistic is greater than or equal to the set minimum statistic, it is considered that the corresponding discrimination threshold meets the requirements of model training.

[0039] For example, in this embodiment, the minimum statistical value is set at about 0.85, see Figure 4and Figure 5 As shown, when TSS reaches 0.87, the first ratio is 0.98 and the second ratio is 0.11; when TSS is 0.94, the training set shows better performance, the first ratio is 0.99 and the second ratio is 0.05. The test results can also further verify the reliability of the training method adopted by the present invention.

[0040] In some embodiments, the method further includes: model application determination, which includes the steps of: S200, obtaining the vertical speed and relative humidity of the aircraft in the flight area; S201, using a first determination mechanism to determine whether the first prediction model is applicable to the flight area; the first determination mechanism requires that the vertical speed is lower than a preset vertical motion threshold, and the relative humidity is higher than a preset relative humidity threshold; for example, in some embodiments, the relative humidity threshold may be approximately 70%; if so, it is recommended to use the first prediction model.

[0041] It is worth noting that the applicant has found that the first prediction model obtained based on secondary training of droplet classification is particularly suitable for flight areas such as layered clouds. Therefore, in order to improve the reliability of model prediction in actual application, the present invention also proposes a corresponding judgment mechanism.

[0042] In some embodiments, before S102, the method further includes: S203, obtaining the altitude below the flight area; S204, using a second determination mechanism to determine whether the first prediction model is applicable to the flight area, and if so, recommending the use of the first prediction model; the second determination mechanism requires the altitude to be lower than a set poster height threshold. For example, in some embodiments, when the altitude is lower than about 800 hPa, it is recommended to use the first prediction model.

[0043] In some embodiments, the present invention further provides another second prediction model directly obtained by training based on a regression algorithm; the corresponding model training steps (i.e., the second training method) include: Acquire a first meteorological sample set, the first meteorological sample set comprising: meteorological sample groups at multiple moments, the meteorological sample groups comprising: temperature ( T )、Air pressure( P )、Relative humidity( RH ), cloud water mass mixing ratio ( Q cloud )、Rainwater mass mixing ratio( Q rain )、Vertical speed( W )、Ice-water mass mixing ratio( Q ice )、Ice water content( IWC )、Raindrop number concentration( Nrain )、ice number concentration( N ice )、Liquid water content( LWC ); The first meteorological sample set is input into a convolutional neural network for iterative calculation. At this time, the input end of the convolutional neural network is: temperature, air pressure, relative humidity, cloud water mass mixing ratio, rain water mass mixing ratio, vertical speed, ice water mass mixing ratio, ice water content, raindrop number concentration, ice number concentration and first liquid water content; the output end of the convolutional neural network includes: second liquid water content and median volume diameter; preferably, the activation function adopts a linear rectification function, and the linear rectification function is used to calculate the second liquid water content and the median volume diameter, and the loss function adopts a mean square error function, and the mean square error function is used to calculate the convergence state of the second liquid water content and the median volume diameter; For example, in some embodiments, the output of WRF T , RH , P , Q cloud , Q rain , IWC , W , Q ice , N ice , N rain 、 LWC Set as the input of the network, and take the LWC and MVD values ​​of the spatiotemporal matching as the output of the network. It can be understood that the samples used in the model training process in this article can be the measurement results collected by the meteorological monitoring system (such as meteorological satellite).

[0044] When the iterative process converges, the corresponding second prediction model (also referred to as CNN-Attention1 model in this article) is obtained. Different from the above solution, the model training in this embodiment directly performs regression training on the entire sample, without introducing the step of distinguishing large and small droplet samples.

[0045] In some embodiments, the above two model training methods can be used interchangeably.

[0046] For example, in some embodiments, different training schemes may be selected according to different training scenarios, such as the number of samples and the distribution of MVD values ​​of the samples.

[0047] For example, in this embodiment, before model training, the step of obtaining a meteorological sample set is also included, which includes:T , RH , P , LWC , Q cloud , Q rain , IWC , W , Q ice , N ice and N rain , and MVD; calculating the difference between at least one maximum MVD value and at least one minimum MVD value in the meteorological sample set; when the difference is less than the set difference interval, recommending the first training method for sample training; otherwise, it may be recommended to use the second training method for sample training; For another example, in some embodiments, before performing model training, the steps of: obtaining a meteorological sample set; calculating the number of samples; when the number of samples is less than the set sample size, recommending the use of a first training method for sample training; otherwise, it may be recommended to use a second training method for sample training.

[0048] It is worth noting that the first training method provided by the present invention, which adopts a mode of secondary training based on droplet classification, has significant prediction advantages in scenarios where the number of samples is limited and the sample data distribution is relatively concentrated. Or in some embodiments, a small number of samples can be used to simultaneously use two training methods to obtain a preliminary prediction model; the recommended training method is selected by comparing the prediction accuracy of the two prediction models. Or in some embodiments, two prediction models can be used simultaneously for prediction to cross-validate the prediction results.

