Meteorological multi-type structure data storage method and system

Abstract

The invention relates to the technical field of data processing, in particular to a meteorological multi-type structure data storage method and system, and aims to realize unified storage according to a preset storage standard by deeply analyzing to-be-stored target multi-mode meteorological monitoring data and accurately identifying key information by means of meteorological type labels. Therefore, the problem that data attributes are inconsistent is solved, and the flexibility and uniformity of data storage are remarkably improved through a series of technical means such as data splitting, strengthening processing and structured attribute vector recognition. The embodiment of the invention aims to provide a more efficient and convenient solution for integration, management and application of meteorological data so as to promote sustainable development of meteorological research and related applications.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a method and system for storing meteorological multi-type structural data. Background Art

[0002] In the field of meteorological monitoring, with the continuous advancement of technology, meteorological data of various structural types can be obtained. However, these data are often difficult to store and manage uniformly due to their inconsistent attributes, which brings great inconvenience to subsequent data analysis and application. Traditional data storage methods often seem to be unable to cope with such multi-modal and multi-attribute meteorological data and cannot meet the high requirements of modern meteorological research for data flexibility and uniformity. Summary of the invention

[0003] In order to improve the above problems, the present application provides a method and system for storing meteorological multi-type structured data.

[0004] The present application provides a method for storing meteorological multi-type structure data, which is applied to a data storage system. The method includes:

[0005] Acquire target multi-mode meteorological monitoring data to be stored, and a meteorological type label corresponding to the target multi-mode meteorological monitoring data, wherein the meteorological type label is used to describe a structured meteorological attribute feature set of the target multi-mode meteorological monitoring data;

[0006] Performing data splitting on the target multi-mode meteorological monitoring data to obtain a meteorological monitoring data set of the target multi-mode meteorological monitoring data, and performing enhancement processing on the meteorological monitoring data set to obtain an enhanced meteorological monitoring data set;

[0007] According to the meteorological type label, the enhanced meteorological monitoring data set is subjected to X rounds of structured attribute vector recognition through a structured storage analysis algorithm to obtain a target structured meteorological attribute feature set of the target multi-mode meteorological monitoring data;

[0008] Among them, during the Y-round structured attribute vector identification in the X-round structured attribute vector identification, the transition meteorological state one-hot coding of the structured storage analysis algorithm is mapped to the meteorological state conduction network obtained in the target multi-mode meteorological monitoring data, and the meteorological state one-hot coding of at least Z network units in the meteorological state conduction network is vector-integrated with the meteorological state one-hot coding of the corresponding neighboring network units, as the object to be processed of the next algorithm branch of the structured storage analysis algorithm, the Z network units include the network units corresponding to the modal identification when the target multi-mode meteorological monitoring data is split, and both X and Z are positive integers, and Y is a positive integer not greater than X.

[0009] Preferably, the method of performing X rounds of structured attribute vector recognition on the enhanced meteorological monitoring data set through a structured storage analysis algorithm based on the meteorological type label to obtain a target structured meteorological attribute feature set of the target multi-mode meteorological monitoring data includes:

[0010] In the u-th round of structured attribute vector recognition, a first transitional meteorological state one-hot encoding generated by a v-th algorithm branch of the structured storage analysis algorithm is obtained, wherein the first transitional meteorological state one-hot encoding is obtained based on the enhanced meteorological monitoring data set and the meteorological type label, wherein v is a positive integer, and u is a positive integer not greater than X;

[0011] When the u-th round of structured attribute vector identification belongs to the Y-th round of structured attribute vector identification, the first transitional meteorological state one-hot encoding is mapped to the target multi-mode meteorological monitoring data to obtain the meteorological state transmission network, and the meteorological state one-hot encodings of the Z network units in the meteorological state transmission network are weighted with the meteorological state one-hot encodings of the corresponding neighboring network units to obtain a first meteorological state weighted one-hot encoding;

[0012] According to the weighted one-hot encoding of the first meteorological state, determining the one-hot encoding to be processed of the v+1th algorithm branch of the structured storage analysis algorithm, and continuing the recognition processing to obtain the uth round of intermediate structured meteorological attribute feature set of the target multi-mode meteorological monitoring data;

[0013] According to the meteorological type label, the structured attribute vector of the u+1th round is identified on the uth round intermediate structured meteorological attribute feature set through the structured storage analysis algorithm to obtain the target structured meteorological attribute feature set.

[0014] Preferably, the obtaining of the first transition meteorological state one-hot encoding generated by the vth algorithm branch of the structured storage analysis algorithm comprises:

[0015] Obtaining the u-1th round of intermediate structured meteorological attribute feature set of the target multi-mode meteorological monitoring data;

[0016] According to the meteorological type label and the u-1th round intermediate structured meteorological attribute feature set, the first transition meteorological state one-hot encoding generated by the vth algorithm branch is obtained.

[0017] Preferably, the step of obtaining the u-1th round of intermediate structured meteorological attribute feature set of the target multi-mode meteorological monitoring data includes:

[0018] In response to u being equal to 1, determining the enhanced meteorological monitoring data set as the u-1th round intermediate structured meteorological attribute feature set;

[0019] In response to u being greater than 1, the structured attribute vector of the u-1th round is identified on the enhanced meteorological monitoring data set through the structured storage analysis algorithm according to the meteorological type label to obtain the u-1th round intermediate structured meteorological attribute feature set.

[0020] Preferably, obtaining the first transitional meteorological state one-hot encoding generated by the vth algorithm branch based on the meteorological type label and the u-1th round intermediate structured meteorological attribute feature set includes:

[0021] When the vth algorithm branch is the first algorithm branch of the structured storage analysis algorithm, the meteorological type label and the u-1th round intermediate structured meteorological attribute feature set are processed by the vth algorithm branch to obtain the first transition meteorological state one-hot encoding;

[0022] When the vth algorithm branch is the transition algorithm branch of the structured storage analysis algorithm, the second transition meteorological state one-hot code generated by the v-1th algorithm branch of the structured storage analysis algorithm is obtained, and the one-hot code to be processed of the vth algorithm branch is determined based on the second transition meteorological state one-hot code, and the one-hot code to be processed of the vth algorithm branch is processed by the vth algorithm branch to obtain the one-hot code of the first transition meteorological state.

[0023] Preferably, the step of processing the meteorological type label and the u-1th round intermediate structured meteorological attribute feature set through the vth algorithm branch to obtain the first transition meteorological state one-hot encoding comprises:

[0024] Mining the meteorological type one-hot encoding of the meteorological type label;

[0025] Mining the linear meteorological state one-hot encoding of the u-1th round intermediate structured meteorological attribute feature set;

[0026] The meteorological type one-hot encoding of the meteorological type label and the linear meteorological state one-hot encoding of the u-1th round intermediate structured meteorological attribute feature set are processed through the vth algorithm branch to obtain the first transition meteorological state one-hot encoding.

[0027] Preferably, determining the one-hot encoding to be processed of the v-th algorithm branch according to the one-hot encoding of the second transitional meteorological state includes:

[0028] Determine the second transitional meteorological state as the one-hot code to be processed in the vth algorithm branch;

[0029] Alternatively, the second transition meteorological state one-hot code is mapped to the target multi-mode meteorological monitoring data to obtain the meteorological state transmission network, and the meteorological state one-hot codes of the Z network units in the meteorological state transmission network are weighted with the meteorological state one-hot codes of the corresponding neighboring network units to obtain the second meteorological state weighted one-hot code, and the one-hot code to be processed of the vth algorithm branch is determined based on the second meteorological state weighted one-hot code.

[0030] Preferably, the meteorological state one-hot coding of the Z network units in the meteorological state transmission network is weighted with the meteorological state one-hot coding of the corresponding neighboring network units to obtain the weighted one-hot coding of the target meteorological state, including:

[0031] For each of the Z network units, weighting the unique hot coding of the meteorological state of the network unit and the unique hot coding of the meteorological state of the neighboring network units of the network unit as the current unique hot coding of the meteorological state of the network unit, and obtaining a meteorological state conduction network after vector integration;

[0032] According to the meteorological state conduction network after the vector integration, the target meteorological state weighted one-hot encoding is obtained, and the target meteorological state weighted one-hot encoding is the first meteorological state weighted one-hot encoding or the second meteorological state weighted one-hot encoding.

[0033] Preferably, obtaining the weighted one-hot encoding of the target meteorological state based on the meteorological state conduction network after the vector integration includes: performing feature encoding on the meteorological state conduction network after the vector integration based on the meteorological monitoring data set to obtain the weighted one-hot encoding of the target meteorological state.

[0034] Preferably, the feature encoding of the meteorological state conduction network after the vector integration is performed based on the meteorological monitoring data set to obtain the weighted one-hot encoding of the target meteorological state includes:

[0035] According to the feature encoding rule between the meteorological monitoring data set and the target multi-mode meteorological monitoring data, the meteorological state conduction network after the vector integration is mapped to the meteorological monitoring data set to obtain a third meteorological state weighted one-hot encoding;

[0036] The weighted one-hot encoding of the target meteorological state is obtained according to the weighted one-hot encoding of the third meteorological state.

[0037] Preferably, the step of obtaining the weighted one-hot encoding of the target meteorological state according to the weighted one-hot encoding of the third meteorological state comprises: adjusting the dimension of the weighted one-hot encoding of the third meteorological state according to the dimension information of the one-hot encoding of the target transition meteorological state to obtain the weighted one-hot encoding of the target meteorological state;

[0038] Among them, the dimension of the target meteorological state weighted unique hot encoding is consistent with the dimension of the target transition meteorological state unique hot encoding. When the target transition meteorological state unique hot encoding is the first transition meteorological state unique hot encoding, the target meteorological state weighted unique hot encoding is the first meteorological state weighted unique hot encoding. When the target transition meteorological state unique hot encoding is the second transition meteorological state unique hot encoding, the target meteorological state weighted unique hot encoding is the second meteorological state weighted unique hot encoding.

