Snowfall calculation method, device, equipment and storage medium

Abstract

The present invention relates to the field of data processing technology, and discloses a snowfall calculation method, device, equipment and storage medium. By acquiring a geographic environment parameter set and snowfall associated parameter information of a target monitoring area, a geographic parameter feature map is constructed using the geographic environment parameter set, and a plurality of migration reference areas are generated in a geographic environment database. An independent reference snowfall prediction model is constructed according to the historical snowfall information of each migration reference area. Then, the reference snowfall prediction model is migrated based on model weight mapping, and the migration strategy is adjusted based on the model optimization period to achieve model migration and optimization between the migration reference area and the target monitoring area. On the one hand, the accuracy of snowfall calculation in the monitoring area under data scarce scenarios is ensured, and on the other hand, the degree of adaptation between the snowfall prediction model and the target monitoring area is improved, so as to avoid the problem of model incompatibility caused by the model migration process, and the model has good scene adaptability.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a snowfall calculation method, device, equipment and storage medium. Background Art

[0002] Snowfall is one of the important parameters in meteorology, and its calculation results are directly related to the accuracy and refinement of weather forecasts. Accurate snowfall forecasts can help governments, businesses and the public prepare in advance to deal with problems such as traffic congestion, energy supply shortages, and crop losses that may be caused by snowfall. At the same time, snowfall data is also crucial for water resources management, ecological environment monitoring, climate change research and other fields. By observing and recording snowfall over a long period of time, we can analyze climate change trends, assess water resources reserves and distribution, and provide a scientific basis for formulating reasonable water resources utilization and protection strategies.

[0003] At present, the snowfall calculation in the monitoring area is usually carried out by manual measurement, weighing, remote sensing measurement, numerical simulation and artificial intelligence-based measurement. Among them, the measurement method based on artificial intelligence technology can integrate multiple data sources, has high calculation accuracy, real-time and refinement, and has the advantages of automation and intelligence. However, in some scenarios, the measurement method based on artificial intelligence technology faces some limitations: for example, in remote areas or areas where the snowfall-related information collection and snowfall monitoring system has just been built, the initial reference data is very limited, which makes it difficult to meet the artificial intelligence model's demand for large amounts of data and to accurately extract features related to the snowfall formation environment, resulting in insufficient model training and insufficient calculation accuracy.

[0004] Therefore, how to improve the accuracy of snowfall calculation in monitored areas using artificial intelligence models in data-scarce scenarios is a technical problem that needs to be solved urgently. Summary of the invention

[0005] The main purpose of the present invention is to provide a snowfall calculation method, device, equipment and storage medium to solve the above-mentioned technical problems.

[0006] To achieve the above object, the present invention provides a snowfall calculation method, comprising the following steps:

[0007] Generate a target geographic parameter feature map according to a set of geographic environment parameters of a target monitoring area; wherein the target geographic parameter feature map includes a first element visualization feature having a first RGB color layer and a second element visualization feature having a second RGB color layer;

[0008] Based on the first RGB color layer and the second RGB color layer of the target geographic parameter characteristic map, performing geographic parameter characteristic map matching in a geographic environment database to determine a plurality of migration reference areas;

[0009] Using several pieces of historical snowfall information stored in the historical snowfall database of the migration reference area, a reference snowfall amount prediction model for the migration reference area is constructed; wherein each piece of historical snowfall information includes historical snowfall associated parameter information and historical snowfall amount information corresponding to the historical monitoring period;

[0010] Perform model migration based on model weight mapping on several reference snowfall prediction models to obtain the initial snowfall prediction model of the target monitoring area;

[0011] Obtain snowfall associated parameter information of the target monitoring area during several model optimization periods and the target monitoring period, input the snowfall associated parameter information of the model optimization period into the initial snowfall prediction model, calculate the predicted snowfall of the target monitoring area during the model optimization period, adjust the mapping relationship of the model migration action based on the predicted snowfall value and the actual snowfall value during the model optimization period, optimize the initial snowfall prediction model of the target monitoring area, and obtain the final snowfall prediction model;

[0012] The snowfall associated parameter information is input into the final snowfall prediction model to calculate the predicted snowfall in the target monitoring area during the target monitoring period.

[0013] Optionally, the first element visualization feature includes a terrain altitude value, and the second element visualization feature includes a vegetation coverage rate; and the step of generating a target geographic parameter feature map according to a geographic environment parameter set of a target monitoring area specifically includes:

[0014] Acquire a set of geographical environment parameters of the target monitoring area; wherein the set of geographical environment parameters includes the terrain altitude value and vegetation coverage rate of each location in the target monitoring area;

[0015] Generate a first RGB color layer according to the terrain altitude value of each position in the target monitoring area and the preset first conversion relationship between the terrain altitude and the first color channel in the RGB color value, and generate a second RGB color layer according to the vegetation coverage rate of each position in the target monitoring area and the preset second conversion relationship between the vegetation coverage rate and the second color channel in the RGB color value;

[0016] Based on the first RGB color layer and the second RGB color layer, a target geographic parameter feature map of the target monitoring area is constructed.

[0017] Optionally, based on the first RGB color layer and the second RGB color layer of the target geographic parameter characteristic map, performing geographic parameter characteristic map matching in a geographic environment database to determine a plurality of migration reference area steps specifically includes:

[0018] Accessing a geographic environment database of a target reference area; wherein the geographic environment database stores a reference geographic parameter characteristic map of the target reference area, and the target reference area is configured to be composed of a plurality of reference sub-areas having collected historical snowfall information that meets training requirements;

[0019] Based on the target geographic parameter characteristic map and the first RGB color layer and the second RGB color layer of the reference geographic parameter characteristic map, a plurality of migration reference areas are determined in the target reference area.