[0049] Embodiment 2: The present invention also provides a verification method for model training, comprising the steps of: S300, obtaining a first meteorological sample set and a corresponding median volume diameter, wherein the first meteorological sample set includes: a meteorological sample group at multiple moments, wherein the meteorological sample group includes: temperature, air pressure, relative humidity, cloud water mass mixing ratio, rain water mass mixing ratio, vertical speed, ice water mass mixing ratio, ice water content, raindrop number concentration, ice number concentration, and liquid water content; S301, inputting the first meteorological sample set into a convolutional neural network for iterative calculation, wherein the input end of the convolutional neural network is: temperature, air pressure, relative humidity, cloud water mass mixing ratio, rain water mass mixing ratio, vertical speed, ice water mass mixing ratio, ice water content, raindrop number concentration, ice number concentration and first liquid water content; the activation function of the convolutional neural network is used to calculate the probability value that the median volume diameter of the sample group belongs to large droplets, and the probability value indicates the probability that the corresponding sample belongs to large droplets or small droplets; correspondingly, the output end of the convolutional neural network includes: probability value; S302, combining the meteorological samples whose probability values ​​of the median volume diameter belong to a set probability interval and whose temperatures are greater than a set temperature threshold to form a second meteorological sample set; S303, inputting the second meteorological sample set into the convolutional neural network again for updating and iterative calculation, and finally obtaining a corresponding first prediction model, wherein the output of the first prediction model includes: a second liquid water content and a median volume diameter; S304, calculating a first verification index, the first verification index comprising: a first ratio R1 of the number of samples correctly classified as large droplets to the actual number of large droplet samples, and / or a second ratio R2 of the number of samples classified as small droplets to the actual number of small droplet samples; S305, determining whether the first verification indicator meets the set first verification condition, if not, adjusting the probability interval, and executing S302 again according to the new probability interval.

[0050] For application scenarios with relatively small sample sizes, the present invention provides a secondary training method based on the probability interval classification of large and small droplets, and uses the classification accuracy of large and small droplets (corresponding to R1 and R2) as a key indicator to guide the adjustment direction of secondary training, and then improves the prediction accuracy of small droplets based on this key indicator. Finally, the accuracy of icing prediction is effectively improved in typical application scenarios such as small training sample sizes or layered clouds.

[0051] For example, in some embodiments, when the first ratio and the second ratio are both greater than corresponding set values, it is considered that the first verification indicator meets the set first verification condition.

[0052] For another example, in some embodiments, the probability interval is set by a discrimination threshold. When the probability value is greater than or equal to the discrimination threshold, it is considered to be a large droplet. When the probability value is less than the discrimination threshold, correspondingly, S305 includes the steps of: calculating multiple receiver operating characteristic curves corresponding to the discrimination threshold according to the first ratio R1 and the second ratio R2; calculating the area size of the receiver operating characteristic curve; when the area size is greater than the set area threshold, it is considered that the first verification indicator determines whether it meets the set verification condition. In some embodiments, it also includes the steps of: when the area size is less than or equal to the area threshold, adjusting the discrimination threshold, and executing S302 and S303 again according to the corresponding new probability interval. In some embodiments, the convolutional neural network includes at least four convolutional layers.

[0053] Embodiment 3: The present invention also provides a cloud physical parameter prediction model application method, comprising the steps of: S401, obtaining a first meteorological sample set including a combination of multiple meteorological elements and a corresponding median volume diameter; S402, selecting a first test sample set from the first meteorological sample set, and respectively training the first test sample set using a first training method and a second training method to obtain a first prediction model and a second prediction model respectively; The step of using the first training method to train a first prediction model includes: Inputting the first test sample set into a convolutional neural network, where the output of the convolutional neural network includes: a probability value of a median volume diameter, wherein the probability value indicates a probability that the corresponding sample belongs to a large droplet or a small droplet; According to the probability value, the sample predicted to be a small droplet is input into the convolutional neural network again to update the convolutional neural network, and the first prediction model is output after the training converges; at this time, the output of the convolutional neural network includes: liquid water content and median volume diameter; The step of using the second training method to train and obtain the second prediction model includes: inputting the first test sample set into a convolutional neural network, the output of the convolutional neural network is the second liquid water content and the median volume diameter, and outputting the second prediction model after the training converges; S403, selecting a second test sample set from the second meteorological sample set, wherein the second test sample set does not completely overlap with the first test sample set; in some embodiments, the second meteorological sample set may be a part of the first meteorological sample set. Alternatively, the second meteorological sample set may also be a meteorological forecast result obtained for a period of time in the future for the current icing prediction task (or flight task).

[0054] S404, using the second test sample set to respectively calculate the prediction accuracy of the first prediction model and the second prediction model; S405: Recommend and select a corresponding prediction model based on the prediction accuracy.

[0055] It is worth noting that the icing mechanism of aircraft is very complex and may vary greatly under different flight scenarios, climate conditions and aircraft types. In addition, it will cost a lot to create targeted modeling for icing prediction in different scenarios.

[0056] In this regard, the applicant chose to approach the issue from the perspective of droplet size prediction accuracy and provided two model training methods. One is to improve the droplet classification accuracy (and improve the effective learning depth for small droplets) in special scenarios (for example, when the amount of sample data is relatively limited, or when flying to cumulus icing weather areas) through a secondary classification training method (i.e., the first training method); the other is to directly use neural networks for regression training (corresponding to the second training method) to quickly model another type of flight scenario (such as a relatively large amount of sample data, or a layered cloud meteorological environment).

[0057] Therefore, when performing icing prediction for a new flight mission, the user can use a small number of samples to complete preliminary test training through the first and second training methods in advance, and then select the recommended training method based on the preliminary prediction results of the two prediction models, and use the recommended training method to complete the final model training.

[0058] For another example, in some embodiments, two prediction models may be used simultaneously to perform synchronous prediction for user reference.