[0039] Preferably, determining the to-be-processed one-hot encoding of the target algorithm branch according to the weighted one-hot encoding of the target meteorological state includes:

[0040] The weighted one-hot code of the target meteorological state is used as the one-hot code to be processed of the target algorithm branch; or the one-hot code of the target transition meteorological state and the weighted one-hot code of the target meteorological state are weighted to obtain the one-hot code to be processed of the target algorithm branch;

[0041] Among them, when the target algorithm branch is the v+1th algorithm branch, the target meteorological state weighted unique hot encoding is the first meteorological state weighted unique hot encoding, and the target transition meteorological state unique hot encoding is the first transition meteorological state unique hot encoding. When the target algorithm branch is the vth algorithm branch, the target meteorological state weighted unique hot encoding is the second meteorological state weighted unique hot encoding, and the target transition meteorological state unique hot encoding is the second transition meteorological state unique hot encoding.

[0042] Preferably, the weighting of the target transition meteorological state one-hot encoding and the target meteorological state weighted one-hot encoding to obtain the one-hot encoding to be processed of the target algorithm branch includes:

[0043] Determine the confidence coefficients corresponding to the one-hot encoding of the target transition meteorological state and the weighted one-hot encoding of the target meteorological state respectively;

[0044] According to the confidence coefficients respectively corresponding to the target transition meteorological state one-hot encoding and the target meteorological state weighted one-hot encoding, the target transition meteorological state one-hot encoding and the target meteorological state weighted one-hot encoding are weighted to obtain the object to be processed of the target algorithm branch.

[0045] Preferably, the determining of the confidence coefficients corresponding to the target transition meteorological state one-hot encoding and the target meteorological state weighted one-hot encoding respectively includes:

[0046] Through the feature focusing strategy, the target transition meteorological state one-hot encoding and the target meteorological state weighted one-hot encoding are processed to obtain the confidence coefficients corresponding to the target transition meteorological state one-hot encoding and the target meteorological state weighted one-hot encoding respectively.

[0047] Preferably, the method further comprises:

[0048] When the u-th round of structured attribute vector identification does not belong to the Y-th round of structured attribute vector identification, the first transitional meteorological state one-hot encoding is determined as the one-hot encoding to be processed of the v+1-th algorithm branch of the structured storage analysis algorithm.

[0049] An embodiment of the present application provides a data storage system, including at least one processor and a memory; the memory stores computer-executable instructions; the at least one processor executes the computer-executable instructions stored in the memory, so that the at least one processor executes the above method.

[0050] An embodiment of the present application provides a computer-readable storage medium on which a computer program is stored. The computer program implements the above method when running.

[0051] In an embodiment of the present application, by conducting an in-depth analysis of the target multi-mode meteorological monitoring data to be stored and accurately identifying key information with the help of meteorological type labels, unified storage according to preset storage standards is achieved. In this way, not only the problem of inconsistent data attributes is solved, but also a series of technical means, such as data splitting, enhanced processing, and structured attribute vector recognition, significantly improve the flexibility and uniformity of data storage. The embodiment of the present application aims to provide a more efficient and convenient solution for the integration, management and application of meteorological data, so as to promote the sustainable development of meteorological research and related applications. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 A flowchart of a method for storing meteorological multi-type structural data provided in an embodiment of the present application.

[0053] Figure 2 A structural schematic diagram of a data storage system provided in an embodiment of the present application. DETAILED DESCRIPTION

[0054] In order to better understand the above technical scheme, the technical scheme of the present application is described in detail below through the accompanying drawings and specific embodiments. It should be understood that the embodiments of the present application and the specific features in the embodiments are detailed descriptions of the technical scheme of the present application, rather than limitations on the technical scheme of the present application. In the absence of conflict, the embodiments of the present application and the technical features in the embodiments can be combined with each other.

[0055] Figure 1 A meteorological multi-type structure data storage method is shown, which is applied to a data storage system. The method includes the following steps 110 to 130.

[0056] In an exemplary meteorological monitoring application scenario, the data storage system is responsible for acquiring, processing and structured storage of multi-mode meteorological monitoring data.

[0057] Step 110: Acquire target multi-mode meteorological monitoring data to be stored, and meteorological type labels corresponding to the target multi-mode meteorological monitoring data, wherein the meteorological type labels are used to describe a structured meteorological attribute feature set of the target multi-mode meteorological monitoring data.

[0058] First, in step 110, the data storage system obtains the target multi-mode meteorological monitoring data to be stored from various meteorological monitoring devices. The target multi-mode meteorological monitoring data may include meteorological information of multiple modes such as temperature, humidity, wind speed, wind direction, and air pressure. At the same time, the data storage system also obtains the meteorological type label corresponding to the target multi-mode meteorological monitoring data. The meteorological type label is a description of the structured meteorological attribute feature set of the target multi-mode meteorological monitoring data, such as "high temperature and high humidity", "low wind speed", etc., which is helpful for subsequent data processing and analysis.

[0059] In detail, the target multi-mode meteorological monitoring data to be stored refers to the raw data collected from various meteorological monitoring devices, which contain meteorological information of multiple modes. For example, temperature data reflects the heat of the environment, humidity data reflects the moisture content in the air, wind speed and wind direction data reveal the strength and direction of the wind, and air pressure data shows the change in atmospheric pressure. These data modes are independent and interrelated, and together constitute complex and changeable meteorological conditions. Since these data come from different monitoring devices and sensors, they are called "multi-mode" data, that is, a collection of data of multiple modes and types. These data are crucial for applications such as meteorological analysis, forecasting, and disaster warning.

[0060] Meteorological type labels are keywords or phrases that classify and describe multi-modal meteorological monitoring data. They are defined based on specific features and attributes of the data and are used to summarize and identify the overall meteorological conditions of the data. For example, a "high temperature and high humidity" label may indicate that both temperature and humidity have reached high levels during a certain period of time, and a "low wind speed" label indicates meteorological conditions with low wind speeds. These labels not only help to quickly understand the core features of the data, but also provide an important reference for subsequent data processing and analysis. With meteorological type labels, researchers can more efficiently screen, compare, and analyze different types of meteorological data.

[0061] The structured meteorological attribute feature set is a set of key features extracted after in-depth analysis and processing of multi-mode meteorological monitoring data. These features are presented in a structured form, that is, they are organized into a set of attributes with clear meanings and relationships. For example, temperature, humidity, wind speed, etc. can all be part of the structured meteorological attributes. These attributes not only reflect the core information of meteorological data, but also facilitate quantitative analysis and comparison. By extracting and integrating these structured features, researchers can have a deeper understanding of the inherent laws and changing trends of meteorological data, and provide strong data support for applications such as meteorological forecasting and disaster warning.

[0062] It can be understood that step 110 is the primary link for the data storage system to process meteorological data, and its core task is to obtain the target multi-mode meteorological monitoring data to be stored, and at the same time obtain the meteorological type labels corresponding to these data. In this step, the data storage system first receives or regularly collects multi-mode meteorological monitoring data from meteorological monitoring equipment all over the country through various interfaces and protocols. These data cover a variety of key meteorological parameters such as temperature, humidity, wind speed, wind direction, and air pressure, and are the basis for subsequent analysis and prediction. At the same time, the data storage system will also obtain meteorological type labels corresponding to these data. These labels are usually generated by meteorological experts or automated algorithms based on the characteristics of the data, and are used to describe the overall meteorological conditions of the data in a concise and clear manner. For example, if the data of a certain period of time shows the characteristics of high temperature and high humidity, then the data of this period of time may be marked as a "high temperature and high humidity" type. The obtained meteorological type labels not only help researchers quickly understand the overall characteristics of the data, but also provide important classification basis for subsequent data processing and analysis. Through these labels, the data storage system can organize and manage massive meteorological data more efficiently, and provide accurate and targeted data support for subsequent meteorological forecasts, disaster warnings and other applications. Therefore, step 110 is a key link to ensure that the data storage system can comprehensively and accurately acquire and understand the multi-mode meteorological monitoring data, laying a solid foundation for subsequent data processing and analysis.

[0063] Step 120: split the target multi-mode meteorological monitoring data to obtain a meteorological monitoring data set of the target multi-mode meteorological monitoring data, and perform enhancement processing on the meteorological monitoring data set to obtain an enhanced meteorological monitoring data set.

[0064] Next, in step 120, the data storage system splits the acquired target multimodal meteorological monitoring data. The purpose of this step is to decompose the complex multimodal data into meteorological monitoring data sets that are easier to process and analyze. For example, wind speed data, wind direction data, etc. may be stored and processed separately. After the data splitting is completed, the data storage system will perform enhanced processing on these meteorological monitoring data sets, such as data cleaning, outlier detection and correction, etc., to improve the quality and accuracy of the data, thereby obtaining an enhanced meteorological monitoring data set.

[0065] In detail, meteorological monitoring datasets refer to data sets collected and collated from various meteorological monitoring devices. These data cover a variety of meteorological parameters, such as temperature, humidity, wind speed, wind direction, air pressure, etc. They are systematically organized and stored for subsequent data analysis and meteorological research. Meteorological monitoring datasets are an important foundation for meteorological research, weather forecasting, climate model construction and other fields. By analyzing these datasets, scientists can better understand the movement of the atmosphere, predict future weather conditions, and even assess the impact of climate change.