[0020] Optionally, based on the target geographic parameter feature map and the first RGB color layer and the second RGB color layer of the reference geographic parameter feature map, a step of determining a plurality of migration reference areas in the target reference area specifically includes:

[0021] Extract the average RGB color value of the target geographic parameter feature map in the first RGB color layer, the RGB color value and position of each inflection point in the first RGB color layer, and the average RGB color value of the region corresponding to the position of each inflection point in the second RGB color layer to construct a target geographic feature set;

[0022] The target geographic feature set and the reference geographic parameter feature map are input into an image region recognition model including a YOLO structure and an image similarity algorithm, and the top N migration reference regions in the target reference region that are most similar to the target monitoring region are output.

[0023] Optionally, the step of constructing a reference snowfall amount prediction model for the migration reference area by using a plurality of pieces of historical snowfall information stored in a historical snowfall database of the migration reference area specifically includes:

[0024] Access the historical snowfall database of each migration reference area, and extract historical snowfall associated parameter information and historical snowfall amount information of each historical monitoring period corresponding to each migration reference area;

[0025] The historical snowfall correlation parameter information and the historical snowfall information of each historical monitoring period are used as a set of training data samples to construct a training sample set for each migration reference area, and the training is input into the recurrent neural network constructed for each migration reference area to obtain a reference snowfall prediction model when the training times reach a preset number or the training converges;

[0026] The recurrent neural network includes an input layer, a hidden layer and an output layer. The input layer transmits the training data sample to the hidden layer after receiving it. The hidden layer updates the hidden state according to the historical snowfall associated parameter information and the historical snowfall information at different times in the time series. The output layer is used to obtain the predicted snowfall according to the hidden state.

[0027] When training the recurrent neural network, the network weights are adjusted through the back propagation algorithm to minimize the error between the predicted value and the actual value, and the final reference snowfall prediction model is obtained.

[0028] Optionally, a model migration action based on model weight mapping is performed on several reference snowfall prediction models to obtain an initial snowfall prediction model step for the target monitoring area, specifically including:

[0029] Obtaining the similarity between each migration reference area and the target monitoring area output by the above-mentioned image area recognition model, and converting the ratio of the similarity into a migration mapping weight of the reference snowfall prediction model corresponding to each migration reference area;

[0030] Based on the migration mapping weight of each reference snowfall prediction model, a conversion relationship from several reference snowfall prediction models to the initial snowfall prediction model is established by adopting the weighted sum of the snowfall prediction value of each reference snowfall prediction model and the corresponding migration mapping weight, and the initial snowfall prediction model of the target monitoring area is obtained, whose output value is the weighted sum of the snowfall prediction value of each reference snowfall prediction model and the corresponding migration mapping weight.

[0031] Optionally, based on the predicted snowfall value and the actual snowfall value during the model optimization period, the mapping relationship of the model migration action is adjusted, the initial snowfall prediction model of the target monitoring area is optimized, and the steps of obtaining the final snowfall prediction model are specifically including:

[0032] Calculate the average snowfall difference between the predicted snowfall value and the actual snowfall value in each model optimization period, and determine whether the average snowfall difference is less than a preset difference;

[0033] If yes, the current initial snowfall prediction model is used as the final snowfall prediction model; if no, the mapping relationship of the model migration action is adjusted to optimize the initial snowfall prediction model of the target monitoring area to obtain the final snowfall prediction model;

[0034] Among them, the initial snowfall prediction model of the target monitoring area is optimized to obtain the final snowfall prediction model, which specifically includes the following execution process:

[0035] Taking the adjustment ratio of the migration mapping weight of each reference snowfall prediction model to be less than the target ratio as a constraint condition, and taking the average value of the snowfall difference between the predicted snowfall value and the actual snowfall value in each model optimization period to be less than the minimum value of the preset difference as the goal, the migration mapping weight of each reference snowfall prediction model in the model migration process is adjusted to obtain the migration mapping weight of each reference snowfall prediction model after optimization, and the final snowfall prediction model is established.

[0036] In addition, in order to achieve the above-mentioned purpose, the present invention also provides a snowfall calculation device, comprising:

[0037] A generating module, used for generating a target geographic parameter characteristic map according to a set of geographic environment parameters of a target monitoring area; wherein the target geographic parameter characteristic map comprises a first element visualization feature having a first RGB color layer and a second element visualization feature having a second RGB color layer;

[0038] A determination module, configured to perform geographic parameter feature map matching in a geographic environment database based on a first RGB color layer and a second RGB color layer of the target geographic parameter feature map, and determine a plurality of migration reference areas;

[0039] A construction module is used to construct a reference snowfall prediction model for the migration reference area by using a number of historical snowfall information stored in a historical snowfall database of the migration reference area; wherein each piece of historical snowfall information includes historical snowfall associated parameter information and historical snowfall information corresponding to a historical monitoring period;

[0040] A migration module is used to perform a model migration action based on model weight mapping on several reference snowfall prediction models to obtain an initial snowfall prediction model for a target monitoring area;

[0041] An optimization module is used to obtain snowfall associated parameter information of a target monitoring area during several model optimization periods and a target monitoring period, input the snowfall associated parameter information of the model optimization period into the initial snowfall prediction model, calculate the predicted snowfall of the target monitoring area during the model optimization period, adjust the mapping relationship of the model migration action based on the predicted snowfall value and the actual snowfall value during the model optimization period, optimize the initial snowfall prediction model of the target monitoring area, and obtain a final snowfall prediction model;

[0042] The calculation module is used to input the snowfall associated parameter information into the final snowfall prediction model to calculate the predicted snowfall in the target monitoring area during the target monitoring period.