[0059] In some embodiments, S403 includes the steps of: selecting the meteorological element combination belonging to the set vertical movement interval and the set relative humidity interval from the second meteorological sample set, and forming the second test sample set according to the meteorological element combination.

[0060] Furthermore, the applicant has noticed that the first prediction model may show extremely high sensitivity in areas with weak vertical motion and high relative humidity. Therefore, when validating the model, it is preferred to observe its prediction accuracy in a specific vertical motion and relative humidity range.

[0061] In some embodiments, S403 includes the steps of: selecting the meteorological element combination belonging to the set altitude interval from the second meteorological sample set, and forming the second test sample set according to the meteorological element combination.

[0062] Furthermore, the applicant has noticed that the first prediction model may show extremely high sensitivity in low altitude areas. Therefore, when validating the model, it is preferred to observe its prediction accuracy in a specific altitude range.

[0063] In some embodiments, the meteorological sample group includes: temperature, air pressure, relative humidity, cloud water mass mixing ratio, rain water mass mixing ratio, vertical velocity, ice water mass mixing ratio, ice water content, raindrop number concentration, ice number concentration and liquid water content.

[0064] In order to illustrate the technical solution and technical effects adopted by the present invention, the following will explain through a specific model training and application example: 1. Source of data sampled for model training and testing: This example uses flight data published by the ICICLE project, which focuses on improving understanding of aircraft icing by studying atmospheric microphysical processes. During the implementation of the project, the National Center for Atmospheric Research (NCAR) conducted multiple flight tests in the Great Lakes region of North America (between 37.588-45.666°N) from January 28 to March 8, 2019, using detection instruments on the aircraft to collect cloud physical parameter data and analyze the thermodynamics, cloud dynamics, and specific cloud and aerosol characteristics of cold clouds and precipitation.

[0065] This example also uses data from airborne instruments on the Canadian National Research Council (NRC) Convair 580 aircraft participating in the project. The flight altitude, longitude, and latitude data are provided by the KVH 1750 IMU sensor installed in the cabin; the microphysical data of liquid and ice particles are obtained by the forward scattering spectrometer (FSSP), the two-dimensional stereo (2DS) probe, and the high-throughput precipitation spectrometer (HVPS). The particle sizes measured by the three instruments are 3-35μm, 40-670μm, and 750-38400μm, respectively. The droplets are divided into small droplets (<100μm) and large droplets (>100μm) with a threshold of 100μm for further analysis.

[0066] The aircraft carried out a total of 30 flight detections. Since the first flight was a test flight, the second and 30th flights did not conduct detection, so the flight detection data from the third to the 29th flights were mainly used. It should be noted that the 10th flight was a transfer flight, and the probe instrument was activated for a very short time, and no relevant cloud physical parameters were detected. The average root mean square error of the detection data is 30s, which is roughly equivalent to a path length of 3km. The LWC data points, large water droplet data points, average LWC, average MVD, and large water droplet percentage within 30s counted by each aircraft are listed in Tables 1.1-1.2.

[0067] Table 1.1 Aircraft statistics table Table 1.2 Aircraft statistics table It can be understood that, in some embodiments, the sample values ​​of LWC and MVD matched in time and space can be obtained through aircraft detection.

[0068] 2. Data processing method - meteorological parameter forecasting method based on WRF The Weather Research and Forecast (WRF) model is a fully compressible, non-hydrostatic numerical model. This example uses the initial and boundary conditions of the fifth generation atmospheric reanalysis dataset (ERA5) of the European Centre for Medium-Range Weather Forecasts (ECMWF), with a horizontal grid spacing of 0.25° and a temporal resolution of 6h. A two-way nesting strategy is used to create two nested attribute domains, with a horizontal resolution of 27 km and 9 km in each region, a vertical resolution of 44 altitude layers, and a temporal resolution of 1h, such as Figure 2 The black line is the flight path of the aircraft. The flight time simulated by the WRF model is at least 6 hours each time to ensure reliable simulation accuracy.

[0069] The WRF parameterization schemes selected in this embodiment are as follows: the microphysics scheme uses the Thompson scheme, which can improve the explicit forecast of aircraft icing and plays an important role in icing forecast; the shortwave radiation scheme adopts the Dudhia scheme, the longwave radiation scheme adopts the rapid radiation transfer model, and the Eta surface layer adopts the Noah land surface model. These configurations are determined based on the research results of winter conditions; the cumulus scheme uses the Kain-Fritsch cumulus scheme.

[0070] In this embodiment, the following meteorological variables are selected to describe the aircraft icing environment, including altitude ( z ),temperature( T )、Air pressure( P )、Relative humidity( RH ), cloud water mass mixing ratio ( Q cloud )、Rainwater mass mixing ratio( Q rain )、Vertical speed( W )、Ice-water mass mixing ratio( Q ice )、Ice water content( IWC )、Raindrop number concentration( N rain )、ice number concentration( N ice )、Liquid water content( LWC), as shown in Table 2. Among them, temperature, altitude, air pressure, relative humidity, cloud water mass mixing ratio, rain water mass mixing ratio, vertical velocity, and ice water mass mixing ratio are provided by WRF output parameters, and ice water content, raindrop number concentration, ice number concentration, and liquid water content are obtained by multiplying the output parameters by air density.