[0066] In the context of data processing, enhancement processing refers to a series of optimization and correction operations on raw data to improve the quality and usability of the data. This process can include data cleaning, which is to remove invalid or erroneous data points; outlier detection and correction, to ensure that all values ​​in the data set are within a reasonable range; data smoothing to reduce the impact of noise and fluctuations on the analysis results; and possible data transformation or standardization to make the data more in line with specific analysis requirements. Enhancement processing is an important step before data analysis, which ensures the accuracy and effectiveness of subsequent analysis.

[0067] Enhanced meteorological monitoring datasets refer to meteorological data sets that have been enhanced. These data have been cleaned, verified, and possibly transformed to ensure their quality, accuracy, and consistency, making them more suitable for advanced data analysis and model building. Enhanced meteorological monitoring datasets are key inputs for applications such as meteorological forecasting, climate research, and disaster warning, as they provide a reliable and optimized data foundation that can support more accurate scientific analysis and decision-making.

[0068] In step 120, the core task of the data storage system is to process and optimize the raw multi-modal meteorological data collected from the meteorological monitoring equipment. First, the system will perform data splitting on the "target multi-modal meteorological monitoring data". Since these data usually contain multiple meteorological parameters, such as temperature, humidity, wind speed, etc., they need to be decomposed in an appropriate way so that they can be processed and analyzed separately. The data splitting process is carried out according to the modal characteristics of the data to ensure that the data of each mode can be extracted and processed separately. After completing the data splitting, the system will obtain multiple single-modal "meteorological monitoring data sets". These data sets correspond to different meteorological parameters and have been preliminarily sorted and formatted, laying the foundation for subsequent data processing. Next, the system will "enhance these meteorological monitoring data sets". This process aims to improve the quality and accuracy of the data and ensure that the data set does not contain erroneous, abnormal or irrelevant information. Enhancement processing can include data cleaning to remove duplicate, invalid or obviously erroneous data points; outlier detection to identify and correct values ​​that are out of the normal range; and data standardization or normalization so that data of different modes can be analyzed on the same scale. After the enhanced processing, the system will obtain "enhanced meteorological monitoring data sets". These data sets are not only cleaner and more accurate, but also optimized and more suitable for complex data analysis and model building. The output of this step provides a high-quality data foundation for subsequent structured storage analysis algorithms, which is a key link in ensuring the effectiveness and accuracy of the entire data storage and processing process.

[0069] Step 130: Based on the meteorological type label, perform X rounds of structured attribute vector recognition on the enhanced meteorological monitoring data set through a structured storage analysis algorithm to obtain a target structured meteorological attribute feature set of the target multi-mode meteorological monitoring data.

[0070] Among them, during the Y-round structured attribute vector identification in the X-round structured attribute vector identification, the transition meteorological state one-hot coding of the structured storage analysis algorithm is mapped to the meteorological state conduction network obtained in the target multi-mode meteorological monitoring data, and the meteorological state one-hot coding of at least Z network units in the meteorological state conduction network is vector-integrated with the meteorological state one-hot coding of the corresponding neighboring network units, as the object to be processed of the next algorithm branch of the structured storage analysis algorithm, the Z network units include the network units corresponding to the modal identification when the target multi-mode meteorological monitoring data is split, and both X and Z are positive integers, and Y is a positive integer not greater than X.

[0071] Entering step 130, the data storage system will use the structured storage analysis algorithm to perform multiple rounds of structured attribute vector recognition on the enhanced meteorological monitoring data set based on the previously acquired meteorological type labels. In this process, the structured storage analysis algorithm will identify and extract key structured features in the data. It is particularly worth mentioning that in a certain round of structured attribute vector recognition, the structured storage analysis algorithm will map the unique hot encoding of the transitional meteorological state to the meteorological state conduction network extracted from the target multi-mode meteorological monitoring data. The meteorological state conduction network (which can be understood as a mapping spectrum) can reflect the mutual influence and conduction relationship between different meteorological states.

[0072] Subsequently, the structured storage analysis algorithm selects at least Z network units in the network, which correspond to different modes when the original multi-mode meteorological monitoring data is split. For these selected network units, the structured storage analysis algorithm integrates the unique hot encoding of their meteorological state with the unique hot encoding of the meteorological state of the adjacent network units. This integration takes into account the spatial relationship between meteorological states and helps to improve the accuracy of data analysis. The integrated vector then becomes the object to be processed by the next algorithm branch of the structured storage analysis algorithm.

[0073] The whole process will be carried out for X rounds to ensure comprehensive and in-depth analysis of the data. In this way, the data storage system can not only effectively store a large amount of multi-mode meteorological monitoring data, but also perform efficient structured processing and analysis on these data, providing accurate data support for subsequent meteorological forecasts, disaster warnings and other applications.

[0074] In detail, structured attribute vector recognition is a data processing technique used to extract key structured features from complex data sets. In the processing of meteorological monitoring data, it specifically refers to the use of algorithms to identify and analyze meteorological attributes in data, and convert these attributes into vector form to facilitate subsequent data analysis, comparison and storage. This process helps to reveal hidden patterns and associations in the data, providing support for meteorological forecasting and decision-making.

[0075] The target structured meteorological attribute feature set refers to the set of key meteorological features extracted from multi-mode meteorological monitoring data through structured attribute vector recognition technology. These features are structured and represented in the form of vectors, which can fully and accurately reflect the key information of the original data and provide strong data support for applications such as meteorological analysis, prediction and decision-making.

[0076] One-hot encoding of transitional weather states is a data representation method used to convert weather states into a computer-processable format. During structured storage analysis, each unique weather state is assigned a unique binary vector in which only one element is 1 and the rest are 0. This encoding method helps the algorithm accurately identify different weather states when processing data, thereby improving the accuracy of the analysis.

[0077] The meteorological state transmission network is a type of mapping graph that is used to represent the mutual influence and transmission relationship between different meteorological states. In this network, nodes represent different meteorological states, and edges represent the transmission paths between states. By analyzing this network, we can gain a deep understanding of the dynamic changes and interaction mechanisms within the meteorological system, and provide a scientific basis for applications such as meteorological forecasting and disaster warning.

[0078] In the meteorological state transmission network, network units refer to the nodes or components that make up the network. Each network unit represents a specific meteorological state or attribute, and is connected to other units through edges to form a complex network structure. By analyzing the connection relationship and attribute characteristics between network units, the internal structure and operation rules of the meteorological system can be revealed.

[0079] Modal identification is a label or tag used to distinguish and identify different modal information in multimodal meteorological monitoring data. In multimodal data, each mode (such as temperature, humidity, wind speed, etc.) has unique data characteristics and representation methods. Modal identification helps the algorithm accurately identify and process these different modal information when processing data, thereby improving the efficiency and accuracy of data processing.

[0080] It can be understood that step 130 is the core link of the data storage system, which aims to deeply mine the key information in the enhanced meteorological monitoring data set through the structured storage analysis algorithm. First, the system performs multiple rounds of structured attribute vector recognition on the enhanced meteorological monitoring data set based on the meteorological type label obtained previously. In this process, the algorithm will carefully analyze each meteorological attribute in the data set and convert it into a vector form with clear meaning. These vectors not only contain the key information of the original data, but also facilitate subsequent data processing and analysis. It is worth noting that in the X rounds of structured attribute vector recognition, one or more rounds will involve the unique hot encoding of the transitional meteorological state. At this stage, the structured storage analysis algorithm will represent the identified meteorological state in the form of unique hot encoding. This encoding method ensures the uniqueness and accuracy of each meteorological state, laying a solid foundation for subsequent data analysis. Next, the algorithm will map the unique hot encoding of these transitional meteorological states to the meteorological state conduction network. This network is a complex mapping map that can clearly show the mutual influence and conduction relationship between different meteorological states. Through network analysis, the internal structure and dynamic changes of the meteorological system can be more deeply understood. During the mapping process, the algorithm will select at least Z network units in the network for further analysis. These network units correspond to different modal identifiers when the original multi-mode meteorological monitoring data is split. The algorithm will vector-integrate the meteorological state unique-hot encodings of these selected network units with the meteorological state unique-hot encodings of their adjacent network units. This integration method comprehensively considers the spatial relationship and mutual influence between meteorological states, thereby improving the accuracy and comprehensiveness of data analysis. Finally, the data after vector integration will become the object to be processed by the next algorithm branch of the structured storage analysis algorithm. Through continuous iteration and optimization, the algorithm can gradually reveal the deep information and laws hidden in the multi-mode meteorological monitoring data. The entire X-round process of structured attribute vector recognition ensures a comprehensive and in-depth analysis of the data, providing strong data support for subsequent applications such as meteorological forecasting and disaster warning.

[0081] The embodiment of the present application achieves accurate identification and efficient storage of key information by conducting in-depth analysis of the target multi-mode meteorological monitoring data to be stored and combining it with the corresponding meteorological type label. This method not only solves the problem of inconsistent attributes of meteorological data of various structural types, but also significantly improves the flexibility and uniformity of data storage through data splitting, enhanced processing and multiple rounds of structured attribute vector recognition. In particular, in the process of structured attribute vector recognition, by cleverly mapping the unique hot coding of the transitional meteorological state to the meteorological state conduction network, and vectorizing the unique hot coding of the meteorological state of the network unit, the embodiment of the present application further enhances the inherent correlation analysis and mining capabilities of the data.

