A Lightning Warning Method, Device and Equipment Based on Weight Integrated Neural Network

Abstract

The present application discloses a lightning warning method, device and equipment based on a weighted integrated neural network, which relates to the technical field of meteorological observation. First, the present application obtains a meteorological parameter combination composed of all meteorological parameters that can be observed in the observation area, and then separately establishes independent lightning warning neural network models for each different meteorological parameter in the meteorological parameter combination; the lightning forecast results of each model are integrated by means of weighted summation. In this process, the genetic algorithm is used to automatically learn the optimal integrated weight coefficient group corresponding to different sub-meteorological parameter combinations, learn the weight coefficient group for each parameter permutation and combination and save it; in the real-time warning stage, for any available meteorological parameter in the current period, the corresponding lightning warning neural network model and weight group are called to give the lightning forecast result, overcoming the problems of poor stability, poor generalization and large modeling time consumption existing in the existing neural network prediction method.

Description

Technical Field

[0001] The present application relates to the technical field of meteorological observation, and particularly to a lightning warning method, device and equipment based on a weighted integrated neural network. Background Art

[0002] The mainstream technology of current lightning warning is to identify thunderstorm clusters through meteorological parameter observation data such as lightning location data and radar observation data, extrapolate their moving directions to obtain the size of future thunderstorm clusters and their possible occurrence positions, and obtain the probability of lightning occurrence in the region through the threshold relationship between radar and other data and lightning occurrence; or establish the threshold relationship or multiple linear equation between meteorological parameters and future lightning occurrence to predict the probability of future lightning occurrence. For example, by establishing a linear relationship between the future lightning occurrence frequency in the target area and the radar observation data in its surrounding area, the frequency and probability of lightning occurrence are predicted. Since the occurrence mechanism of lightning has not been clarified and the relationship between its occurrence and other meteorological parameters is very complex and may not follow a linear relationship, the prediction accuracy of such methods is not high; such methods need to be adjusted manually for weights and thresholds in different regions and seasons, so the generalization ability is also poor.

[0003] With the development of technology, with the update of modern equipment such as radar and satellite, a large amount of relevant observation data of the atmosphere has been accumulated and the quality has been greatly improved, which has improved the effect of lightning short-term prediction methods and also gave birth to the emergence of neural network lightning warning methods. Compared with using traditional methods, the lightning warning neural network model established by using neural network methods relies on a large number of learnable parameters and can more accurately identify the complex non-linear relationship between lightning and other meteorological parameters. Therefore, today with a large amount of accumulated data and high quality, its warning effect gradually surpasses the mainstream linear methods and makes it one of the important methods for lightning warning. However, the neural network models currently used for lightning warning can only accept data in a fixed form as input, that is, the spatio-temporal resolution and types of meteorological parameters of the input data must be fixed, which leads to the following problems.

[0004] 1. The stability of the model is poor. In the process of real-time warning, it is very likely that the meteorological parameter observations used by the warning model will be missing for a short time or the quality inspection of some observation data will not pass, resulting in some parameters being unavailable. At this time, the warning model cannot continue to make predictions, and this situation will become more frequent as the number of used parameters increases.

[0005] 2. The generalization ability of the model is poor. There are differences in available meteorological parameters in different regions, which will cause the lightning warning neural network model established in the current region to not be able to be extended to other regions with different available meteorological parameters.

[0006] 3. The time consumption for parameter selection during model establishment is large. Currently, the method for selecting meteorological parameters used in lightning warning neural network model establishment generally involves establishing models through permutations and combinations of various parameters and selecting the parameter combination with the optimal modeling effect. The time consumption of this method increases exponentially with the increase in the number of parameters. Summary of the Invention

[0007] The purpose of this application is to provide a lightning warning method, device, and equipment based on a weighted integrated neural network to overcome the problems of poor stability, poor generalization, and large modeling time consumption existing in existing neural network prediction methods.

[0008] To achieve the above purpose, this application provides the following solutions.

[0009] In the first aspect, this application provides a lightning warning method based on a weighted integrated neural network, including the following steps.

[0010] Obtain the historical observation data sequence of each meteorological parameter within the meteorological parameter combination; the meteorological parameter combination is the combination of meteorological parameters that can be observed in the observation area.

[0011] Using the sliding window method, establish a sample set based on the historical observation data sequence of each meteorological parameter respectively, and obtain the sample set corresponding to each meteorological parameter.

[0012] Use the sample set corresponding to each meteorological parameter to train a neural network model respectively, and obtain a lightning warning neural network model corresponding to each meteorological parameter.