[0071] Table 2 Variables and definitions list 3. Training Tools This embodiment preferably adopts a feedforward neural network model based on convolution (also known as a convolutional neural network model, or a CNN network model). The CNN model consists of a convolution layer, a pooling layer, and a fully connected layer. Among them, the convolution layer is composed of multiple convolution units, and the parameters of each convolution unit are optimized by a back propagation algorithm. In this embodiment, the goal of the convolution operation is to extract various features of the input. The input data passes through the convolution layer to obtain a feature map, and the feature map convolves the input data through the neural network to represent the features in the neural space. The pooling layer is used to sample the feature map, mainly by reducing the number of network parameters to reduce the computational complexity, so as to reduce the network size and obtain the invariant features of the input data. The fully connected layer mainly combines all local features, multiplies the local features by their corresponding weights, and then sums them through a convolution operation to form a global feature for calculating the final score of each category.

[0072] The CNN model used in this embodiment can reduce the complexity of meteorological sample data and extract data features; by introducing the attention mechanism in the CNN model, the data dimension of the meteorological sample data can be reduced and the training efficiency can be improved.

[0073] Among them, the calculation formula for data feature extraction using the CNN-attention mechanism model is as follows: (1); (2); in x is the input data, y is the output data, σ is the activation function, W k and b k are the weight coefficient and the bias function respectively, k is the number of convolution kernels (indicating discrete convolution operations), α and β is the size parameter of the convolution kernel, W m,n represents the weight coefficient of the convolution kernel of the weight matrix, m and nRespectively represent the row index and column index of the eigenvalues ​​in the convolution kernel, i, j Represents the row and column indices of the input data.

[0074] IV. Sampling Strategy for Model-Observation Comparison In terms of time, in order to ensure that the model output results are completely matched with the flight detection data, this embodiment matches the aircraft detection results 15 minutes before and after the model output time with the model output. In each 30-minute sampling interval, the output value of a model grid point is selected as the matching value of the average observation. In terms of space, this embodiment uses the vertical linear interpolation method to model the aircraft sampling height output to achieve optimal matching of the flight altitude. The aircraft detection data and WRF prediction data after spatiotemporal matching are divided into two parts: a training set (70%) and a test set (30%). Among them, the training set is used to train the CNN-attention model, and the test set is used to evaluate the accuracy of the model.

[0075] Specifically, the model evaluation statistical method used in this embodiment is: This embodiment uses the first ratio (also known as: true positive rate (TPR)), the second ratio (also known as: false positive rate (FPR)), the third ratio (also known as: true negative rate (TNR)), the fourth ratio (also known as: false negative rate (FNR)) and accuracy to evaluate the recognition ability of the CNN-attention model for water droplets. Large droplets are positive and small droplets are negative. TPR is the ratio of the number of samples correctly classified as large droplets to the actual number of large droplet samples, FPR is the ratio of the number of samples classified as small droplets to the actual number of small droplet samples, TNR is the ratio of the number of samples correctly classified as small droplets to the actual number of small droplet samples, FNR is the ratio of the number of samples classified as large droplets to the actual number of large droplet samples, and accuracy is the ratio of the number of correctly classified droplet samples (including large droplets and small droplets) to the total number of samples.

[0076] The root mean square error (RMSE) and Pearson correlation coefficient (Pearson Corr) were used to evaluate the prediction ability of the model for LWC and MVD in the case of small droplets.

[0077] The formula for calculating the root mean square error is: (3); The calculation formula of Pearson correlation coefficient is: (4); Among them, RMSE is the root mean square error, MSE is the mean square error, Corr(output, output) is the Pearson correlation coefficient, n is the number of samples, N is the sample size, output is the predicted value, and observe is the observed value.

[0078] 5. Numerical simulation of icing: This embodiment uses LEWICE software to perform numerical simulation of aircraft icing. By inputting the wing model, aircraft state parameters, cloud microphysical parameters and exposure time, the icing pattern of the aircraft and the ice thickness of each area of ​​the wing can be obtained, and the maximum icing rate can be calculated.

[0079] By adjusting different parameters, a maximum icing rate lookup table (LUT) for the aircraft is established to determine the severity of icing on the aircraft. The selection of parameter range is crucial for accurate aircraft icing prediction. This embodiment is based on the previous results and refers to Appendix C of FAR Part 25. By counting tens of thousands of icing events, the meteorological elements covering 95% of icing events are extracted as the input parameter range of the LEWICE software, as shown in Table 3. Among them, the temperature is set to -28-4℃, the relative humidity is set to 70%-100%, the flight speed is set to 60-120m / s, and the LWC is set to 0-1g / m 3 , MVD was set to 5-100 μm.

[0080] The method for establishing the maximum icing rate lookup table (LUT) in this embodiment can refer to the aircraft icing degree prediction method based on icing numerical simulation disclosed in patent application CN117493738A.

[0081] Table 3 LEWICE parameter setting table The wing model used in this embodiment is the NACA0012 airfoil. Figure 3 shown. Figure 3 Figure (a) is a schematic diagram of the NAC0012 airfoil; Figure (b) is a curve showing the change of ice type at the leading edge of the wing with exposure time under given conditions; the lower right corner of Figure (b) is a schematic diagram of the change of ice type at the leading edge of the wing from 1 to 20 minutes under given conditions.

[0082] This airfoil is a standard numerical model in Lewis wind tunnels and is widely used in icing simulation studies. In order to examine the correlation between the exposure time to icing conditions and the resulting ice accumulation, the airfoil was tested under fixed environmental parameters (temperature -8°C, flight speed 80m / s, relative humidity 100%, MVD 20µm, LWC 0.8g / cm 3 A series of experiments were conducted with the exposure time ranging from 1 to 20 minutes and the test interval of 1 minute. The test results show that the maximum ice thickness increases evenly with the increase of exposure time, such as Figure 3 This indicates that the growth rate of ice thickness is not affected by exposure time and can be used to assess the severity of aircraft icing.