[0082] Compared with the prior art, the embodiment of the present application can not only store various types of meteorological data in a single data file according to a unified standard, but also support a series of data-driven functions such as arbitrary attribute extension, parallel compression and decompression, encryption and decryption, thereby greatly improving the comprehensive efficiency of data storage and processing. In this way, it not only brings revolutionary convenience to the integration and management of meteorological data, but also lays a solid foundation for subsequent data analysis and application, and is expected to play an important role in meteorological forecasting, disaster prevention and control and other fields.

[0083] In some optional embodiments, based on the meteorological type label, the enhanced meteorological monitoring data set is subjected to X rounds of structured attribute vector recognition through a structured storage analysis algorithm to obtain a target structured meteorological attribute feature set of the target multimodal meteorological monitoring data, including: in the u-th round of structured attribute vector recognition, obtaining the first transition meteorological state unique hot code generated by the v-th algorithm branch of the structured storage analysis algorithm, the first transition meteorological state unique hot code is obtained based on the enhanced meteorological monitoring data set and the meteorological type label, v is a positive integer, and u is a positive integer not greater than X; when the u-th round of structured attribute vector recognition belongs to the Y-th round of structured attribute vector recognition, mapping the first transition meteorological state unique hot code to the The meteorological state transmission network is obtained from the target multi-mode meteorological monitoring data, and the meteorological state unique-hot codes of the Z network units in the meteorological state transmission network are weighted with the meteorological state unique-hot codes of the corresponding neighboring network units to obtain a first meteorological state weighted unique-hot code; based on the first meteorological state weighted unique-hot code, the to-be-processed unique-hot code of the v+1th algorithm branch of the structured storage analysis algorithm is determined, and recognition processing is continued to obtain the u-th round intermediate structured meteorological attribute feature set of the target multi-mode meteorological monitoring data; based on the meteorological type label, the u+1th round structured attribute vector recognition is performed on the u-th round intermediate structured meteorological attribute feature set through the structured storage analysis algorithm to obtain the target structured meteorological attribute feature set.

[0084] Based on the above embodiments, the data storage system will perform a series of complex operations to process and analyze multi-mode meteorological monitoring data. One of the key links in this process is to perform multiple rounds of structured attribute vector recognition on the enhanced meteorological monitoring data set through a structured storage analysis algorithm.

[0085] For example, when the data storage system reaches the uth round of structured attribute vector recognition, it will first obtain the first transition weather state one-hot encoding generated by the vth algorithm branch of the structured storage analysis algorithm. This one-hot encoding is based on the enhanced weather monitoring data set and the corresponding weather type label. The "one-hot encoding" here is a method of converting categorical variables into a format that can be better processed by machine learning algorithms; in this scenario, it represents a specific weather state.

[0086] If the u-th round structured attribute vector identification happens to belong to one of the Y rounds that need special treatment, the data storage system will perform an additional step. It will map the first transition meteorological state one-hot encoding to the meteorological state transmission network extracted from the target multi-mode meteorological monitoring data. This network reveals the connection and influence between different meteorological states.

[0087] Next, the system selects Z network units in the meteorological state transmission network, which correspond to the modal identification when the data is split. Then, the system weights the meteorological state unique hot encodings of these selected network units and the meteorological state unique hot encodings of their adjacent network units to obtain the first meteorological state weighted unique hot encoding. Weighted processing is a mathematical method that considers the importance of different data points, and is used here to strengthen the impact of key meteorological states.

[0088] With this weighted one-hot encoding, the data storage system can determine the one-hot encoding to be processed for the v+1th algorithm branch of the structured storage analysis algorithm. The system then continues the recognition process to generate the uth round of intermediate structured meteorological attribute feature sets for the target multimodal meteorological monitoring data. This feature set captures the key attributes of the data in this round of processing.

[0089] Finally, based on the meteorological type label, the data storage system will use the structured storage analysis algorithm to perform the next round of structured attribute vector recognition on this intermediate feature set, that is, the u+1th round. Through such rounds of processing, the system can eventually extract the target structured meteorological attribute feature set, which is a data set that comprehensively and accurately describes the core attributes of multi-mode meteorological monitoring data.

[0090] Through the detailed operation of the above-mentioned embodiment, the data storage system not only realizes the efficient processing of multi-modal and multi-attribute meteorological data, but also deeply mines the key information in the data through multiple rounds of recognition of structured storage analysis algorithms. This processing method significantly improves the flexibility of data storage and the uniformity of data, so that different types of meteorological data can be integrated and stored according to unified standards. In addition, through the application of weighted processing and meteorological state transmission network, the system further enhances the depth and accuracy of data analysis, and provides more reliable data support for subsequent meteorological forecasts, disaster warnings and other applications. In general, the embodiment of the data storage system demonstrates an advanced and efficient multi-mode meteorological monitoring data processing method, and the beneficial effects it brings are reflected in the accuracy of data processing, the uniformity of storage and the extensiveness of subsequent applications.

[0091] In order to understand the above technical solution more intuitively, it can be explained in detail through a specific numerical feature vector example.

[0092] For example, there is an enhanced meteorological monitoring dataset, which contains various meteorological parameters such as temperature, humidity, wind speed, etc. These data have been preprocessed and enhanced for subsequent structured attribute vector recognition.

[0093] Example:

[0094] Weather type labels: There are three weather type labels, namely "sunny", "rainy" and "snowy", which can be represented by one-hot encoding as:

[0095] Sunny: [1, 0, 0];

[0096] Rainy day: [0, 1, 0];

[0097] Snowy day: [0, 0, 1].

[0098] Structured storage analysis algorithm: perform 3 rounds (X=3) of structured attribute vector recognition.

[0099] First round of recognition (u=1):

[0100] In the first round, the data of "sunny day" is processed, and the unique hot encoding of its weather type label is [1, 0, 0].

[0101] The first algorithm branch (v=1) of the structured storage analysis algorithm generates a first transition weather state unique hot encoding, such as [0.8, 0.1, 0.1], indicating that the current data is more inclined to "sunny".

[0102] Because the first round does not belong to the special Y round, it directly proceeds to the next round of identification.

[0103] Second round of recognition (u=2, this round is one of Y rounds):

[0104] In this round, we continue to process the "clear sky" data, but now we consider the meteorological state transmission network.

[0105] There are 5 network units (Z=5) in the network, and each unit represents a different meteorological observation point.

[0106] For each network unit, there is a one-hot encoding of its weather state, such as [0.9, 0.05, 0.05] for unit 1, [0.85, 0.1, 0.05] for unit 2, and so on.

[0107] These one-hot encodings are weighted with the one-hot encodings of their neighboring units. For example, the weighted one-hot encoding of unit 1 may be the average of its own encoding and the encodings of neighboring units (such as unit 2 and unit 3).

[0108] After weighting, a weighted one-hot encoding of the first meteorological state is obtained, such as [0.87, 0.08, 0.05].

[0109] This weighted one-hot encoding becomes the input to the next algorithm branch (v+1), and the recognition process continues.

[0110] The third round of recognition (u=3):

[0111] In this round, the weighted one-hot encoding obtained in the previous round is used as input.

[0112] The structured storage analysis algorithm continues to process this weighted one-hot encoding and finally outputs a target structured meteorological attribute feature set.

[0113] This feature set may be a more detailed description of the meteorological state, such as a vector containing multiple attributes such as temperature, humidity, wind speed, etc.

[0114] Through this example, we can see that each round of structured attribute vector recognition is gradually refining and strengthening the characteristics of meteorological data. In particular, in round Y, by considering the meteorological state conduction network and weighted processing, it is possible to more accurately capture the complex relationships and patterns in the data. Ultimately, this technical solution helps to obtain a comprehensive and accurate target structured meteorological attribute feature set, providing strong support for subsequent meteorological analysis and forecasting.

[0115] In some preferred embodiments, the obtaining of the one-hot encoding of the first transition meteorological state generated by the vth algorithm branch of the structured storage analysis algorithm includes: obtaining the u-1th round intermediate structured meteorological attribute feature set of the target multimodal meteorological monitoring data; and obtaining the one-hot encoding of the first transition meteorological state generated by the vth algorithm branch based on the meteorological type label and the u-1th round intermediate structured meteorological attribute feature set.

[0116] Based on this embodiment, when executing the structured storage analysis algorithm, the data storage system will obtain the unique hot encoding of the first transition meteorological state generated by the vth algorithm branch of the algorithm. This process is one of the key steps for the algorithm to process multi-mode meteorological monitoring data, which helps the system gradually extract the key features of the data.

[0117] In detail, when the data storage system is ready to perform the u-th round of structured attribute vector recognition, it first reviews and obtains the previous round, that is, the u-1th round of intermediate structured meteorological attribute feature set. This feature set is the key meteorological attributes extracted in the previous round of recognition, which reflects the main characteristics of the data at that stage.

[0118] Next, the data storage system will analyze the current weather type label and the intermediate structured weather attribute feature set of round u-1. The weather type label provides the system with the overall background of the data, while the intermediate structured weather attribute feature set provides the detailed characteristics of the data in the previous stage. By comprehensively considering these two aspects of information, the system can more accurately grasp the current state of the data.

[0119] Then, based on the above comprehensive analysis, the data storage system will obtain the one-hot encoding of the first transitional meteorological state generated by the vth algorithm branch. This one-hot encoding is a temporary, intermediate meteorological state representation, which will be further refined and optimized in the subsequent processing.

[0120] For example, the data storage system is processing a round of data about the "rainy day" weather type. In the previous round of recognition, the system has extracted some key meteorological attributes related to rainy days, such as humidity, precipitation, etc. When the system proceeds to the uth round of recognition, it will combine these key attributes and the "rainy day" weather type label to generate a more accurate first transition weather state unique hot encoding. This encoding may not only contain basic information about rainy days, but also reflect the specific meteorological characteristics extracted in the previous round of recognition.