[0013] Adopt a genetic algorithm to determine the weights of the lightning forecast results output by the lightning warning neural network models corresponding to each meteorological parameter in each sub-meteorological parameter combination as the weights corresponding to each meteorological parameter in each sub-meteorological parameter combination; each sub-meteorological parameter combination is obtained by permuting and combining each meteorological parameter in the meteorological parameter combination.

[0014] Obtain the current observation data sequence of each meteorological parameter in the current meteorological parameter combination; the current meteorological parameter combination is the combination of meteorological parameters in the observation area that can be observed currently.

[0015] Input the current observation data sequence of each meteorological parameter in the current meteorological parameter combination into the lightning warning neural network model corresponding to each meteorological parameter in the current meteorological parameter combination respectively, and obtain the lightning forecast results corresponding to each meteorological parameter in the current meteorological parameter combination.

[0016] Determine the weights of each meteorological parameter in the sub-meteorological parameter combination whose types of meteorological parameters are consistent with those included in the current meteorological parameter combination as the weights of each meteorological parameter in the current meteorological parameter combination.

[0017] Calculate the final prediction result according to the lightning forecast results corresponding to each meteorological parameter in the current meteorological parameter combination and the weights corresponding to each meteorological parameter in the current meteorological parameter combination.

[0018] In a second aspect, the present application provides a lightning warning device based on a weighted integrated neural network. The lightning warning device based on a weighted integrated neural network applies the above-mentioned lightning warning method based on a weighted integrated neural network. The lightning warning device based on a weighted integrated neural network includes the following modules.

[0019] A historical observation data acquisition module, configured to acquire a historical observation data sequence of each meteorological parameter within a meteorological parameter combination; the meteorological parameter combination is a combination of meteorological parameters that can be observed in the observation area.

[0020] A sample set construction module, configured to use a sliding window method to respectively establish a sample set based on the historical observation data sequences of each meteorological parameter, and obtain a sample set corresponding to each meteorological parameter.

[0021] A model training module, configured to respectively train a neural network model using the sample set corresponding to each meteorological parameter, and obtain a lightning warning neural network model corresponding to each meteorological parameter.

[0022] A weight determination module, configured to use a genetic algorithm to determine the weights of the lightning forecast results output by the lightning warning neural network models corresponding to each meteorological parameter in each sub-meteorological parameter combination, as the weights corresponding to each meteorological parameter in each sub-meteorological parameter combination; each sub-meteorological parameter combination is obtained by performing permutation and combination on each meteorological parameter in the meteorological parameter combination.

[0023] A current observation data acquisition module, configured to acquire a current observation data sequence of each meteorological parameter in the current meteorological parameter combination; the current meteorological parameter combination is a combination of meteorological parameters of the observation area that can be observed currently.

[0024] A prediction module, configured to respectively input the current observation data sequences of each meteorological parameter in the current meteorological parameter combination into the lightning warning neural network models corresponding to each meteorological parameter in the current meteorological parameter combination, and obtain the lightning forecast results corresponding to each meteorological parameter in the current meteorological parameter combination.

[0025] A weight selection module, configured to determine the weights of each meteorological parameter in the sub-meteorological parameter combination that is consistent with the types of meteorological parameters included in the current meteorological parameter combination, as the weights of each meteorological parameter in the current meteorological parameter combination.

[0026] A comprehensive calculation module, configured to calculate the final prediction result according to the lightning forecast results corresponding to each meteorological parameter in the current meteorological parameter combination and the weights corresponding to each meteorological parameter in the current meteorological parameter combination.

[0027] In a third aspect, the present application provides a computer device, including: a memory, a processor, and a computer program stored on the memory and executable on the processor, where the processor executes the computer program to implement the above-mentioned lightning warning method based on a weighted integrated neural network.

[0028] According to the specific embodiments provided by the present application, the present application has the following technical effects.

[0029] The present application provides a lightning warning method, device, and equipment based on a weighted integrated neural network. The present application first obtains a meteorological parameter combination composed of all meteorological parameters that can be observed in the observation area, and then separately establishes independent lightning warning neural network models for each different meteorological parameter in the meteorological parameter combination; integrates the lightning forecast results of each model by weighted summation. In this process, the genetic algorithm is used to automatically learn the optimal integrated weight coefficient group corresponding to different sub-meteorological parameter combinations, learn the weight coefficient group for each parameter permutation and combination (i.e., sub-meteorological parameter combination) and save it; in the real-time warning stage, for any available meteorological parameter in the current time period, call the corresponding model and weight coefficient group to give the lightning forecast result, overcoming the problems of poor stability, poor generalization, and large modeling time consumption of the existing neural network prediction methods. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for use in the embodiments. Obviously, the drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0031] Figure 1 It is a schematic flowchart of a lightning warning method based on a weighted integrated neural network provided by an embodiment of the present application.