[0083] 6. Model Validation The CNN-attention model was used to classify the size of the droplets. Multiple thresholds were systematically established in the interval [0,1], each of which was a discriminative criterion: if the predicted value exceeded the threshold, the diagnosis was classified as a large droplet, while the value below the threshold was considered a small droplet. Subsequently, the true positive rate and false positive rate were calculated for these different thresholds to evaluate the diagnostic ability of the CNN-attention model. The receiver operating characteristic curve (ROC) was used to show the relationship between the true positive rate and the false positive rate at different threshold levels, and the area under the curve (AUC) was a quantitative indicator of the model's discriminative ability. When the AUC value is 0.5, it means that there is no discrimination ability; the closer the AUC value is to 1, the better the prediction performance. Figure 4 This is the ROC curve for large droplet classification based on the CNN-Attention0 model. The results show that the model has good training ability and the AUC value of the entire sample set reaches 0.979.

[0084] In order to determine the most appropriate threshold for large liquid classification, this example also calculates the true skill statistics (TSS) at different thresholds. TSS integrates the true positive rate and false positive rate to provide an overall measure of the classification performance of the model. Figure 4 It shows that as the threshold increases from 0 to 1.0, TSS first increases and then gradually decreases, reaching a maximum value at a threshold of 0.12. Therefore, the decisive criterion threshold for distinguishing large and small droplets in this embodiment is set to 0.12. Considering that the missed detection of large droplets in the cloud may lead to serious icing consequences, and the false alarm may lead to unnecessary anti-icing system activation measures, through comprehensive consideration, 0.85 is used as the minimum TSS value in this embodiment to strike a balance between false alarms and missed detections. When TSS reaches 0.87, the true positive rate is 0.98 and the false positive rate is 0.11; when TSS is 0.94, the training set shows better performance, with a true positive rate of 0.99 and a false positive rate of 0.05.

[0085] The true positive rate, false negative rate, and accuracy of the CNN-Attention0 model and the CNN-Attention1 model for large and small droplet classification evaluation of all sample data are listed in Table 4. The results show that the true positive rate is higher than that of the CNN-Attention0 model in the process of large and small droplet classification, reaching 95%, indicating that it has a stronger ability to accurately diagnose large droplets; at the same time, its accuracy rate also reached 90.1%, reflecting its overall accuracy in distinguishing large and small droplets.

[0086] Table 4 Model prediction accuracy table The CNN-Attention0 model is further used to predict microphysical meteorological parameters, including vertical velocity-relative humidity, liquid water content-air pressure, temperature-ice water content, raindrop number concentration-ice number concentration, such as Figure 6 As shown, the applicant also found that the CNN-Attention0 model has significant advantages in predicting flight areas such as layered clouds. Especially in areas with weak vertical motion and high relative humidity (such as Figure 6 The applicant predicts that this is because the CNN-Attention0 model can accurately identify droplets through droplet classification and secondary training and optimize the model based on the accurate identification results. The two precise characteristics make it more accurate when applied to scenes such as layered clouds with complex meteorological characteristics such as reduced airflow and increased moisture. For example, when the RH level is greater than about 70%, the prediction reliability is significantly enhanced.

[0087] from Figure 6 It can also be seen from (b) that CNN-Attention0 also shows significant prediction advantages in areas with low altitude and low liquid water content. For example, when the altitude is less than or equal to 800hPa or the LWC is less than or equal to 0.3g / m 3 , the applicability of the model is significantly enhanced. Figure 6 (c) in Figure 2 shows that the samples where the CNN-Attention0 model correctly predicts large droplets are mainly found in environments with low ice water content. This low ice water content setting is conducive to the growth of large droplets, because excessive ice water content will trigger freezing, thereby hindering the formation and maintenance of large droplets. In addition, from Figure 6 It can also be seen from (d) that the CNN-Attention0 model is particularly suitable for predicting N rain Higher area.

[0088] Prediction ability of CNN-Attention model for MVD and LWC of small droplets: The performance of two CNN-Attention models in predicting LWC and MVD for small droplets (MVD < 100 μm) was evaluated, with RMSE and Corr as evaluation indicators, and the results are listed in Table 5. It can be seen that the two models show basically similar abilities in predicting LWC, and the performance of CNN-Attention0 model is slightly better than that of CNN-Attention1 model. In terms of MVD prediction, CNN-Attention0 model performs well, with RMSE reduced to 25.5 and correlation coefficient reaching 0.75, showing significant effect compared with CNNAttention1 model. This may be due to the ability of CNN-Attention0 model to accurately distinguish between large and small droplets. This targeted focus enables the model to converge more effectively, thereby achieving high-precision prediction of MVD.