[0043] In addition, in order to achieve the above-mentioned purpose, the present invention also provides a snowfall calculation device, which includes: a memory, a processor, and a snowfall calculation program stored in the memory and executable on the processor, and when the snowfall calculation program is executed by the processor, the steps of the snowfall calculation method as described above are implemented.

[0044] In addition, in order to achieve the above-mentioned purpose, the present invention also provides a storage medium, on which a snowfall calculation program is stored, and when the snowfall calculation program is executed by a processor, the steps of the above-mentioned snowfall calculation method are implemented.

[0045] The beneficial effects of the present invention are as follows: a snowfall calculation method, device, equipment and storage medium are proposed, by obtaining a geographic environment parameter set and snowfall associated parameter information of a target monitoring area, a geographic parameter feature map is constructed using the geographic environment parameter set, and then several migration reference areas are generated in a geographic environment database, and independent reference snowfall prediction models are constructed according to the historical snowfall information of each migration reference area, and then a model migration action based on model weight mapping is performed on several reference snowfall prediction models, and the migration strategy is adjusted based on the model optimization period to realize model migration and optimization between the migration reference area and the target monitoring area. On the one hand, it ensures the accuracy of snowfall calculation in the monitoring area using an artificial intelligence model in data-scarce scenarios, and on the other hand, it improves the degree of adaptation of the snowfall prediction model to the target monitoring area, avoids the problem of model incompatibility caused by the model migration process, and has good scene adaptability. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0047] Figure 1 The figure is a schematic diagram of the device structure of the hardware operating environment involved in the embodiment of the present invention.

[0048] Figure 2 Schematic diagram of the process of an embodiment of a method for calculating snowfall amount according to the present invention.

[0049] Figure 3 The figure is a structural block diagram of a snowfall calculation device in an embodiment of the present invention.

[0050] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0051] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0052] like Figure 1 As shown, Figure 1 It is a schematic diagram of the device structure of the hardware operating environment involved in the embodiment of the present invention.

[0053] like Figure 1 As shown, the device may include: a processor 1001, such as a CPU, a communication bus 1002, a user interface 1003, a network interface 1004, and a memory 1005. Among them, the communication bus 1002 is used to realize the connection and communication between these components. The user interface 1003 may include a display screen (Display), an input unit such as a keyboard (Keyboard), and the optional user interface 1003 may also include a standard wired interface and a wireless interface. The network interface 1004 may optionally include a standard wired interface and a wireless interface (such as a WI-FI interface). The memory 1005 may be a high-speed RAM memory, or it may be a stable memory (non-volatile memory), such as a disk memory. The memory 1005 may also be a storage device independent of the aforementioned processor 1001.

[0054] Those skilled in the art will understand that Figure 1 The structure of the device shown in the figure does not constitute a limitation of the device, and may include more or less components than shown in the figure, or combine certain components, or arrange the components differently.

[0055] like Figure 1 As shown, the memory 1005 as a computer storage medium may include an operating system, a network communication module, a user interface module, and a snowfall calculation program.

[0056] exist Figure 1 In the terminal shown, the network interface 1004 is mainly used to connect to the backend server and perform data communication with the backend server; the user interface 1003 is mainly used to connect to the client (user end) and perform data communication with the client; and the processor 1001 can be used to call the snowfall calculation program stored in the memory 1005 and perform the following operations:

[0057] Generate a target geographic parameter feature map according to a set of geographic environment parameters of a target monitoring area; wherein the target geographic parameter feature map includes a first element visualization feature having a first RGB color layer and a second element visualization feature having a second RGB color layer;

[0058] Based on the first RGB color layer and the second RGB color layer of the target geographic parameter characteristic map, performing geographic parameter characteristic map matching in a geographic environment database to determine a plurality of migration reference areas;

[0059] Using several pieces of historical snowfall information stored in the historical snowfall database of the migration reference area, a reference snowfall amount prediction model for the migration reference area is constructed; wherein each piece of historical snowfall information includes historical snowfall associated parameter information and historical snowfall amount information corresponding to the historical monitoring period;

[0060] Perform model migration based on model weight mapping on several reference snowfall prediction models to obtain the initial snowfall prediction model of the target monitoring area;

[0061] Obtain snowfall associated parameter information of the target monitoring area during several model optimization periods and the target monitoring period, input the snowfall associated parameter information of the model optimization period into the initial snowfall prediction model, calculate the predicted snowfall of the target monitoring area during the model optimization period, adjust the mapping relationship of the model migration action based on the predicted snowfall value and the actual snowfall value during the model optimization period, optimize the initial snowfall prediction model of the target monitoring area, and obtain the final snowfall prediction model;

[0062] The snowfall associated parameter information is input into the final snowfall prediction model to calculate the predicted snowfall in the target monitoring area during the target monitoring period.

[0063] The specific embodiments of the present invention applied to the device are basically the same as the embodiments of the snowfall calculation method described below, and will not be described in detail here.

[0064] The embodiment of the present invention provides a method for calculating snowfall, referring to Figure 2 , Figure 2 Schematic diagram of the process of an embodiment of a method for calculating snowfall amount according to the present invention.