[0121] In this way, the data storage system can more effectively utilize the historical recognition results and the current meteorological type label to generate a more accurate first transition meteorological state unique hot encoding. This not only improves the accuracy of data processing, but also provides more reliable basic data for subsequent meteorological state analysis and prediction. Overall, this embodiment enhances the intelligence and accuracy of the data storage system when processing multi-mode meteorological monitoring data, and provides strong technical support for applications in the meteorological field.

[0122] In some other preferred embodiments, the method of obtaining the u-1th round intermediate structured meteorological attribute feature set of the target multi-mode meteorological monitoring data includes: in response to u being equal to 1, determining the enhanced meteorological monitoring data set as the u-1th round intermediate structured meteorological attribute feature set; in response to u being greater than 1, performing the u-1th round structured attribute vector identification on the enhanced meteorological monitoring data set through the structured storage analysis algorithm according to the meteorological type label to obtain the u-1th round intermediate structured meteorological attribute feature set.

[0123] Based on this embodiment, the data storage system adopts a flexible and efficient method when executing the structured storage analysis algorithm to obtain the u-1th round of intermediate structured meteorological attribute feature sets of the target multi-mode meteorological monitoring data.

[0124] When the data storage system reaches the first round of structured attribute vector recognition, that is, u is equal to 1, since there is no previous round of recognition results to refer to, the system will directly use the enhanced meteorological monitoring data set as the intermediate structured meteorological attribute feature set for round 0 (that is, round u-1). This is because in the initial stage, the enhanced meteorological monitoring data set itself already contains rich meteorological information and can be used as the starting point for the first round of recognition.

[0125] For example, the data storage system has just started and is ready for the first round of recognition, at which point u = 1. Since there is no recognition result from the previous round, the system will use the enhanced meteorological monitoring data set, which can include multiple meteorological indicators such as temperature, humidity, and wind speed, as the intermediate structured meteorological attribute feature set of round 0 for the first round of structured attribute vector recognition.

[0126] When the data storage system reaches the second round or later, that is, when u is greater than 1, the system will perform the u-1 round of structured attribute vector recognition on the enhanced meteorological monitoring data set based on the current meteorological type label through the structured storage analysis algorithm. This means that the system will use the recognition results of the previous round and the current meteorological type label to conduct a deeper analysis and processing of the data set to extract more accurate meteorological attribute characteristics.

[0127] Taking the second round of recognition as an example, u=2, at this time the system has completed the first round of recognition and obtained the first round of intermediate structured meteorological attribute feature set. Before the second round of recognition, the system will combine the current meteorological type labels, such as "sunny", "rainy", etc., and use the structured storage analysis algorithm to perform the first round of structured attribute vector recognition on the enhanced meteorological monitoring data set to obtain the first round (i.e., the u-1th round) of intermediate structured meteorological attribute feature set. This feature set will serve as the input of the second round of recognition to help the system more accurately identify the second round of meteorological attribute features.

[0128] In this way, the data storage system can flexibly adjust the acquisition method of the intermediate structured meteorological attribute feature set according to the actual situation. In the initial stage, the system can directly use the enhanced meteorological monitoring data set for identification, which improves the processing efficiency; in the subsequent stage, the system can combine the historical identification results and the current meteorological type label for a more in-depth analysis, thereby improving the accuracy of identification. Overall, this embodiment makes the data storage system more flexible and efficient in processing multi-mode meteorological monitoring data, and provides more powerful technical support for applications in the meteorological field.

[0129] In some further preferred embodiments, the first transition meteorological state unique-hot encoding generated by the vth algorithm branch is obtained based on the meteorological type label and the u-1th round intermediate structured meteorological attribute feature set, including: when the vth algorithm branch is the first algorithm branch of the structured storage analysis algorithm, the meteorological type label and the u-1th round intermediate structured meteorological attribute feature set are processed by the vth algorithm branch to obtain the first transition meteorological state unique-hot encoding; when the vth algorithm branch is the transition algorithm branch of the structured storage analysis algorithm, the second transition meteorological state unique-hot encoding generated by the v-1th algorithm branch of the structured storage analysis algorithm is obtained, and the to-be-processed unique-hot encoding of the vth algorithm branch is determined based on the second transition meteorological state unique-hot encoding, and the to-be-processed unique-hot encoding of the vth algorithm branch is processed by the vth algorithm branch to obtain the first transition meteorological state unique-hot encoding.

[0130] Based on this embodiment, when the data storage system obtains the one-hot encoding of the first transition meteorological state generated by the vth algorithm branch based on the meteorological type label and the intermediate structured meteorological attribute feature set of the u-1th round, it will adopt different processing strategies according to the type of algorithm branch.

[0131] When the vth algorithm branch is the first algorithm branch of the structured storage analysis algorithm, the data storage system will directly use this branch to process the meteorological type label and the u-1th round of intermediate structured meteorological attribute feature set. In detail, the system will pass these input data through a series of calculations and transformations in the first algorithm branch, and finally generate the first transition meteorological state one-hot encoding. This one-hot encoding is a preliminary and temporary digital representation of the current meteorological state, which will serve as the basis for subsequent algorithm branch processing.

[0132] To illustrate with a specific example, the data storage system is performing the first round of structured attribute vector recognition, and is currently processing data of the meteorological type "sunny". The first algorithm branch receives the meteorological type label of "sunny" and the intermediate structured meteorological attribute feature set of round 0 (i.e., initial), and converts this information into a one-hot encoding of the first transitional meteorological state, such as a one-hot encoding representing the "sunny" state, through specific algorithm logic.

[0133] When the vth algorithm branch is a transition algorithm branch of the structured storage analysis algorithm, the data processing method will be different. In this case, the data storage system will first obtain the second transition meteorological state unique hot encoding generated by the v-1th algorithm branch of the structured storage analysis algorithm. This second transition meteorological state unique hot encoding is the result of the previous algorithm branch processing, and it contains richer meteorological state information.

[0134] Next, the data storage system determines the pending one-hot encoding of the vth algorithm branch based on the second transitional meteorological state one-hot encoding. This usually involves further analysis or adjustment of the second transitional meteorological state one-hot encoding. Finally, the vth algorithm branch processes the pending one-hot encoding, refines and updates the meteorological state information through specific algorithm logic, and finally generates the first transitional meteorological state one-hot encoding.

[0135] For example, the data storage system has completed the processing of the first two algorithm branches and now enters the processing stage of the third algorithm branch (a transition algorithm branch). The system will first obtain the second transition meteorological state one-hot encoding generated by the second algorithm branch, which can contain more detailed meteorological feature information. Then, the system will determine the unprocessed one-hot encoding of the third algorithm branch based on this encoding, and may optimize or adjust it. Finally, the third algorithm branch will process this unprocessed one-hot encoding and generate a new first transition meteorological state one-hot encoding, which more accurately reflects the current meteorological state.

[0136] In this way, the data storage system can flexibly process meteorological type labels and intermediate structured meteorological attribute feature sets according to the characteristics of different algorithm branches, and generate accurate first transition meteorological state unique hot encodings. This processing method not only improves the efficiency and accuracy of data processing, but also provides a more reliable data basis for subsequent meteorological state analysis and prediction. Overall, this embodiment enhances the adaptability and accuracy of the data storage system when processing complex meteorological data, and brings significant technological progress to the research and application of the meteorological field.

[0137] In the next step, the meteorological type label and the u-1-th round intermediate structured meteorological attribute feature set are processed through the v-th algorithm branch to obtain the first transition meteorological state unique hot coding, including: mining the meteorological type unique hot coding of the meteorological type label; mining the linear meteorological state unique hot coding of the u-1-th round intermediate structured meteorological attribute feature set; processing the meteorological type unique hot coding of the meteorological type label and the linear meteorological state unique hot coding of the u-1-th round intermediate structured meteorological attribute feature set through the v-th algorithm branch to obtain the first transition meteorological state unique hot coding.

[0138] Based on this embodiment, the data storage system will perform a series of operations to process the meteorological type label and the intermediate structured meteorological attribute feature set of the u-1th round through the vth algorithm branch, thereby obtaining the first transition meteorological state unique hot encoding. This process is an important part of the data storage system when performing meteorological data analysis, and is intended to convert the raw data into a format that is easier to process and understand.

[0139] First, the data storage system mines the weather type one-hot encoding of the weather type label. One-hot encoding is a method of converting categorical variables into a format that is easy for machine learning algorithms to exploit. In this case, each weather type (such as sunny, rainy, snowy, etc.) is converted into a unique binary vector in which only one element is 1 and the rest are 0. For example, if the weather type label is "sunny", the corresponding one-hot encoding may be a vector in which the position representing "sunny" is 1 and the rest are 0.

[0140] Secondly, the data storage system mines the linear meteorological state one-hot encoding of the intermediate structured meteorological attribute feature set of round u-1. The "linear meteorological state" here may refer to the meteorological state features extracted by some linear model or method, which are also converted into one-hot encoding format for subsequent processing. The purpose of this step is to extract key information from the structured meteorological attributes of the previous round and represent it in the form of one-hot encoding.

[0141] Finally, the data storage system processes these two sets of unique hot codes (i.e., the unique hot code of the weather type label and the linear weather state unique hot code of the intermediate structured weather attribute feature set of the u-1th round) through the vth algorithm branch. This processing may include weighting, merging, conversion, or other complex calculations, depending on the design of the algorithm branch. Ultimately, this processing step generates the first transition weather state unique hot code, which is an intermediate representation that combines the weather type and structured weather attribute features, providing a basis for subsequent weather state recognition and prediction.