[0089] Table 5 Evaluation of two CNN-Attention models on LWC and MVD prediction In order to get a more detailed understanding of the proficiency of the two models in predicting LWC and MVD under small droplet conditions, a density scatter plot of model predictions versus actual observations was generated for the entire data set. The closeness of the scatter plots to the diagonal line indicates the prediction accuracy, as shown in Figure 2. Figure 7-10 The LWC scatter plots of the two models ( Figure 7-10 ) shows a clear concentration of points along the diagonal, indicating that both models have shown remarkable ability in predicting the LWC of small droplets, and the predictions are very close to the actual observations. However, the performance of the two models differed in MVD prediction ( Figure 7-Figure 8 ). The points of the scatter plot of the CNN-Attention1 model are more dispersed, while the scatter plot of the CNN-Attention0 model shows a tighter clustering around the diagonal, indicating that its estimation of the MVD value is more accurate. Therefore, the CNN-Attention0 model has lower systematic bias and overall error in predicting cloud microphysical parameters, especially for small droplets. Overall, the performance evaluation on the full sample dataset shows that both CNN-Attention models perform well in distinguishing large and small droplets. In the current sample dataset, the CNN-Attention0 model is the preferred option due to its higher true positive rate and accuracy.

[0090] VII. Aircraft Icing Severity Forecast 7.1 Icing severity forecast algorithm: It includes two components. The first part obtains the meteorological parameters (T, RH, LWC and MVD) of the target area through meteorological parameter forecasting, and distinguishes between large and small droplets. This embodiment adopts a hybrid model combining WRF and CNN-Attention0 model. The second part is the icing severity forecast, which uses meteorological parameters and aircraft fuselage parameters as input, linearly interpolates the maximum icing rate, and realizes rapid evaluation and prediction of the maximum icing rate in the target area. The icing severity based on the icing rate is divided as follows: trace icing <0.6mm / min, light icing is 0.6-1.0mm / min, moderate icing is 1.1-2.0mm / min, and severe icing is >2.0mm / min.

[0091] 7.2 Evaluation of icing severity prediction algorithms for typical icing events: 1) Single flight The 17th flight of the ICICLE project provided a valuable case study for evaluating the accuracy of the icing severity forecast algorithm. According to the flight crew report, the aircraft took off from Terre Haute Airport at 12:04 pm and flew south. There was no icing between 12:04 pm and 12:50 pm. At 13:13 pm, the pilot detected slight icing above the aircraft using radar. Between 13:26 pm and 13:39 pm, the aircraft entered the cloud layer and encountered SLD. Between 13:39 pm and 15:23 pm, SLD was encountered on the way between Bloomington Normal and Springfield. The aircraft then returned to Terre Haute Airport. Based on the icing severity forecast algorithm, the icing conditions in the altitude range of 750 to 900 hPa from 13:00 to 15:00 on February 17, 2019 were simulated. The simulation results showed that for the time period of 14:00 and 15:00, the aircraft may encounter moderate icing or SLD in the pressure layer of 750-800 hPa, and the simulation results are consistent with the pilot's flight report.

[0092] 2) All flights In order to comprehensively evaluate the accuracy of the icing severity prediction algorithm proposed in this paper, the LEWICE software was used to calculate all data points in the sampling area, and the TPR, FNR and accuracy of the aircraft icing severity algorithm were calculated. The results are shown in Table 6 below: Table 6 TPR, FNR and accuracy of icing severity forecast It can be found that the true positive rate of the icing severity prediction algorithm for slight icing, mild icing, moderate icing and severe icing is 90%, 76%, 72% and 50% respectively, and it has a low false negative rate, which can accurately realize the aircraft icing severity prediction. In summary, the present invention uses the flight data published by the US ICICLE project as input, establishes a CNN model based on the attention mechanism, provides a new method to distinguish large and small droplets of liquid water in the cloud, and can accurately predict the cloud physical parameters such as LWC and MVD in the case of small droplets. First, the cloud physical parameters output by the WRF model are obtained based on the Thomson scheme. It is found that the WRF model underestimates the LWC results and needs further correction. Subsequently, two CNN-Attention models are established to train the output parameters. It is found that the CNN-Attention0 model has a significant advantage in the prediction accuracy of droplet classification, which can reach 90.1%. At the same time, it has extremely high prediction capabilities for small droplet MVD and LWC. Finally, based on the WRF mode, CNN-Attention0 model and icing numerical simulation calculation, an aircraft icing severity prediction algorithm is established to realize the prediction of aircraft icing severity. In general, this embodiment adopts a classification simulation method to exclude large droplets and obtain accurate predictions of icing microphysical parameters (MVD and LWC) under small droplet conditions at low cost. The CNN-Attention0 model constructed in the study shows good capabilities in droplet classification and MVD and LWC prediction under small droplet conditions, and combined with icing numerical simulation calculation, it significantly improves the classification prediction accuracy of aircraft icing severity. The map data shown in this article comes from: World Standard Map (Approval Number: GS (2021) No. 5444).

[0093] It can be understood that the present invention also provides corresponding system products corresponding to the above methods. For example, the present invention also provides a prediction system for cloud physical parameters, a model training subsystem, which includes: a sample acquisition module, used to obtain a first meteorological sample set, the first meteorological sample set includes: a meteorological sample group at multiple times, the meteorological sample group includes: temperature, air pressure, relative humidity, cloud water mass mixing ratio, rain water mass mixing ratio, vertical speed, ice water mass mixing ratio, ice water content, raindrop number concentration, ice number concentration and first liquid water content; a first training module, used to input the first meteorological sample set into a convolutional neural network for iterative calculation, at this time, the input end of the convolutional neural network is: temperature, air pressure, relative humidity, cloud water mass mixing ratio, rain water mass mixing ratio, vertical speed, ice water mass mixing ratio, ice water content, raindrop number concentration, ice number concentration, first liquid water content; the output end of the convolutional neural network is: the probability value of the median volume diameter belonging to a large droplet; wherein the activation function adopts a normalized exponent Function, the normalized exponential function is used to calculate the probability value corresponding to the meteorological sample group, the loss function adopts the binary intersection function, and the binary intersection function is used to calculate the convergence state of the current probability value; a sample screening module is used to combine the meteorological samples whose probability values ​​belong to the set probability interval and whose temperature is greater than the set temperature threshold to form a second meteorological sample set; a second training module is used to input the second meteorological sample set into the convolutional neural network again for update iterative calculation, and finally obtain the corresponding first prediction model, the output of the first prediction model is: the second liquid water content and the median volume diameter; at this time, the activation function adopts the linear rectification function, and the linear rectification function is used to calculate the second liquid water content and the median volume diameter, and the loss function adopts the mean square error function, and the mean square error function is used to calculate the convergence state of the second liquid water content and the median volume diameter.