[0065] In this embodiment, the snowfall calculation method includes the following steps:

[0066] S100: generating a target geographic parameter feature map according to a set of geographic environment parameters of a target monitoring area; wherein the target geographic parameter feature map comprises a first element visualization feature having a first RGB color layer and a second element visualization feature having a second RGB color layer;

[0067] S200: Based on the first RGB color layer and the second RGB color layer of the target geographic parameter characteristic map, performing geographic parameter characteristic map matching in a geographic environment database to determine a plurality of migration reference areas;

[0068] S300: constructing a reference snowfall prediction model for the migration reference area using a plurality of pieces of historical snowfall information stored in a historical snowfall database of the migration reference area; wherein each piece of historical snowfall information includes historical snowfall associated parameter information and historical snowfall information corresponding to a historical monitoring period;

[0069] S400: performing a model migration action based on model weight mapping on a plurality of reference snowfall prediction models to obtain an initial snowfall prediction model for a target monitoring area;

[0070] S500: obtaining snowfall associated parameter information of the target monitoring area in several model optimization periods and the target monitoring period, inputting the snowfall associated parameter information of the model optimization period into the initial snowfall prediction model, calculating the predicted snowfall of the target monitoring area in the model optimization period, adjusting the mapping relationship of the model migration action based on the predicted snowfall value and the actual snowfall value in the model optimization period, optimizing the initial snowfall prediction model of the target monitoring area, and obtaining the final snowfall prediction model;

[0071] S600: Input the snowfall associated parameter information into the final snowfall prediction model to calculate the predicted snowfall in the target monitoring area during the target monitoring period.

[0072] It should be noted that in some scenarios, the snowfall calculation method based on artificial intelligence technology faces some limitations: for example, in remote areas or areas where snowfall-related information collection and snowfall monitoring systems have just been built, due to the very limited initial reference data, it is difficult to meet the artificial intelligence model's demand for a large amount of data, and it is difficult to accurately extract features related to the snowfall formation environment, resulting in insufficient model training and insufficient calculation accuracy. In order to solve the above problems, this embodiment obtains the geographic environment parameter set and snowfall-related parameter information of the target monitoring area, uses the geographic environment parameter set to construct a geographic parameter feature map, and then generates several migration reference areas in the geographic environment database, and constructs an independent reference snowfall prediction model based on the historical snowfall information of each migration reference area, and then performs a model migration action based on model weight mapping on several reference snowfall prediction models, and adjusts the migration strategy based on the model optimization period to achieve model migration and optimization between the migration reference area and the target monitoring area. On the one hand, it ensures the accuracy of snowfall calculation in the monitoring area using the artificial intelligence model in data-scarce scenarios, and on the other hand, it improves the degree of adaptation between the snowfall prediction model and the target monitoring area, avoids the problem of model incompatibility caused by the model migration process, and has good scene adaptability.

[0073] In a preferred embodiment, the first element visualization feature includes the terrain altitude value, and the second element visualization feature includes the vegetation coverage rate; according to the geographical environment parameter set of the target monitoring area, the step of generating the target geographical parameter feature map specifically includes:

[0074] S110: Acquire a set of geographical environment parameters of the target monitoring area; wherein the set of geographical environment parameters includes a terrain altitude value and a vegetation coverage rate at each location in the target monitoring area;

[0075] S120: generating a first RGB color layer according to the terrain altitude value of each position in the target monitoring area and a preset first conversion relationship between the terrain altitude and the first color channel in the RGB color value, and generating a second RGB color layer according to the vegetation coverage rate of each position in the target monitoring area and a preset second conversion relationship between the vegetation coverage rate and the second color channel in the RGB color value;

[0076] S130: Constructing a target geographic parameter feature map of a target monitoring area based on the first RGB color layer and the second RGB color layer.

[0077] On this basis, based on the first RGB color layer and the second RGB color layer of the target geographic parameter characteristic map, geographic parameter characteristic map matching is performed in the geographic environment database to determine several migration reference area steps, specifically including:

[0078] S131: Accessing a geographic environment database of a target reference area; wherein the geographic environment database stores a reference geographic parameter feature map of the target reference area, and the target reference area is configured to be composed of a plurality of reference sub-areas having collected historical snowfall information that meets training requirements;

[0079] S132: Determine a plurality of migration reference areas in the target reference area based on the target geographic parameter feature map and the first RGB color layer and the second RGB color layer of the reference geographic parameter feature map.

[0080] Furthermore, based on the target geographic parameter characteristic map and the first RGB color layer and the second RGB color layer of the reference geographic parameter characteristic map, a plurality of migration reference area steps are determined in the target reference area, specifically including:

[0081] S210: extracting the average RGB color value of the target geographic parameter feature map in the first RGB color layer, the RGB color value and position of each inflection point in the first RGB color layer, and the average RGB color value of the region corresponding to the position of each inflection point in the second RGB color layer, and constructing a target geographic feature set;

[0082] S220: Input the target geographic feature set and the reference geographic parameter feature map into an image region recognition model including a YOLO structure and an image similarity algorithm, and output the top N migration reference regions in the target reference region that are most similar to the target monitoring region.

[0083] In this embodiment, considering that the factors affecting snowfall generally include geographical constant elements (such as terrain altitude and vegetation coverage; the terrain altitude is usually constant and different morphological characteristics of the terrain altitude will affect changes in snowfall formation factors such as air flow lift, water vapor condensation and air flow humidity; the vegetation coverage is usually constant in the short term and different distributions of vegetation coverage will affect changes in snowfall formation factors such as water vapor humidity and air stability) and atmospheric random elements (such as atmospheric vertical temperature, water vapor content, rising airflow and cloud morphology, which directly affect snowfall formation and snowfall amount), when selecting the migration reference area in the target reference area, the color feature matching degree between the target geographic parameter feature map and the reference geographic parameter feature map in the geographical constant elements is mainly based on the image area recognition model including the YOLO structure and the image similarity algorithm to output the top N migration reference areas with the highest similarity.