[0142] For example, the data storage system is processing a data set containing a weather type label of "sunny" and a series of meteorological attributes such as temperature and humidity. The system will first convert "sunny" into the corresponding one-hot encoding, and then extract the linear meteorological state one-hot encoding of attributes such as temperature and humidity. Then, through the specific calculation logic of the vth algorithm branch, these two sets of one-hot encodings are combined to generate a first transition meteorological state one-hot encoding that combines the meteorological type and attribute characteristics. This code not only reflects that it is currently "sunny", but also contains the specific meteorological attribute state under this weather type.

[0143] In this way, the data storage system can effectively convert the meteorological type label and the structured meteorological attribute feature set into the first transition meteorological state unique hot encoding. This conversion not only simplifies the representation of the data, making it easier to be processed by the machine learning algorithm, but also retains the key information of the original data. Therefore, this embodiment improves the efficiency and accuracy of the data storage system in processing and analyzing meteorological data, and provides strong support for subsequent meteorological state identification and prediction.

[0144] In the next step, the one-hot coding to be processed of the vth algorithm branch is determined based on the one-hot coding of the second transition meteorological state, including: determining the one-hot coding of the second transition meteorological state as the one-hot coding to be processed of the vth algorithm branch; or mapping the one-hot coding of the second transition meteorological state to the target multi-mode meteorological monitoring data to obtain the meteorological state transmission network, and weighting the meteorological state one-hot coding of Z network units in the meteorological state transmission network with the meteorological state one-hot coding of the corresponding neighboring network units to obtain the second meteorological state weighted one-hot coding, and determining the one-hot coding to be processed of the vth algorithm branch based on the second meteorological state weighted one-hot coding.

[0145] Based on this embodiment, the data storage system will determine the to-be-processed one-hot encoding of the vth algorithm branch according to the one-hot encoding of the second transitional meteorological state. This process is one of the key links in the data storage system processing multi-mode meteorological monitoring data, which ensures the smooth transmission and correct processing of data between algorithm branches.

[0146] One direct way is that the data storage system directly determines the second transition meteorological state one-hot encoding as the one-hot encoding to be processed by the vth algorithm branch. In this case, the second transition meteorological state one-hot encoding, as the output of the previous algorithm branch, directly becomes the input of the next algorithm branch, maintaining the consistency of data processing.

[0147] For example, the data storage system has obtained a one-hot encoding of the second transitional meteorological state representing "cloudy to overcast" through the previous algorithm branch. In this encoding, the state bits of "cloudy" and "overcast" may be activated (set to 1), and the remaining state bits are 0. When this encoding is directly determined as the one-hot encoding to be processed by the v-th algorithm branch, the v-th algorithm branch will perform subsequent processing based on this "cloudy to overcast" state.

[0148] Another way is that the data storage system maps the second transition meteorological state unique hot encoding to the target multi-mode meteorological monitoring data to construct a meteorological state conduction network. In this network, each network unit represents a specific meteorological state, and the connection between network units represents the conduction relationship between states. The system weights the meteorological state unique hot encoding of each network unit and the meteorological state unique hot encoding of its neighboring network unit to obtain the weighted unique hot encoding of the second meteorological state. This weighting process takes into account the mutual influence between different meteorological states, making data processing more refined and accurate.

[0149] Continuing with the above-mentioned "cloudy to overcast" example, in the meteorological state transmission network, "cloudy" and "overcast" may be adjacent network units. When performing weighted processing, the system will consider the impact of "cloudy" on "overcast" and the impact of "overcast" on "cloudy", and weight them according to the preset weights. The weighted one-hot encoding of the second meteorological state obtained in this way not only reflects the current meteorological state, but also contains the transition trend between states. Finally, this weighted one-hot encoding will be determined as the one-hot encoding to be processed for the vth algorithm branch.

[0150] In this way, the data storage system can flexibly and accurately determine the unique hot encoding to be processed for the vth algorithm branch. Whether directly using the second transition meteorological state unique hot encoding or weighted processing through the meteorological state conduction network, the continuity and accuracy of data processing can be ensured. This method not only improves the ability of the data storage system to process complex meteorological data, but also provides a more reliable data basis for subsequent meteorological analysis and prediction. Overall, this embodiment enhances the adaptability and accuracy of the data storage system in meteorological data processing.

[0151] In some alternative embodiments, the meteorological state unique-hot codes of the Z network units in the meteorological state transmission network are weighted with the meteorological state unique-hot codes of the corresponding neighboring network units to obtain a weighted unique-hot code of the target meteorological state, including: for each of the Z network units, weighting the meteorological state unique-hot code of the network unit with the meteorological state unique-hot code of the neighboring network units of the network unit as the current meteorological state unique-hot code of the network unit to obtain a vector-integrated meteorological state transmission network; based on the vector-integrated meteorological state transmission network, obtaining the target meteorological state weighted unique-hot code, the target meteorological state weighted unique-hot code is a first meteorological state weighted unique-hot code or a second meteorological state weighted unique-hot code.

[0152] Based on this embodiment, the data storage system will perform a series of complex operations to weight the unique hot encoding of the meteorological state of each network unit in the meteorological state transmission network. Here, the implementation details of this process are described in detail.

[0153] First, consider the Z network units contained in the weather state transmission network, each of which has a corresponding one-hot encoding of the weather state. One-hot encoding is a method of converting categorical variables into a format that can be used by machine learning algorithms. In one-hot encoding, only one element is 1 (indicating "hot"), and the rest are 0. For example, if there are three weather states: sunny, rainy, and cloudy, then the one-hot encoding of sunny days may be [1, 0, 0], rainy days are [0, 1, 0], and cloudy days are [0, 0, 1].

[0154] Next, the data storage system will perform the following steps for each of the Z network units: Identify the neighbor network units of the network unit. In the network structure, neighbors refer to other units directly connected to the network unit. Obtain the unique hot codes of the weather status of the network unit and its neighbor network units. Weight these unique hot codes. The weighting method can be determined according to the specific application requirements and network structure. A simple example is that if a network unit has three neighbors, the unique hot code of the weather status of each neighbor will be assigned a weight, and then these weighted codes will be added to the unique hot code of the weather status of the network unit itself to obtain a new weighted unique hot code. After the above processing, each network unit will obtain a current weighted unique hot code of the weather status, and these codes together constitute the weather status transmission network after vector integration. Finally, based on this vector integrated network, the data storage system will generate a weighted unique hot code of the target weather status. This target code can be a weighted unique hot code of the first weather status, representing a specific weather condition or mode, or a weighted unique hot code of the second weather status, representing another condition or mode.

[0155] For example, there is a simple weather state transmission network, which contains five network units, each of which can be in one of the three states of sunny, rainy or cloudy. The data storage system will first obtain the one-hot encoding for each network unit and its neighbors. For example, if the states of a network unit and its two neighbors are rainy, sunny and cloudy, respectively, then their one-hot encodings may be [0, 1, 0], [1, 0, 0] and [0, 0, 1]. Then, the system will assign weights to these codes, such as 0.5, 0.3 and 0.2, and perform weighted summation to obtain a new weighted one-hot encoding. This process will be repeated for each unit in the network, eventually forming a weighted weather state transmission network.

[0156] In this way, the data storage system can effectively integrate the meteorological status information of each unit in the network and generate a more representative weighted unique hot encoding of the target meteorological status. This not only improves the efficiency and accuracy of data processing, but also helps to more comprehensively understand the distribution and change trend of the meteorological status of the entire network. In this way, by weighting the unique hot encoding of the meteorological status of the network unit, a more comprehensive and accurate analysis of the meteorological status transmission network is achieved. This method not only enhances the data processing capabilities, but also provides strong technical support for meteorological forecasting, environmental monitoring and other fields, and helps to improve the performance and accuracy of related applications.

[0157] In some other alternative embodiments, the meteorological state conduction network after the vector integration is used to obtain the weighted one-hot encoding of the target meteorological state, including: based on the meteorological monitoring data set, feature encoding the meteorological state conduction network after the vector integration is performed to obtain the weighted one-hot encoding of the target meteorological state.

[0158] In this embodiment, the data storage system will take more complex steps to obtain the weighted unique hot encoding of the target meteorological state based on the meteorological state conduction network after vector integration. This process involves in-depth analysis of the meteorological monitoring data set and the application of feature encoding technology.

[0159] First, the data storage system obtains the meteorological monitoring data set, which is a collection of a large amount of meteorological observation data, including temperature, humidity, wind speed, wind direction, air pressure and other meteorological parameters. These data are the basis for feature encoding.

[0160] Next, the data storage system uses this data to feature encode the meteorological state transmission network after vector integration. Feature encoding is a process of converting raw data into a format that can be more easily understood by machine learning models. In this context, the purpose of feature encoding is to extract features related to meteorological conditions from meteorological monitoring datasets and embed these features into the meteorological state transmission network after vector integration.

[0161] In detail, the data storage system analyzes the various parameters in the meteorological monitoring data set and identifies features that are closely related to changes in meteorological conditions. For example, the system may find that within a certain temperature and humidity range, the meteorological condition is more likely to change from sunny to rainy. Based on such a finding, the system will feature encode the corresponding nodes in the meteorological state transmission network after vector integration to reflect the possibility of such a meteorological state transition.

[0162] When encoding features, the data storage system also considers time and space factors. For example, the system may analyze the evolution of meteorological conditions in different time periods, or their distribution in geographic space, and encode this information into the network.