[0032] Figure 2 It is a schematic diagram of the principle of a lightning warning method based on a weighted integrated neural network provided by an embodiment of the present application.

[0033] Figure 3 It is a schematic diagram of the overall structure of the AutoFitNet neural network model provided by an embodiment of the present application.

[0034] Figure 4 It is a schematic diagram of the structure of the Multi-Scale Conv layer provided by an embodiment of the present application.

[0035] Figure 5 It is a schematic flowchart of the lightning forecast process provided by an embodiment of the present application.

[0036] Figure 6 A schematic structural diagram of a computer device provided by an embodiment of the present application. Specific implementation manners

[0037] Next, the technical solutions in the embodiments of the present application will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present application.

[0038] To make the above objects, features, and advantages of the present application more obvious and understandable, the present application will be further described in detail below in conjunction with the accompanying drawings and specific implementation manners.

[0039] In an exemplary embodiment, as Figure 1 and Figure 2 shown, a lightning warning method based on a weighted integrated neural network is provided, including the following steps 101-step 108.

[0040] Step 101, obtaining a historical observation data sequence of each meteorological parameter in the meteorological parameter combination; the meteorological parameter combination is a combination of meteorological parameters that can be observed in the observation area.

[0041] Step 102, using a sliding window method to respectively establish a sample set based on the historical observation data sequence of each meteorological parameter, and obtaining a sample set corresponding to each meteorological parameter.

[0042] Step 103, respectively training a neural network model by using the sample set corresponding to each meteorological parameter, and obtaining a lightning warning neural network model corresponding to each meteorological parameter.

[0043] Step 104, using a genetic algorithm to determine the weight of the lightning forecast result output by the lightning warning neural network model corresponding to each meteorological parameter in each sub-meteorological parameter combination, as the weight corresponding to each meteorological parameter in each sub-meteorological parameter combination; each sub-meteorological parameter combination is obtained by permuting and combining each meteorological parameter in the meteorological parameter combination.

[0044] Step 105, obtaining a current observation data sequence of each meteorological parameter in the current meteorological parameter combination; the current meteorological parameter combination is a combination of meteorological parameters in the observation area that can be observed currently; wherein, the current observation data sequence is a sequence formed by arranging the observation data in the sliding window including the current observation data in chronological order, and the sliding window is the sliding window applied in the sliding window method.

[0045] Step 106: Input the current observation data sequences of each meteorological parameter in the current meteorological parameter combination into the lightning warning neural network model corresponding to each meteorological parameter in the current meteorological parameter combination, and obtain the lightning forecast results corresponding to each meteorological parameter in the current meteorological parameter combination.

[0046] Step 107: Determine the weights of each meteorological parameter in the sub-meteorological parameter combination that is consistent with the types of meteorological parameters included in the current meteorological parameter combination, and use them as the weights of each meteorological parameter in the current meteorological parameter combination.

[0047] Step 108: Calculate the final prediction result based on the lightning forecast results corresponding to each meteorological parameter in the current meteorological parameter combination and the weights corresponding to each meteorological parameter in the current meteorological parameter combination.

[0048] Implement the above steps 101 to 108. Based on the weighted integration neural network, through the methods of model separation and weight integration, it allows the input of the model to change, thereby enabling the model to continue running in the case of partial data loss, making the model output more stable forecast results, and at the same time enabling the model to be extended to different regions with different available meteorological parameters. The weight integration mechanism of this neural network will automatically assign higher (lower) weights to meteorological parameters with high (low) correlation with the future lightning occurrence probability, without the need to model permutations and combinations of parameters for screening. Therefore, its modeling time is linearly related to the number of parameters. Compared with traditional neural networks, its modeling time is significantly reduced.