[0094] In some embodiments, it also includes: a probability setting system, which is used to: select multiple discrimination thresholds between [a, b], the discrimination thresholds are used to judge large droplets and small droplets, and a and b are set probability interval values; set a first verification indicator, the first verification indicator includes: a first ratio R1 of the number of samples correctly classified as large droplets to the actual number of large droplet samples, and a second ratio R2 of the number of samples classified as small droplets to the actual number of small droplet samples; wherein, when the probability value is greater than or equal to the discrimination threshold, it is considered to be a large droplet, and when the probability value is less than the discrimination threshold, it is considered to be a small droplet; calculate multiple receiver operating characteristic curves corresponding to the multiple discrimination thresholds based on the first ratio R1 and the second ratio R2; calculate the area size of the receiver operating characteristic curve; select at least one discrimination threshold corresponding to the largest area size as the discrimination threshold in the model training process.

[0095] The present invention also provides a verification system for model training, including: a sample acquisition module, used to acquire a first meteorological sample set and a corresponding median volume diameter, wherein the first meteorological sample set includes: a meteorological sample group at multiple moments, wherein the meteorological sample group includes: temperature, air pressure, relative humidity, cloud water mass mixing ratio, rain water mass mixing ratio, vertical speed, ice water mass mixing ratio, ice water content, raindrop number concentration, ice number concentration and first liquid water content; a first training module, used to input the first meteorological sample set into a convolutional neural network for iterative calculation, at which time, the input end of the convolutional neural network is: temperature, air pressure, relative humidity, cloud water mass mixing ratio, rain water mass mixing ratio, vertical speed, ice water mass mixing ratio, ice water content, raindrop number concentration, ice number concentration and first liquid water content; the activation function of the convolutional neural network is used to calculate the probability value of the median volume diameter of the sample group, wherein the probability value indicates the probability that the corresponding sample belongs to a large droplet or a small droplet; correspondingly, the output end of the convolutional neural network is: probability value; A sample screening module is used to combine the meteorological samples corresponding to the probability values ​​belonging to the set probability interval and the temperature being greater than the set temperature threshold into a second meteorological sample set; a second training module is used to input the second meteorological sample set into the convolutional neural network again for update iterative calculation, and finally obtain the corresponding first prediction model, the output of the first prediction model is: the second liquid water content and the median volume diameter; a first verification module is used to calculate a first verification index, the first verification index includes: a first ratio R1 of the number of samples correctly classified as large droplets to the actual number of large droplet samples, and / or, a second ratio R2 of the number of samples classified as small droplets to the actual number of small droplet samples; a second verification module is used to determine whether the first verification index meets the set first verification condition, if not, adjust the probability interval and enter the second training module again.

[0096] Furthermore, the probability interval is set by a discrimination threshold. When the probability value is greater than or equal to the discrimination threshold, it is considered to be a large droplet. When the probability value is less than the discrimination threshold, correspondingly, the second verification module is further used to perform the following steps: calculating multiple receiver operating characteristic curves corresponding to the discrimination threshold based on the first ratio R1 and the second ratio R2; calculating the area size of the receiver operating characteristic curve; when the area size is greater than the set area threshold, it is considered that the first verification indicator determines whether it meets the set verification conditions.

[0097] In some embodiments, when the area size is less than or equal to the area threshold, the discrimination threshold is adjusted, and the sample screening module is entered again according to the corresponding new probability interval. In some embodiments, the second sample acquisition module is used to select the meteorological element combination belonging to the set vertical movement interval and the set relative humidity interval from the second meteorological sample set, and form the second test sample set according to the meteorological element combination. In some embodiments, the second sample acquisition module is used to select the meteorological element combination belonging to the set altitude interval from the second meteorological sample set, and form the second test sample set according to the meteorological element combination.

[0098] The present invention also provides an electronic device, comprising: a memory for storing a computer program; a processor for implementing the steps of the method for predicting cloud physical parameters described in any embodiment when executing the computer program. The present invention also provides a computer-readable storage medium, in which a computer program is stored, and when the computer program is executed by the processor, the steps of the method for predicting cloud physical parameters described in any embodiment are implemented. The present invention also provides a computer program product, including a computer program / instruction, which implements the steps of the method for predicting cloud physical parameters described in any embodiment when executed by the processor. It should be noted that the activation function and loss function required by the present invention are the preferred schemes adopted for small sample training scenarios. When the present invention is applied to other types (such as large sample training scenarios), other activation functions or loss functions can also be selected.