[0163] After completing the feature encoding, the data storage system will generate a weighted one-hot encoding of the target meteorological state. This encoding not only reflects the current meteorological state of each network unit, but also incorporates the feature information extracted from the meteorological monitoring data set, making it more representative and more valuable for prediction.

[0164] For example, the data storage system is processing a meteorological monitoring data set that contains temperature, humidity, and wind speed data. By analyzing this data, the system finds that when the temperature and humidity reach a certain threshold, an increase in wind speed often indicates an upcoming rainy day. Therefore, when the system performs feature encoding on the meteorological state transmission network after vector integration, it pays special attention to those network units that meet this condition and reflects the high probability of rainy day state in its weighted unique hot encoding.

[0165] In this way, the data storage system can more accurately capture and predict changes in meteorological conditions, providing more reliable data support for applications such as weather forecasting and disaster warning. Furthermore, by encoding the characteristics of the meteorological state transmission network after vector integration based on the meteorological monitoring data set, a more accurate and comprehensive weighted unique hot encoding of the target meteorological state is obtained, which improves the processing efficiency and accuracy of meteorological data.

[0166] In the next step, the meteorological state conduction network after the vector integration is feature encoded based on the meteorological monitoring data set to obtain the weighted unique hot encoding of the target meteorological state, including: according to the feature encoding rules between the meteorological monitoring data set and the target multi-mode meteorological monitoring data, the meteorological state conduction network after the vector integration is mapped to the meteorological monitoring data set to obtain the third meteorological state weighted unique hot encoding; according to the third meteorological state weighted unique hot encoding, the target meteorological state weighted unique hot encoding is obtained.

[0167] Based on this technical solution, the data storage system will take a series of refined operations to feature encode the meteorological state transmission network after vector integration according to the meteorological monitoring data set, and finally obtain the weighted unique hot encoding of the target meteorological state.

[0168] First, the data storage system will clarify the feature encoding rules between the meteorological monitoring data set and the target multi-mode meteorological monitoring data. These rules define how to convert the original meteorological monitoring data into a format that can be understood by machine learning or data analysis algorithms. For example, the rules may include how to discretize continuous variables such as temperature, humidity, wind speed, or how to combine multiple meteorological parameters to form a comprehensive feature.

[0169] Subsequently, according to these feature encoding rules, the data storage system maps the vector-integrated meteorological state transmission network to the meteorological monitoring data set. This mapping process aims to find the correspondence between each node or edge in the network and the specific features in the meteorological monitoring data. Through this mapping, the system can link the abstract meteorological state in the network with the actual meteorological observation data.

[0170] During the mapping process, the data storage system generates a third weather state weighted one-hot encoding for each network unit. This encoding is generated based on the specific performance of the corresponding features of the network unit in the weather monitoring data set. For example, if a network unit represents a specific weather pattern, then its third weather state weighted one-hot encoding will be determined based on the frequency, intensity, and other characteristics of the pattern in the weather monitoring data set.

[0171] Finally, based on these third meteorological state weighted unique hot codes, the data storage system will derive the target meteorological state weighted unique hot code. This process may involve aggregation, averaging or other statistical processing of multiple third meteorological state weighted unique hot codes, so as to obtain a target code that can fully reflect the current state of the entire meteorological state transmission network.

[0172] For example, the data storage system is processing a meteorological monitoring data set that contains multiple parameters such as temperature, humidity, and wind speed. The system will first convert these parameters into features suitable for machine learning models according to feature encoding rules. Then, the system will map each node in the meteorological state conduction network after vector integration to these features, and generate a third meteorological state weighted one-hot encoding for each node. These encodings may reflect the correlation between different meteorological parameters, such as high temperature and high humidity may indicate upcoming rainfall. Finally, the system will comprehensively process these third meteorological state weighted one-hot encodings to obtain a target meteorological state weighted one-hot encoding that can represent the current meteorological state of the entire network.

[0173] In this way, the data storage system can more effectively extract useful information from the meteorological monitoring data set and integrate it into the analysis of the meteorological state transmission network, thereby improving the ability to predict and understand changes in meteorological states. In this way, by mapping the vector-integrated meteorological state transmission network to the meteorological monitoring data set based on the feature encoding rules between the meteorological monitoring data set and the target multi-mode meteorological monitoring data, an accurate and comprehensive weighted unique hot encoding of the target meteorological state is obtained. This method not only enhances the ability of the data storage system to process complex meteorological data, but also provides more accurate and reliable data support for research and application in the fields of meteorological forecasting and environmental monitoring, demonstrating significant technological progress and practical application value.

[0174] Under an optional technical idea, the weighted unique hot coding of the target meteorological state is obtained based on the weighted unique hot coding of the third meteorological state, including: according to the dimension information of the target transition meteorological state unique hot coding, the dimension of the third meteorological state weighted unique hot coding is adjusted to obtain the weighted unique hot coding of the target meteorological state; wherein, the dimension of the target meteorological state weighted unique hot coding is consistent with the dimension of the target transition meteorological state unique hot coding, when the target transition meteorological state unique hot coding is the first transition meteorological state unique hot coding, then the target meteorological state weighted unique hot coding is the first meteorological state weighted unique hot coding, and when the target transition meteorological state unique hot coding is the second transition meteorological state unique hot coding, then the target meteorological state weighted unique hot coding is the second meteorological state weighted unique hot coding.

[0175] Under this technical idea, the operations performed by the data storage system will be more complex and sophisticated, especially when processing the weighted one-hot encoding of the meteorological state. Here we will elaborate on how to derive the weighted one-hot encoding of the target meteorological state from the weighted one-hot encoding of the third meteorological state.

[0176] First, the data storage system obtains the third meteorological state weighted one-hot encoding, which is the mapping result of the meteorological state transmission network after the feature encoding and vector integration of the meteorological monitoring data set in the previous step. The third meteorological state weighted one-hot encoding reflects the meteorological state and its weight of each node or area in the network.

[0177] Next, the system will refer to the dimension information of the target transitional weather state one-hot encoding. The transitional weather state one-hot encoding represents the temporary state during the weather state transition process, and its dimension information indicates the number and representation of different weather states. For example, if there are five different weather states, then the dimension of the transitional weather state one-hot encoding is five, and each dimension corresponds to a specific weather state.

[0178] Next, the data storage system will adjust the dimension of the weighted one-hot encoding of the third meteorological state according to the dimension information of the one-hot encoding of the target transition meteorological state. The purpose of this step is to ensure the structural consistency of the weighted one-hot encoding of the target meteorological state and the one-hot encoding of the transition meteorological state, so as to facilitate subsequent data analysis and model training. Dimension adjustment can include operations such as increasing or decreasing the dimension of the encoding and reallocating weights.

[0179] After the dimension adjustment is completed, the data storage system will obtain the weighted one-hot encoding of the target meteorological state. This encoding not only inherits the meteorological state information in the weighted one-hot encoding of the third meteorological state, but also adapts to the dimension structure of the one-hot encoding of the transitional meteorological state, so it has more analytical value and prediction accuracy.

[0180] In detail, if the target transitional meteorological state unique hot encoding is the first transitional meteorological state unique hot encoding, then the dimensionally adjusted target meteorological state weighted unique hot encoding is called the first meteorological state weighted unique hot encoding. Similarly, if the target transitional meteorological state unique hot encoding is the second transitional meteorological state unique hot encoding, then the obtained target meteorological state weighted unique hot encoding is the second meteorological state weighted unique hot encoding.

[0181] For example, the weighted one-hot encoding of the third meteorological state originally obtained by the data storage system is a four-dimensional vector, representing four different meteorological states and their weights. Now, the system learns that the one-hot encoding of the target transition meteorological state is five-dimensional, which means that five meteorological states need to be considered. Therefore, the system will adjust the dimension of the weighted one-hot encoding of the third meteorological state, which may be to add a dimension on the basis of the original four dimensions and redistribute the weights of each dimension to obtain a five-dimensional weighted one-hot encoding of the target meteorological state.

[0182] In this way, the data storage system can flexibly adapt to the requirements of meteorological state representation of different dimensions and complexity, and improve the versatility and accuracy of data processing. In this way, by adjusting the dimension of the weighted unique hot encoding of the third meteorological state according to the dimension information of the unique hot encoding of the target transition meteorological state, the weighted unique hot encoding of the target meteorological state is obtained, which is consistent with the dimension of the unique hot encoding of the transition meteorological state. This method not only enhances the ability of the data storage system to process variable meteorological data, but also provides more accurate and flexible data support for meteorological forecasting and analysis.

[0183] In some preferred embodiments, the to-be-processed unique-hot coding of the target algorithm branch is determined based on the weighted unique-hot coding of the target meteorological state, including: taking the weighted unique-hot coding of the target meteorological state as the to-be-processed unique-hot coding of the target algorithm branch; or, weighting the target transition meteorological state unique-hot coding and the weighted unique-hot coding of the target meteorological state to obtain the to-be-processed unique-hot coding of the target algorithm branch; wherein, when the target algorithm branch is the v+1th algorithm branch, the weighted unique-hot coding of the target meteorological state is the weighted unique-hot coding of the first meteorological state, and the unique-hot coding of the target transition meteorological state is the unique-hot coding of the first transition meteorological state; when the target algorithm branch is the vth algorithm branch, the weighted unique-hot coding of the target meteorological state is the weighted unique-hot coding of the second meteorological state, and the unique-hot coding of the target transition meteorological state is the unique-hot coding of the second transition meteorological state.