[0049] In the above steps 101 - 108, multi-source observation parameters (including but not limited to lightning location, radar reflectivity, meteorological station observation, satellite observation) within the observation area can be used to predict the lightning occurrence probability (hereinafter referred to as lightning probability) in the target area within the next 0 - 1 hour. This method includes: selecting the observation area, determining the available observation parameters within this area, preprocessing the parameter data, and integrating them into a dataset available for neural network training and testing; using neural networks to establish independent lightning warning neural network models for each different parameter respectively; integrating the forecast results of each lightning warning neural network model through weighted summation, automatically learning the optimal integrated weight coefficient group through genetic algorithms, learning the weight coefficient group for different parameter permutations and combinations respectively and saving them; in the real-time warning stage, input any available parameters in the current period into the model, and call the corresponding lightning warning neural network model and weight group to give the lightning forecast result.

[0050] After the lightning warning neural network model is established, it can still use any permutation and combination of meteorological parameters as input for lightning warning. This enables the model to continue lightning warning when the observation of a certain meteorological parameter is interrupted during real-time forecasting. Therefore, compared with the prior art, the model can perform real-time lightning warning more stably while ensuring the accuracy rate.

[0051] In another exemplary embodiment of the present application, performing step 102 is a process of observing data quality detection and preprocessing. The staff determines the forecasting area and available meteorological parameter observation data, performs quality control and preprocessing on them, and constructs a historical data set (i.e., a sample set) for neural network model training. The specific implementation method is as follows: perform quality detection on the historical observation data of the selected meteorological parameters, remove the unqualified data, interpolate the smaller part of the historical observation data vacancy for each parameter (using methods such as linear interpolation and Gaussian diffusion, and select the method according to the data characteristics), obtain the historical observation data sequence of each meteorological parameter, and use the sliding window method to establish a sample set for each meteorological parameter respectively. Exemplarily, set the lengths of the sliding window and the prediction window to be both 1 hour, use the observation data within one hour as samples (the time resolutions of each meteorological parameter are different), and the lightning probability within the next hour as the annotation (the time resolution is 10 minutes, and each frame contains the lightning that occurs within the next 10 minutes at this time point). Divide the training set, validation set, and test set in a ratio of 8:1:1 in chronological order. The purpose of dividing the sample set in chronological order is to prevent overestimation of the model effect caused by data leakage.

[0052] Exemplarily, the above step 102 specifically includes the following steps 201 - 202.

[0053] Step 201, set the length of the sliding window of the kth meteorological parameter; k = 1, 2,..., K; K is the number of types of meteorological parameters in the meteorological parameter combination.

[0054] Step 202, perform a sliding window operation on the historical observation data sequence of the kth meteorological parameter according to the length of the sliding window of the kth meteorological parameter, determine each historical observation data of the kth meteorological parameter within the sliding window as the sample input, and determine the lightning probability at each time step within the prediction window as the sample annotation, and construct the sample set corresponding to the kth meteorological parameter; the prediction window is located behind the sliding window.

[0055] In another exemplary embodiment of the present application, in the above steps 103 and 104, use the sample set to construct a lightning warning neural network model, and obtain the lightning warning neural network model corresponding to each meteorological parameter and the weight group corresponding to each parameter permutation and combination.

[0056] In another exemplary embodiment of the present application, the neural network model in step 103 above is an AutoFitNet (Auto Adaptation Network) neural network model, and this AutoFitNet neural network model is improved from a ConvLSTM (Convolution Long Short-Term Memory) model. As Figure 3 and Figure 4 shown, the improvement method is to add a fully connected reduction layer and two multi-scale convolutional layers at the input end of the ConvLSTM model, and add two multi-scale convolutional layers and a fully connected amplification layer at the output end of the ConvLSTM model.

[0057] As Figure 3 shown, the AutoFitNet neural network model of the present application adds a fully connected reduction layer (Multi-Layer Perceptron Down, MLP Down), a fully connected amplification layer (Multi-Layer Perceptron Up, MLP Up) and a multi-scale convolutional layer (Multi-Scale Conv) on the basis of the ConvLSTM model. The fully connected reduction layer enables the model to adapt to inputs of any size without changing the structure; the multi-scale convolutional layer can extract spatial features of multiple resolutions in the input.

[0058] Denote the input as [X 1 , X 2 ,... X Tn , the number of observed frames within its sliding window (i.e., 1 hour) is Tn, and the Tn values of each parameter are inconsistent; denote the output as [Y 1 , Y 2 ,... Y Tm , the number of frames within its 1 hour is Tm, and Tm can be adjusted according to requirements. The specific working process of the AutoFitNet neural network model is as follows:

[0059] Step1: Use the fully connected reduction layer (MLP Down) to compress the features of the input. The formula for using the fully connected reduction layer is shown in Equation (1).