[0084] It should be noted that the target reference area is configured as an area with a mature snowfall-related information collection and snowfall monitoring system and has collected data that meets the requirements of artificial intelligence model training. When selecting the migration reference area, the terrain altitude value and vegetation coverage rate of each location in the target monitoring area are first converted into corresponding RGB color values, and the first RGB color layer and the second RGB color layer are constructed. The position information and color information of each pixel in the color layer are used to represent the position and geographic constant elements in the actual area. On the one hand, it facilitates data storage and management, and on the other hand, it improves the visualization of geographic constant elements, which can help users manually screen target reference areas and reduce data processing load. Then, the average RGB color value of the first RGB color layer in the geographical constant element (characterizing the average altitude), the inflection point RGB color value (characterizing the altitude of mountains, peaks, valleys, basins and other areas) and the inflection point position (characterizing the position coordinates of mountains, peaks, valleys, basins and other areas), the regional average RGB color value of the inflection point position corresponding to the second RGB color layer (characterizing the vegetation coverage of mountains, peaks, valleys, basins and other areas) and other parameters are used as matching features to perform similarity matching of regional geographical constant elements, and obtain the top N migration reference areas with the highest similarity, so that the obtained migration reference area has the same or similar snowfall geographical environment model as the target monitoring area. In this way, the migration reference area determined based on the same or similar snowfall geographical environment model is used to realize the migration of the snowfall model and improve the accuracy of snowfall calculation.

[0085] In a preferred embodiment, the steps of constructing a reference snowfall amount prediction model for the migration reference area by using several pieces of historical snowfall information stored in the historical snowfall database of the migration reference area specifically include:

[0086] S310: Accessing the historical snowfall database of each migration reference area, extracting historical snowfall associated parameter information and historical snowfall amount information of each historical monitoring period corresponding to each migration reference area;

[0087] S320: Taking the historical snowfall associated parameter information and the historical snowfall information of each historical monitoring period as a set of training data samples, constructing a training sample set for each migration reference area, inputting the training into the recurrent neural network constructed for each migration reference area, and obtaining a reference snowfall prediction model when the training times reach a preset number or the training converges;

[0088] The recurrent neural network includes an input layer, a hidden layer and an output layer. The input layer transmits the training data sample to the hidden layer after receiving it. The hidden layer updates the hidden state according to the historical snowfall associated parameter information and the historical snowfall information at different times in the time series. The output layer is used to obtain the predicted snowfall according to the hidden state.

[0089] When training the recurrent neural network, the network weights are adjusted through the back propagation algorithm to minimize the error between the predicted value and the actual value, and the final reference snowfall prediction model is obtained.

[0090] On this basis, several reference snowfall prediction models are subjected to model migration based on model weight mapping to obtain the initial snowfall prediction model steps for the target monitoring area, including:

[0091] S410: Obtaining the similarity between each migration reference area and the target monitoring area output by the image area recognition model, and converting the similarity ratio into a migration mapping weight of the reference snowfall prediction model corresponding to each migration reference area;

[0092] S420: Based on the migration mapping weight of each reference snowfall prediction model, a conversion relationship from several reference snowfall prediction models to the initial snowfall prediction model is established by summing the snowfall prediction value of each reference snowfall prediction model and the corresponding migration mapping weight, and an initial snowfall prediction model for the target monitoring area is obtained, whose output value is the weighted sum of the snowfall prediction value of each reference snowfall prediction model and the corresponding migration mapping weight.

[0093] In this embodiment, after determining several migration reference areas, the historical snowfall associated parameter information and the historical snowfall information in several pieces of historical snowfall information stored in the historical snowfall database of each migration reference area are used to input the initial neural network model for training, and the association between the snowfall and the snowfall associated parameters in the time series in these several migration reference areas is explored to obtain the reference snowfall prediction model of each migration reference area. It should be noted that the historical snowfall associated parameter information includes the above-mentioned factors affecting the snowfall (i.e., geographical constant factors and atmospheric random factors). The reference snowfall prediction model obtained by training in this way can characterize the relationship between different and random atmospheric environmental characteristics and snowfall on the basis of geographical constant factors similar to the target monitoring area, which provides a basis for model migration. After that, the similarity between each migration reference area and the target monitoring area is used as the output cumulative weight, so as to use the reference snowfall prediction model of each migration reference area to establish the initial snowfall prediction model of the target monitoring area, realize the model migration call from the first N migration reference areas with the highest similarity to the target monitoring area, and improve the accuracy of snowfall calculation.

[0094] In a preferred embodiment, based on the predicted snowfall value and the actual snowfall value during the model optimization period, the mapping relationship of the model migration action is adjusted, the initial snowfall prediction model of the target monitoring area is optimized, and the steps of obtaining the final snowfall prediction model are specifically included:

[0095] S510: Calculate the average snowfall difference between the predicted snowfall value and the actual snowfall value in each model optimization period, and determine whether the average snowfall difference is less than a preset difference;

[0096] S520: If yes, use the current initial snowfall prediction model as the final snowfall prediction model; if no, adjust the mapping relationship of the model migration action, optimize the initial snowfall prediction model of the target monitoring area, and obtain the final snowfall prediction model;

[0097] Among them, the initial snowfall prediction model of the target monitoring area is optimized to obtain the final snowfall prediction model, which specifically includes the following execution process:

[0098] Taking the adjustment ratio of the migration mapping weight of each reference snowfall prediction model to be less than the target ratio as a constraint condition, and taking the average value of the snowfall difference between the predicted snowfall value and the actual snowfall value in each model optimization period to be less than the minimum value of the preset difference as the goal, the migration mapping weight of each reference snowfall prediction model in the model migration process is adjusted to obtain the migration mapping weight of each reference snowfall prediction model after optimization, and the final snowfall prediction model is established.

[0099] In this embodiment, considering that the initial snowfall prediction model is obtained by model migration between several migration reference areas of the same or similar (not exactly the same) geographical constant elements, the differences in geographical constant elements will still affect the role of atmospheric random elements in snowfall calculation, and will still affect the adaptability of the snowfall prediction model to the target monitoring area and the accuracy of snowfall calculation. After obtaining the initial snowfall prediction model, it is also necessary to adjust the migration strategy during the model optimization period. By monitoring the difference between the predicted snowfall value and the actual snowfall value during the model optimization period, the migration mapping weight of each reference snowfall prediction model is adjusted during the model migration action to achieve model migration and optimization between the migration reference area and the target monitoring area. On the one hand, it ensures the accuracy of snowfall calculation in the monitoring area using artificial intelligence models in data-scarce scenarios. On the other hand, it improves the adaptability of the snowfall prediction model to the target monitoring area, avoids the problem of model incompatibility caused by the model migration process, and has good scene adaptability.