[0099] It should be noted that, in this article, the term "includes", "comprising" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "includes a ..." does not exclude the presence of other identical elements in the process, method, article or device including the element. Through the description of the above implementation mode, a person skilled in the art can clearly understand that the above-mentioned embodiment method can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware, but in many cases the former is a better implementation mode. Based on such an understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, disk, CD), and includes several instructions to enable a computer terminal (which can be a mobile phone, computer, server, or network equipment, etc.) to execute the methods described in each embodiment of the present invention. The embodiments of the present invention are described above in conjunction with the accompanying drawings, but the present invention is not limited to the above-mentioned specific implementation modes, which are merely illustrative rather than restrictive. Under the guidance of the present invention, ordinary technicians in this field can also make many forms without departing from the scope of protection of the present invention and the claims, all of which are within the protection of the present invention.

Claims

1. A cloud physical parameter prediction model application method, characterized in that: Includes steps: S401, obtaining a first meteorological sample set including a combination of multiple meteorological elements and a corresponding median volume diameter; S402, selecting a first test sample set from the first meteorological sample set, and respectively training the first test sample set using a first training method and a second training method to obtain a first prediction model and a second prediction model respectively; The step of using the first training method to train and obtain the first prediction model includes: Inputting the first test sample set into a neural convolutional network, where the output of the neural convolutional network includes: a probability value of a median volume diameter, wherein the probability value indicates a probability that the corresponding sample belongs to a large droplet or a small droplet; According to the probability value, the sample predicted to be a small droplet is inputted into the neural convolution network again to update the neural convolution network, and the first prediction model is outputted after the training converges; at this time, the output of the neural convolution network includes: the second liquid water content and the median volume diameter; The step of using the second training method to train a second prediction model includes: Inputting the first test sample set into a neural convolutional network, the output of the neural convolutional network includes: a second liquid water content and a median volume diameter, and outputting a second prediction model after training convergence; S403, selecting a second test sample set from the second meteorological sample set, where the second test sample set does not completely overlap with the first test sample set; S404, using the second test sample set to respectively calculate the prediction accuracy of the first prediction model and the second prediction model; S405: Recommend and select a corresponding prediction model based on the prediction accuracy.

2. The cloud physical parameter prediction model application method according to claim 1, characterized in that: S403 includes the steps of: selecting the meteorological element combination belonging to the set vertical movement interval and the set relative humidity interval from the second meteorological sample set, and forming the second test sample set according to the meteorological element combination.

3. The cloud physical parameter prediction model application method according to claim 1, characterized in that: S403 includes the steps of: selecting the meteorological element combination belonging to the set altitude range from the second meteorological sample set, and forming the second test sample set according to the meteorological element combination.

4. The cloud physical parameter prediction model application method according to claim 1, characterized in that: The meteorological sample group includes: temperature, air pressure, relative humidity, cloud water mass mixing ratio, rain water mass mixing ratio, vertical velocity, ice water mass mixing ratio, ice water content, raindrop number concentration, ice number concentration and first liquid water content.

5. A cloud physical parameter prediction model application system, characterized in that: include: A first sample acquisition module, used to acquire a first meteorological sample set including a combination of multiple meteorological elements and a corresponding median volume diameter; A training module, used for selecting a first test sample set from the first meteorological sample set, and respectively using a first training method and a second training method to train the first test sample set to obtain a first prediction model and a second prediction model; The step of using the first training method to train and obtain the first prediction model includes: Inputting the first test sample set into a neural convolutional network, where the output of the neural convolutional network includes: a probability value of a median volume diameter, wherein the probability value indicates a probability that the corresponding sample belongs to a large droplet or a small droplet; According to the probability value, the sample predicted to be a small droplet is inputted into the neural convolution network again to update the neural convolution network, and the first prediction model is output after the training converges; at this time, the output of the neural convolution network includes: the second liquid water content and the median volume diameter; The step of using the second training method to train a second prediction model includes: Inputting the first test sample set into a neural convolutional network, the output of the neural convolutional network includes: liquid water content and median volume diameter, and outputting a second prediction model after training convergence; A second sample acquisition module, configured to select a second test sample set from the second meteorological sample set, wherein the second test sample set does not completely overlap with the first test sample set; A prediction accuracy calculation module, used to respectively calculate the prediction accuracy of the first prediction model and the second prediction model using the second test sample set; The recommendation module is used to recommend the selection of a corresponding prediction model according to the prediction accuracy.

6. A cloud physical parameter prediction model application system according to claim 5, characterized in that: The second sample acquisition module is used to select the meteorological element combination belonging to the set vertical movement interval and the set relative humidity interval from the second meteorological sample set, and form the second test sample set according to the meteorological element combination.

7. The cloud physical parameter prediction model application system according to claim 5, characterized in that: The second sample acquisition module is used to select the meteorological element combination belonging to the set altitude range from the second meteorological sample set, and form the second test sample set according to the meteorological element combination.

8. The cloud physical parameter prediction model application system according to claim 5, characterized in that: The meteorological sample group includes: temperature, air pressure, relative humidity, cloud water mass mixing ratio, rain water mass mixing ratio, vertical velocity, ice water mass mixing ratio, ice water content, raindrop number concentration, ice number concentration and first liquid water content.

9. An electronic device, characterized in that: include: Memory for storing computer programs; A processor, configured to implement the steps of the cloud physical parameter prediction model application method according to any one of claims 1 to 4 when executing the computer program.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the cloud physical parameter prediction model application method according to any one of claims 1 to 4 are implemented.

Images