[0060]

[0061] Among them, is the feature map output by the fully connected reduction layer, is the observed value of the meteorological parameter of the feature point in the th row and th column of the feature map input by the fully connected reduction layer, that is is 's th row Observation values of meteorological parameters of the feature points in the column Is the Row Weight parameters of the feature points in the column And Are the number of rows and columns of the feature points of the feature map input to the fully connected reduction layer respectively

[0062] The input of the fully connected reduction layer is data with any resolution, and the output is a feature map with 128 channels and a resolution of 32×32

[0063] Step2: Stack two multi-scale convolutional (Multi-Scale Conv) layers (whose structure is as Figure 4 Shown) to extract the spatial features of the input, and use the shortcut (shortest path) technology to accelerate the model convergence speed

[0064] Step3: Use the ConvLSTM layer in the ConvLSTM model to learn the temporal relationship features between the input and output. When the ConvLSTM layer runs, the following calculations need to be performed for each time step The following calculations are performed

[0065] Input gate:

[0066] (2)

[0067] Forget gate:

[0068] (3)

[0069] Candidate cell state:

[0070] (4)

[0071] Cell state update:

[0072] (5)

[0073] Output gate:

[0074] (6)

[0075] Hidden state update:

[0076] (7)

[0077] Among them, Is the output state of the input gate at time step Is the convolution kernel of the input state of the input gate ​is the time step of the input state, is the convolution kernel of the hidden state of the input gate, is the time step of the hidden state, is the bias of the input gate, is the Sigmod activation function, represents the convolution operation. is the output state of the forget gate at time step , is the convolution kernel of the input state of the forget gate, is the convolution kernel of the hidden state of the forget gate, is the bias of the forget gate. is the time step of the output state of the output gate, is the convolution kernel of the input state of the output gate, is the convolution kernel of the hidden state of the output gate, is the bias of the output gate. is the time step of the hidden state, represents element-wise multiplication of matrix elements, is the time step of the cell state, is the time step of the cell state, is the time step of the candidate cell state, and are the convolution kernels of the input state and the hidden state of the hidden layer respectively, is the bias of the hidden layer, is the tanh activation function.

[0078] The ConvLSTM model of this application is divided into an encoding module and a decoding module. The encoding module is used for learning input time features, and the decoding module is used for learning output time features. Both the encoding module and the decoding module contain two ConvLSTM layers. In the encoding module, the input of each time step of the ConvLSTM layer is , and the inputs of the cell state and the hidden state are the cell state and the hidden state at time step , and the output is the hidden state at time step . The initial input cell state and hidden state are zero tensors; in the decoding module, the input of one time step of the ConvLSTM layer is the hidden state at time step , and the inputs of the cell state and the hidden state are at time step Cell state and hidden state , the output is the hidden state [H 1 ,H 2 ,...,H m . The initial input cell state and hidden state are the outputs of the last time step of the encoding module (as Figure 3 shown).

[0079] The hidden state [H 1 ,H 2 ,...,H m output by the decoding module is mapped into the annotation [Y 1 ,Y 2 ,...,Y m through multiple multi-scale convolutional layers and fully connected amplification layers.

[0080] In the above step 103, the following loss function is used for model training.

[0081] (8)

[0082] where is the loss function, is the length of the prediction window, is the error weight at time step within the prediction window, used to adjust the prediction time points that the model focuses on, is the total number of feature points, , and are the number of rows and columns of the feature points of the target feature map respectively, is the adjustment weight to make the model tend to learn the features of small samples with higher lightning content, is the adjustment coefficient, and the target feature map is the feature map at time step of the annotation or the feature map at time step output by the neural network model, is the total number of lightning in the feature map at time step of the annotation, is the lightning probability of the feature point at the th row and th column of the feature map at time step of the annotation, is the lightning probability of the feature point at the th row and th column of the feature map at time step output by the neural network model.

[0083] After training, a lightning warning neural network model corresponding to each meteorological parameter is obtained.

[0084] In another exemplary embodiment, in the above step 104, the sample is input into the established model to output the lightning forecast result, and the annotation calculation error is calculated (using formula (8)), and its optimization problem is shown as follows.

[0085] (9)

[0086] Wherein, is the weight vector, , , , , are the weights of the lightning forecast results output by the lightning warning neural network models corresponding to the first, second, th, th meteorological parameters, is a set of real numbers of dimension, is the rd loss function value of the th meteorological parameter in the th sample of the th sub-meteorological parameter combination, is the number of types of meteorological parameters in the th sub-meteorological parameter combination, and

[0087] Output the lightning warning neural network models corresponding to the above each parameter, and the weight groups corresponding to the parameter permutations and combinations.