[0100] Reference Figure 3 , Figure 3 It is a structural block diagram of an embodiment of a snowfall calculation device of the present invention.

[0101] like Figure 3 As shown, the snowfall calculation device proposed in the embodiment of the present invention includes:

[0102] A generating module 10 is used to generate a target geographic parameter feature map according to a set of geographic environment parameters of a target monitoring area; wherein the target geographic parameter feature map includes a first element visualization feature having a first RGB color layer and a second element visualization feature having a second RGB color layer;

[0103] A determination module 20, configured to perform geographic parameter feature map matching in a geographic environment database based on the first RGB color layer and the second RGB color layer of the target geographic parameter feature map, and determine a plurality of migration reference areas;

[0104] A construction module 30 is used to construct a reference snowfall amount prediction model for the migration reference area by using a plurality of pieces of historical snowfall information stored in a historical snowfall database of the migration reference area; wherein each piece of historical snowfall information includes historical snowfall associated parameter information and historical snowfall amount information corresponding to a historical monitoring period;

[0105] A migration module 40 is used to perform a model migration action based on a model weight mapping on a plurality of reference snowfall prediction models to obtain an initial snowfall prediction model for a target monitoring area;

[0106] The optimization module 50 is used to obtain snowfall associated parameter information of the target monitoring area in several model optimization periods and the target monitoring period, input the snowfall associated parameter information of the model optimization period into the initial snowfall prediction model, calculate the predicted snowfall of the target monitoring area in the model optimization period, adjust the mapping relationship of the model migration action based on the predicted snowfall value and the actual snowfall value in the model optimization period, optimize the initial snowfall prediction model of the target monitoring area, and obtain the final snowfall prediction model;

[0107] The calculation module 60 is used to input the snowfall associated parameter information into the final snowfall prediction model to calculate the predicted snowfall in the target monitoring area during the target monitoring period.

[0108] In this embodiment, a snowfall calculation device is provided. The device obtains the geographic environment parameter set and snowfall associated parameter information of the target monitoring area, constructs a geographic parameter feature map using the geographic environment parameter set, and then generates several migration reference areas in the geographic environment database. An independent reference snowfall prediction model is constructed according to the historical snowfall information of each migration reference area. Then, the model migration action based on model weight mapping is performed on the several reference snowfall prediction models, and the migration strategy is adjusted based on the model optimization period to realize the model migration and optimization between the migration reference area and the target monitoring area. On the one hand, it ensures the accuracy of snowfall calculation in the monitoring area using artificial intelligence models in data-scarce scenarios. On the other hand, it improves the adaptability of the snowfall prediction model to the target monitoring area, avoids the model incompatibility problem caused by the model migration process, and has good scene adaptability.

[0109] Other embodiments or specific implementations of the snowfall calculation device of the present invention can refer to the above-mentioned method embodiments, which will not be described in detail here.

[0110] In addition, the present invention also proposes a snowfall calculation device, which includes: a memory, a processor, and a snowfall calculation program stored in the memory and executable on the processor, and when the snowfall calculation program is executed by the processor, the steps of the snowfall calculation method described above are implemented.

[0111] The specific implementation of the snowfall calculation device of the present application is basically the same as the above-mentioned embodiments of the snowfall calculation method, and will not be repeated here.

[0112] In addition, the present invention also provides a readable storage medium, which includes a computer-readable storage medium on which a snowfall calculation program is stored. The readable storage medium may be Figure 1The memory 1005 in the terminal may also be at least one of a ROM (Read-Only Memory) / RAM (Random Access Memory), a magnetic disk, and an optical disk. The readable storage medium includes a number of instructions for enabling a snowfall calculation device with a processor to execute the snowfall calculation method described in each embodiment of the present invention.

[0113] The specific implementation in the readable storage medium of the present application is basically the same as the above-mentioned embodiments of the snowfall calculation method, and will not be repeated here.

[0114] It is understood that, in the description of this specification, the description with reference to the terms "one embodiment", "another embodiment", "other embodiments", or "first to Nth embodiments" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0115] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or system 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 system. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the existence of other identical elements in the process, method, article or system including the element.

[0116] The serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.

[0117] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. 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, magnetic disk, optical disk) as described above, and includes a number of instructions for a terminal device (which can be a mobile phone, a computer, a server, or a network device, etc.) to execute the methods described in each embodiment of the present invention.

[0118] Finally, it should be noted that the above embodiments are only used to illustrate rather than limit the technical solutions of the present invention. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that the present invention can still be modified or replaced by equivalents. Any modification or partial replacement that does not depart from the spirit and scope of the present invention should be included in the scope of the claims of the present invention.

Claims

1. A snowfall calculation method, characterized in that: The following steps are involved: Generate a target geographic parameter feature map according to a set of geographic environment parameters of a target monitoring area; wherein the target geographic parameter feature map includes a first element visualization feature having a first RGB color layer and a second element visualization feature having a second RGB color layer; Based on the first RGB color layer and the second RGB color layer of the target geographic parameter characteristic map, performing geographic parameter characteristic map matching in a geographic environment database to determine a plurality of migration reference areas; The first element visualization feature includes the terrain altitude value, and the second element visualization feature includes the vegetation coverage rate; based on the first RGB color layer and the second RGB color layer of the target geographic parameter feature map, performing geographic parameter feature map matching in the geographic environment database to determine a number of migration reference area steps specifically includes: accessing the geographic environment database of the target reference area; wherein the geographic environment database stores a reference geographic parameter feature map of the target reference area, and the target reference area is configured to be composed of a number of reference sub-areas that have collected historical snowfall information that meets the training requirements; based on the target geographic parameter feature map and the first RGB color layer and the second RGB color layer of the reference geographic parameter feature map, determining a number of migration reference areas in the target reference area; Based on the target geographic parameter feature map and the first RGB color layer and the second RGB color layer of the reference geographic parameter feature map, a step of determining a plurality of migration reference areas in the target reference area specifically includes: extracting the average RGB color value of the target geographic parameter feature map in the first RGB color layer, the RGB color value and position of each inflection point in the first RGB color layer, and the average RGB color value of the area corresponding to the position of each inflection point in the second RGB color layer, and constructing a target geographic feature set; inputting the target geographic feature set and the reference geographic parameter feature map into an image region recognition model including a YOLO structure and an image similarity algorithm, and outputting the top N migration reference areas in the target reference area that have the highest similarity with the target monitoring area; Using several pieces of historical snowfall information stored in the historical snowfall database of the migration reference area, a reference snowfall amount prediction model for the migration reference area is constructed; wherein each piece of historical snowfall information includes historical snowfall associated parameter information and historical snowfall amount information corresponding to the historical monitoring period; Perform model migration actions based on model weight mapping on several reference snowfall prediction models to obtain the initial snowfall prediction model of the target monitoring area; The steps of performing a model migration action based on model weight mapping on several reference snowfall prediction models to obtain an initial snowfall prediction model for a target monitoring area specifically include: obtaining the similarity between each migration reference area and the target monitoring area output by the above-mentioned image area recognition model, and converting the ratio of the similarities into a migration mapping weight of the reference snowfall prediction model corresponding to each migration reference area; based on the migration mapping weight of each reference snowfall prediction model, using the weighted sum of the snowfall prediction value of each reference snowfall prediction model and the corresponding migration mapping weight, establishing a conversion relationship from several reference snowfall prediction models to the initial snowfall prediction model, and obtaining an initial snowfall prediction model for the target monitoring area whose output value is the weighted sum of the snowfall prediction value of each reference snowfall prediction model and the corresponding migration mapping weight; Obtain snowfall associated parameter information of the target monitoring area during several model optimization periods and the target monitoring period, input the snowfall associated parameter information of the model optimization period into the initial snowfall prediction model, calculate the predicted snowfall of the target monitoring area during the model optimization period, adjust the mapping relationship of the model migration action based on the predicted snowfall value and the actual snowfall value during the model optimization period, optimize the initial snowfall prediction model of the target monitoring area, and obtain the final snowfall prediction model; Based on the predicted snowfall value and the actual snowfall value in the model optimization period, the mapping relationship of the model migration action is adjusted, the initial snowfall prediction model of the target monitoring area is optimized, and the steps of obtaining the final snowfall prediction model specifically include: calculating the average snowfall difference between the predicted snowfall value and the actual snowfall value in each model optimization period, and judging whether the average snowfall difference is less than a preset difference; if so, taking the current initial snowfall prediction model as the final snowfall prediction model; if not, adjusting the mapping relationship of the model migration action, optimizing the initial snowfall prediction model of the target monitoring area, and obtaining the final snowfall prediction model; Among them, the mapping relationship of the model migration action is adjusted, the initial snowfall prediction model of the target monitoring area is optimized, and the final snowfall prediction model is obtained, which specifically includes the following execution process: taking the migration mapping weight adjustment ratio of each reference snowfall prediction model less than the target ratio as a constraint condition, taking the average value of the snowfall difference between the predicted snowfall value and the actual snowfall value in each model optimization period as less than the minimum value of the preset difference as the goal, adjusting the migration mapping weight of each reference snowfall prediction model during the model migration action, obtaining the migration mapping weight of each reference snowfall prediction model after optimization, and establishing the final snowfall prediction model; The snowfall associated parameter information is input into the final snowfall prediction model to calculate the predicted snowfall in the target monitoring area during the target monitoring period.

2. The snowfall calculation method according to claim 1, characterized in that: The first element visualization feature includes the terrain altitude value, and the second element visualization feature includes the vegetation coverage rate; according to the geographic environment parameter set of the target monitoring area, the steps of generating the target geographic parameter feature map specifically include: Acquire a set of geographical environment parameters of the target monitoring area; wherein the set of geographical environment parameters includes the terrain altitude value and vegetation coverage rate of each location in the target monitoring area; Generate a first RGB color layer according to the terrain altitude value of each position in the target monitoring area and the preset first conversion relationship between the terrain altitude and the first color channel in the RGB color value, and generate a second RGB color layer according to the vegetation coverage rate of each position in the target monitoring area and the preset second conversion relationship between the vegetation coverage rate and the second color channel in the RGB color value; A target geographic parameter feature map of the target monitoring area is constructed based on the first RGB color layer and the second RGB color layer.

3. The snowfall calculation method according to claim 1, characterized in that: The steps of constructing a reference snowfall prediction model for the migration reference area by using several pieces of historical snowfall information stored in the historical snowfall database of the migration reference area include: Access the historical snowfall database of each migration reference area, and extract historical snowfall associated parameter information and historical snowfall amount information of each historical monitoring period corresponding to each migration reference area; The historical snowfall correlation parameter information and the historical snowfall information of each historical monitoring period are used as a set of training data samples to construct a training sample set for each migration reference area, and the training is input into the recurrent neural network constructed for each migration reference area to obtain a reference snowfall prediction model when the training times reach a preset number or the training converges; The recurrent neural network includes an input layer, a hidden layer and an output layer. The input layer transmits the training data sample to the hidden layer after receiving it. The hidden layer updates the hidden state according to the historical snowfall associated parameter information and the historical snowfall information at different times in the time series. The output layer is used to obtain the predicted snowfall according to the hidden state. When training the recurrent neural network, the network weights are adjusted through the back propagation algorithm to minimize the error between the predicted value and the actual value, and the final reference snowfall prediction model is obtained.