[0088] In another exemplary embodiment, the above steps 105-step 108 are the process of performing lightning real-time warning, as Figure 5 shown, specifically: automatically obtain the observation data of the selected meteorological parameters within 1 hour before the current moment (that is, the current observation data sequence of each meteorological parameter in the current meteorological parameter combination), perform automated data quality inspection and preprocessing, and screen out the parameters with observations and meeting the standards, call the model corresponding to the parameter and the weight group corresponding to the current meteorological parameter combination to obtain the final prediction result. This process does not require manual operation and is fully automated.

[0089] Based on the same inventive concept, an embodiment of the present application further provides a lightning warning device based on a weighted integrated neural network for implementing the above-mentioned lightning warning method based on a weighted integrated neural network. The implementation solutions provided by this device to solve problems are similar to the implementation solutions described in the above method. Therefore, the specific limitations in one or more embodiments of the lightning warning device based on a weighted integrated neural network provided below can refer to the limitations on the lightning warning method based on a weighted integrated neural network in the foregoing, and will not be elaborated here.

[0090] In an exemplary embodiment, a lightning warning device based on a weighted integrated neural network is provided, including the following modules.

[0091] A historical observation data acquisition module, configured to acquire a historical observation data sequence of each meteorological parameter within a meteorological parameter combination; the meteorological parameter combination is a combination of meteorological parameters that can be observed in the observation area.

[0092] A sample set construction module, configured to use a sliding window method to respectively establish a sample set based on the historical observation data sequence of each meteorological parameter, and obtain a sample set corresponding to each meteorological parameter.

[0093] A model training module, configured to respectively train a neural network model using the sample set corresponding to each meteorological parameter, and obtain a lightning warning neural network model corresponding to each meteorological parameter.

[0094] A weight determination module, configured to use a genetic algorithm to determine the weight of the lightning forecast result output by the lightning warning neural network model corresponding to each meteorological parameter in each sub-meteorological parameter combination, as the weight corresponding to each meteorological parameter in each sub-meteorological parameter combination; each sub-meteorological parameter combination is obtained by permuting and combining each meteorological parameter in the meteorological parameter combination.

[0095] A current observation data acquisition module, configured to acquire a current observation data sequence of each meteorological parameter in the current meteorological parameter combination; the current meteorological parameter combination is a combination of meteorological parameters of the observation area that can be observed currently; wherein, the current observation data sequence is a sequence formed by arranging the observation data within the sliding window including the current observation data in chronological order, and the sliding window is the sliding window applied in the sliding window method.

[0096] A prediction module, configured to respectively input the current observation data sequence of each meteorological parameter in the current meteorological parameter combination into the lightning warning neural network model corresponding to each meteorological parameter in the current meteorological parameter combination, and obtain the lightning forecast result corresponding to each meteorological parameter in the current meteorological parameter combination.

[0097] A weight selection module, configured to determine the weight of each meteorological parameter in a sub-meteorological parameter combination that is consistent with the types of meteorological parameters included in the current meteorological parameter combination, and use it as the weight of each meteorological parameter in the current meteorological parameter combination.

[0098] An integrated calculation module, configured to calculate a final prediction result according to the lightning forecast results corresponding to each meteorological parameter in the current meteorological parameter combination and the weights corresponding to each meteorological parameter in the current meteorological parameter combination.

[0099] In an exemplary embodiment, a computer device is provided. The computer device can be a server or a terminal, and its internal structure diagram can be as Figure 6 shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, abbreviated as I / O), and a communication interface. Among them, the processor, the memory, and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and external devices. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, it implements a lightning warning method based on a weighted integrated neural network.

[0100] Those skilled in the art can understand that Figure 6 the structure shown in

[0101] is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different component layout. In an exemplary embodiment, a computer device is provided, including a memory and a processor. A computer program is stored in the memory, and when the processor executes the computer program, the steps in the above method embodiments are implemented.

[0102] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above methods. Among them, any reference to a memory, database, or other medium used in the embodiments provided in this application can include at least one of non-volatile and volatile memories. Non-volatile memories can include read-only memory (ROM), magnetic tapes, floppy disks, flash memories, optical memories, high-density embedded non-volatile memories, resistive random access memories (ReRAM), magnetoresistive random access memories (MRAM), ferroelectric random access memories (FRAM), phase change memories (PCM), graphene memories, etc. Volatile memories can include random access memory (RAM) or external cache memories, etc. By way of illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc.