4. A snowfall calculation device, characterized in that: include: A generating module, used for generating a target geographic parameter characteristic map according to a set of geographic environment parameters of a target monitoring area; wherein the target geographic parameter characteristic map comprises a first element visualization feature having a first RGB color layer and a second element visualization feature having a second RGB color layer; A determination module, configured to perform geographic parameter feature map matching in a geographic environment database based on a first RGB color layer and a second RGB color layer of the target geographic parameter feature map, and determine a plurality of migration reference areas; The first element visualization feature includes the terrain altitude value, and the second element visualization feature includes the vegetation coverage rate; based on the first RGB color layer and the second RGB color layer of the target geographic parameter feature map, performing geographic parameter feature map matching in the geographic environment database to determine a number of migration reference areas, specifically including: accessing the geographic environment database of the target reference area; wherein the geographic environment database stores a reference geographic parameter feature map of the target reference area, and the target reference area is configured to be composed of a number of reference sub-areas that have collected historical snowfall information that meets the training requirements; based on the target geographic parameter feature map and the first RGB color layer and the second RGB color layer of the reference geographic parameter feature map, determining a number of migration reference areas in the target reference area; Based on the target geographic parameter feature map and the first RGB color layer and the second RGB color layer of the reference geographic parameter feature map, a number of migration reference areas are determined in the target reference area, specifically including: extracting the average RGB color value of the target geographic parameter feature map in the first RGB color layer, the RGB color value and position of each inflection point in the first RGB color layer, and the average RGB color value of the area corresponding to the position of each inflection point in the second RGB color layer, and constructing a target geographic feature set; inputting the target geographic feature set and the reference geographic parameter feature map into an image area recognition model including a YOLO structure and an image similarity algorithm, and outputting the top N migration reference areas in the target reference area that have the highest similarity with the target monitoring area; A construction module is used to construct a reference snowfall prediction model for the migration reference area by using a number of historical snowfall information stored in a historical snowfall database of the migration reference area; wherein each piece of historical snowfall information includes historical snowfall associated parameter information and historical snowfall information corresponding to a historical monitoring period; A migration module is used to perform a model migration action based on model weight mapping on several reference snowfall prediction models to obtain an initial snowfall prediction model for a target monitoring area; The model migration action based on model weight mapping is performed on several reference snowfall prediction models to obtain the initial snowfall prediction model of the target monitoring area, including: Obtaining the similarity between each migration reference area and the target monitoring area output by the above-mentioned image area recognition model, and converting the ratio of the similarity into a migration mapping weight of the reference snowfall prediction model corresponding to each migration reference area; Based on the migration mapping weight of each reference snowfall prediction model, a conversion relationship from several reference snowfall prediction models to an initial snowfall prediction model is established by using the weighted sum of the snowfall prediction value of each reference snowfall prediction model and the corresponding migration mapping weight, and an initial snowfall prediction model for the target monitoring area whose output value is the weighted sum of the snowfall prediction value of each reference snowfall prediction model and the corresponding migration mapping weight is obtained; An optimization module is used to obtain snowfall associated parameter information of a target monitoring area during several model optimization periods and a target monitoring period, input the snowfall associated parameter information of the model optimization period into the initial snowfall prediction model, calculate the predicted snowfall of the target monitoring area during the model optimization period, adjust the mapping relationship of the model migration action based on the predicted snowfall value and the actual snowfall value during the model optimization period, optimize the initial snowfall prediction model of the target monitoring area, and obtain a final snowfall prediction model; Based on the predicted snowfall value and the actual snowfall value in the model optimization period, the mapping relationship of the model migration action is adjusted, the initial snowfall prediction model of the target monitoring area is optimized, and the final snowfall prediction model is obtained, which specifically includes: calculating the average snowfall difference between the predicted snowfall value and the actual snowfall value in each model optimization period, and judging whether the average snowfall difference is less than a preset difference; if so, taking the current initial snowfall prediction model as the final snowfall prediction model; if not, adjusting the mapping relationship of the model migration action, optimizing the initial snowfall prediction model of the target monitoring area, and obtaining the final snowfall prediction model; Among them, the mapping relationship of the model migration action is adjusted, the initial snowfall prediction model of the target monitoring area is optimized, and the final snowfall prediction model is obtained, which specifically includes the following execution process: taking the migration mapping weight adjustment ratio of each reference snowfall prediction model less than the target ratio as a constraint condition, taking the average value of the snowfall difference between the predicted snowfall value and the actual snowfall value in each model optimization period as less than the minimum value of the preset difference as the goal, adjusting the migration mapping weight of each reference snowfall prediction model during the model migration action, obtaining the migration mapping weight of each reference snowfall prediction model after optimization, and establishing the final snowfall prediction model; The calculation module is used to input the snowfall associated parameter information into the final snowfall prediction model to calculate the predicted snowfall in the target monitoring area during the target monitoring period.

5. A snowfall calculation device, characterized in that: The snowfall calculation device includes: a memory, a processor, and a snowfall calculation program stored in the memory and executable on the processor. When the snowfall calculation program is executed by the processor, the steps of the snowfall calculation method according to any one of claims 1 to 3 are implemented.

6. A storage medium, characterized in that: The storage medium stores a snowfall calculation program, which, when executed by a processor, implements the steps of the snowfall calculation method according to any one of claims 1 to 3.

Images