[0103] The databases involved in the embodiments provided in this application can include at least one of relational databases and non-relational databases. Non-relational databases can include distributed databases based on blockchain, etc., without limitation. The processors involved in the embodiments provided in this application can be general-purpose processors, central processors, graphics processors, digital signal processors, programmable logics, data processing logics based on quantum computing, etc., without limitation.

[0104] The technical features of the above embodiments can be combined arbitrarily. For the sake of concise description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered to be within the scope described in this specification.

[0105] In this article, specific examples are used to elaborate on the principles and implementation manners of this application. The description of the above embodiments is only used to help understand the method and its core idea of this application; at the same time, for those of ordinary skill in the art, according to the idea of this application, there will be changes in the specific implementation manners and application scopes. In summary, the content of this specification should not be construed as a limitation to this application.

Claims

1. A lightning warning method based on weighted integrated neural network, characterized in that: include: Acquire a historical observation data sequence of each meteorological parameter in a meteorological parameter combination; the meteorological parameter combination is a combination of meteorological parameters that can be observed in the observation area; Using the sliding window method, sample sets are established based on the historical observation data series of each meteorological parameter to obtain the sample sets corresponding to each meteorological parameter; The neural network model is trained respectively using the sample set corresponding to each meteorological parameter to obtain the lightning warning neural network model corresponding to each meteorological parameter; A genetic algorithm is used to determine the weight of the lightning forecast result output by the lightning warning neural network model corresponding to each meteorological parameter in each sub-meteorological parameter combination as the weight corresponding to each meteorological parameter in each sub-meteorological parameter combination; each sub-meteorological parameter combination is obtained by arranging and combining each meteorological parameter in the meteorological parameter combination; Obtaining a current observation data sequence of each meteorological parameter in a current meteorological parameter combination; the current meteorological parameter combination is a combination of meteorological parameters of an observation area that can be currently observed; Inputting the current observation data sequence of each meteorological parameter in the current meteorological parameter combination into the lightning warning neural network model corresponding to each meteorological parameter in the current meteorological parameter combination, respectively, to obtain the lightning forecast result corresponding to each meteorological parameter in the current meteorological parameter combination; Determine the weight of each meteorological parameter in the sub-meteorological parameter combination that is consistent with the type of meteorological parameters included in the current meteorological parameter combination as the weight of each meteorological parameter in the current meteorological parameter combination; Calculate the final prediction result according to the lightning forecast result corresponding to each meteorological parameter in the current meteorological parameter combination and the weight corresponding to each meteorological parameter in the current meteorological parameter combination; The neural network model is obtained by improving the ConvLSTM model; The improvement method is to add a fully connected reduction layer and two multi-scale convolution layers at the input end of the ConvLSTM model, and add two multi-scale convolution layers and a fully connected amplification layer at the output end of the ConvLSTM model; The optimization goal in the process of using genetic algorithm to determine the weight of the lightning forecast result output by the lightning warning neural network model corresponding to each meteorological parameter in each sub-meteorological parameter combination is: ； in, is the weight vector, , , , , For the first, second, Type The weights of the lightning forecast results output by the lightning warning neural network model corresponding to the meteorological parameters, for dimensional set of real numbers, For the The first The meteorological parameters The loss function value of samples is For the The number of meteorological parameter types in each meteorological parameter combination. is the number of samples for each meteorological parameter.

2. The lightning warning method based on weighted integrated neural network according to claim 1 is characterized in that: Using the sliding window method, sample sets are established based on the historical observation data series of each meteorological parameter, and the sample sets corresponding to each meteorological parameter are obtained, including: Set the length of the sliding window of the kth meteorological parameter; k=1, 2, ..., K; K is the number of meteorological parameters in the meteorological parameter combination; A sliding window operation is performed on the historical observation data sequence of the kth meteorological parameter according to the length of the sliding window of the kth meteorological parameter, each historical observation data of the kth meteorological parameter in the sliding window is determined as a sample input, and the lightning probability of each time step in the prediction window is determined as a sample label to construct a sample set corresponding to the kth meteorological parameter; the prediction window is located after the sliding window.

3. The lightning warning method based on weighted integrated neural network according to claim 1 is characterized in that: The fully connected reduction layer is used to compress the size of the input feature using the following formula; ； in, is the feature map output by the fully connected reduction layer, is the first feature map in the fully connected reduction layer input Line The observed values ​​of meteorological parameters of the characteristic points of the column, is the first feature map in the fully connected reduction layer input Line The weight parameter of the feature points of the column, and are the number of rows and columns of feature points in the feature graph input to the fully connected reduction layer, respectively. is the ReLU activation function.

4. The lightning warning method based on weighted integrated neural network according to claim 1 is characterized in that: The ConvLSTM model includes an encoding module and a decoding module; the encoding module and the decoding module each include two ConvLSTM layers.

5. The lightning warning method based on weighted integrated neural network according to claim 4 is characterized in that: The ConvLSTM layer includes: an input gate, a forget gate and an output gate; The formula of the input gate is: ； in, is the time step The output state of the input gate, is the convolution kernel of the input state of the input gate, is the time step The input state, is the convolution kernel of the hidden state of the input gate, is the time step The hidden state of is the bias of the input gate, is the Sigmod activation function, Represents the convolution operation; The formula of the forget gate is: ； in, is the time step The output state of the forget gate is is the convolution kernel of the input state of the forget gate, is the convolution kernel of the hidden state of the forget gate, is the bias of the forget gate; The formula for the output gate is: ； in, is the time step The output state of the output gate, is the convolution kernel of the input state of the output gate, is the convolution kernel of the hidden state of the output gate, is the bias of the output gate; The update formula for the hidden state of the ConvLSTM layer is: ； ； ； in, is the time step The hidden state of means that the matrix elements are multiplied one by one, is the time step The cell state, is the time step The cell state, is the time step Candidate cell states, and are the convolution kernels of the input state and hidden state of the hidden layer, is the bias of the hidden layer, is the tanh activation function.

6. The lightning warning method based on weighted integrated neural network according to claim 1 is characterized in that: The loss function for training the neural network model is: ； in, is the loss function, is the length of the prediction window, is the time step in the prediction window The error weight, is the total number of feature points, , and are the number of rows and columns of feature points in the target feature map, respectively. is the adjustment coefficient, and the target feature map is the marked time step The feature map or the time step of the neural network model output The feature map of The time step of the label The total number of lightning flashes in the feature graph of The time step of the label The feature map of Line The lightning probability of the characteristic points of the column, The time step output by the neural network model The feature map of Line The lightning probability of the characteristic points of the column.

7. A lightning warning device based on a weighted integrated neural network, characterized in that: The lightning warning device based on the weighted integrated neural network applies the lightning warning method based on the weighted integrated neural network according to any one of claims 1 to 6, and the lightning warning device based on the weighted integrated neural network comprises: A historical observation data acquisition module is used to acquire a historical observation data sequence of each meteorological parameter in a meteorological parameter combination; the meteorological parameter combination is a combination of meteorological parameters that can be observed in the observation area; A sample set building module is used to use a sliding window method to build sample sets based on the historical observation data sequence of each meteorological parameter, so as to obtain a sample set corresponding to each meteorological parameter; A model training module is used to train a neural network model using a sample set corresponding to each meteorological parameter to obtain a lightning warning neural network model corresponding to each meteorological parameter; A weight determination module is used to determine the weight of the lightning forecast result output by the lightning warning neural network model corresponding to each meteorological parameter in each sub-meteorological parameter combination by using a genetic algorithm as the weight corresponding to each meteorological parameter in each sub-meteorological parameter combination; each sub-meteorological parameter combination is obtained by arranging and combining each meteorological parameter in the meteorological parameter combination; A current observation data acquisition module is used to acquire a current observation data sequence of each meteorological parameter in a current meteorological parameter combination; the current meteorological parameter combination is a combination of meteorological parameters of an observation area that can be currently observed; A prediction module is used to input the current observation data sequence of each meteorological parameter in the current meteorological parameter combination into the lightning warning neural network model corresponding to each meteorological parameter in the current meteorological parameter combination, and obtain the lightning forecast result corresponding to each meteorological parameter in the current meteorological parameter combination; A weight selection module is used to determine the weight of each meteorological parameter in the sub-meteorological parameter combination that is consistent with the type of meteorological parameters included in the current meteorological parameter combination as the weight of each meteorological parameter in the current meteorological parameter combination; The comprehensive calculation module is used to calculate the final prediction result according to the lightning forecast result corresponding to each meteorological parameter in the current meteorological parameter combination and the weight corresponding to each meteorological parameter in the current meteorological parameter combination.

8. A computer device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the lightning warning method based on a weighted integrated neural network according to any one of claims 1 to 6.

Images