[0184] In this embodiment, the data storage system will adopt a more flexible and accurate strategy when determining the one-hot encoding to be processed of the target algorithm branch. This step is a key link in the data processing flow, which directly affects the processing effect and accuracy of the subsequent algorithm branches.

[0185] First, the data storage system may directly use the weighted unique hot code of the target meteorological state as the pending unique hot code of the target algorithm branch. This approach is suitable for scenarios where the meteorological state has a significant impact on the algorithm branch and there is no need to consider the transitional meteorological state. For example, under certain specific meteorological conditions, such as high temperature and dryness, the algorithm branch may need to pay special attention to the risk of fire. At this time, it is reasonable to directly use the corresponding meteorological state weighted unique hot code as the pending code.

[0186] However, in more cases, the data storage system considers the impact of the target transitional meteorological state one-hot encoding. The transitional meteorological state represents the transition process from one meteorological state to another meteorological state, and it can contain important dynamic information. Therefore, the system performs weighted processing on the target transitional meteorological state one-hot encoding and the target meteorological state weighted one-hot encoding to obtain the one-hot encoding to be processed of the target algorithm branch.

[0187] The specific method of weighted processing may vary depending on the application scenario, but it usually involves numerically weighted summation of the corresponding dimensions of the two codes. In this way, the processed one-hot code combines both the current meteorological state and the trend of the meteorological state, providing more comprehensive input information for the algorithm branch.

[0188] It is worth noting that when the target algorithm branch is different, the weighted unique hot coding of the meteorological state and the unique hot coding of the transitional meteorological state involved will also be different. For example, when the target algorithm branch is the v+1th algorithm branch, the system will use the weighted unique hot coding of the first meteorological state and the unique hot coding of the first transitional meteorological state for weighted processing; and when the target algorithm branch is the vth algorithm branch, the second meteorological state weighted unique hot coding and the second transitional meteorological state unique hot coding will be used.

[0189] For example: the data storage system is processing a meteorological data set containing parameters such as temperature, humidity, and wind speed, and needs to determine the fire risk level of a certain area based on these data. At a certain point in time, the system detects a sudden rise in temperature and a sudden drop in humidity, which may indicate an increase in fire risk. At this point, the system will weight this meteorological state transition trend (i.e., transitional meteorological state) with the current meteorological state (high temperature, low humidity) to obtain a unique hot code to be processed that combines dynamic and static information. This code will then be input into the corresponding algorithm branch for a more accurate fire risk assessment.

[0190] In this way, the data storage system can flexibly generate unprocessed unique hot codes containing rich information according to the needs of different algorithm branches. This not only improves the pertinence and accuracy of data processing, but also provides a solid foundation for subsequent data analysis and decision support. In this way, by flexibly weighting the target meteorological state weighted unique hot code or weighting the target transition meteorological state unique hot code, the unprocessed unique hot code that adapts to the needs of different algorithm branches is obtained. This method not only improves the sophistication and accuracy of data processing, but also provides more comprehensive and dynamic data support for various application scenarios, showing significant technological progress and practical application value.

[0191] In some other preferred embodiments, the target transition meteorological state unique-hot coding and the target meteorological state weighted unique-hot coding are weighted to obtain the target algorithm branch's unprocessed unique-hot coding, including: determining the confidence coefficients corresponding to the target transition meteorological state unique-hot coding and the target meteorological state weighted unique-hot coding respectively; weighting the target transition meteorological state unique-hot coding and the target meteorological state weighted unique-hot coding according to the confidence coefficients corresponding to the target transition meteorological state unique-hot coding and the target meteorological state weighted unique-hot coding respectively to obtain the target algorithm branch's unprocessed object.

[0192] Based on this embodiment, when performing weighted processing of target transition meteorological state one-hot encoding and target meteorological state weighted one-hot encoding, the data storage system will adopt a more sophisticated method to determine the one-hot encoding to be processed. This process involves two main steps: determination of confidence coefficient and weighted processing.

[0193] First, the data storage system determines the confidence coefficients corresponding to the target transition meteorological state unique hot encoding and the target meteorological state weighted unique hot encoding. The confidence coefficient is an indicator used to measure the reliability of data or information, and its value is usually between 0 and 1. Here, the confidence coefficient reflects the system's trust or accuracy assessment of the two codes. For example, if the target meteorological state weighted unique hot encoding is calculated based on a large amount of historical data and an accurate model, it may have a higher confidence coefficient; if the target transition meteorological state unique hot encoding is based on recent meteorological change trend forecasts, its confidence coefficient may be slightly lower.

[0194] Next, the data storage system will perform weighted processing on the two codes according to their corresponding confidence coefficients. Specifically, the system will multiply each dimension value of each code by its corresponding confidence coefficient, and then add the two weighted codes together to obtain the object to be processed of the target algorithm branch. This process ensures that the one-hot code to be processed takes into account both the current meteorological state (reflected by the weighted one-hot code of the target meteorological state) and the dynamic change trend of the meteorological state (reflected by the one-hot code of the target transition meteorological state), and makes reasonable weight distribution according to their respective confidence levels.

[0195] For example: the data storage system obtains a target meteorological state weighted one-hot encoding [0.8, 0.1, 0.1, 0] and a target transition meteorological state one-hot encoding [0.2, 0.2, 0.5, 0.1], representing four different meteorological states respectively. At the same time, the system determines that the confidence coefficient of the weighted one-hot encoding of the target meteorological state is 0.9, and the confidence coefficient of the one-hot encoding of the target transition meteorological state is 0.7. When performing weighted processing, the system multiplies each dimension value of each code by its confidence coefficient and then adds them together. For example, for the first dimension, the weighted value is 0.80.9+0.20.7=0.86. Similarly, the weighted values ​​of other dimensions can be calculated to obtain the final one-hot encoding to be processed.

[0196] Through this method, the data storage system can comprehensively consider the reliability of the current meteorological status and the trend of meteorological status changes, and generate more accurate and comprehensive unique hot codes to be processed. This not only improves the scientificity and rationality of data processing, but also provides a more reliable data foundation for subsequent data analysis and decision support.

[0197] It can be seen that the data storage system obtains more accurate and comprehensive processing objects of the target algorithm branch by determining the confidence coefficients of the target transition meteorological state unique hot coding and the target meteorological state weighted unique hot coding, and performing weighted processing according to these confidence coefficients. This method not only improves the accuracy and reliability of data processing, but also provides more solid and reliable data support for various application scenarios, showing significant technological progress and practical application potential.

[0198] In an exemplary embodiment, determining the confidence coefficients corresponding to the target transition meteorological state unique-hot encoding and the target meteorological state weighted unique-hot encoding, respectively, includes: processing the target transition meteorological state unique-hot encoding and the target meteorological state weighted unique-hot encoding through a feature focusing strategy to obtain the confidence coefficients corresponding to the target transition meteorological state unique-hot encoding and the target meteorological state weighted unique-hot encoding, respectively.

[0199] In this embodiment, the data storage system adopts a feature focusing strategy when determining the confidence coefficients corresponding to the target transition meteorological state one-hot encoding and the target meteorological state weighted one-hot encoding. This strategy aims to evaluate the accuracy and reliability of the encoding through in-depth analysis, thereby providing a scientific basis for subsequent weighted processing.

[0200] First, the data storage system applies a feature focusing strategy, which is a method of evaluating the quality or credibility of data by extracting and analyzing its key features. In this scenario, the system carefully processes the target transition meteorological state one-hot encoding and the target meteorological state weighted one-hot encoding.

[0201] In detail, for the one-hot encoding of the target transitional meteorological state, the system may focus on factors such as the consistency of its transition trend, the degree of consistency with historical data, and the accuracy of the prediction model. For example, if the transition of a certain transitional meteorological state is highly consistent with the common pattern in historical data, and the prediction model performs more accurately in this state, then the confidence coefficient of the encoding may be increased accordingly.

[0202] Similarly, for the weighted one-hot encoding of the target meteorological state, the system will evaluate the rationality of its weight distribution, the matching degree with the real-time meteorological data, and the scientific nature of the weighting strategy. For example, if the weighted one-hot encoding can accurately reflect the current meteorological state and the weight distribution is consistent with the principles of meteorology, then its confidence coefficient will also increase accordingly.

[0203] Through the comprehensive evaluation of feature focusing strategies, the data storage system can generate a reasonable confidence coefficient for each code. These confidence coefficients not only reflect the accuracy and reliability of the code, but also provide an important reference for subsequent weighted processing.

[0204] For example: the data storage system is processing a meteorological data set containing parameters such as temperature, humidity, and wind speed. The system will first apply a feature focus strategy to conduct an in-depth analysis of the target transition meteorological state one-hot encoding and the target meteorological state weighted one-hot encoding. During the analysis process, the system may find that the transformation of a certain transition meteorological state is highly consistent with the common pattern in historical data, and the prediction model has a high accuracy rate in this state, so it assigns a higher confidence coefficient to the transition meteorological state one-hot encoding. At the same time, for weighted one-hot encoding, the system may evaluate the rationality of its weight allocation based on real-time meteorological data and meteorological principles, and determine its confidence coefficient accordingly.

[0205] Finally, after obtaining these confidence coefficients, the data storage system will perform weighted processing on the target transition meteorological state one-hot encoding and the target meteorological state weighted one-hot encoding based on them, thereby obtaining a more accurate and comprehensive one-hot encoding to be processed. This method not only improves the scientificity and accuracy of data processing, but also provides a more reliable data foundation for subsequent data analysis and decision support.

[0206] It can be seen that this data storage system uses a feature focusing strategy to determine the confidence coefficient of the target transition meteorological state one-hot encoding and the target meteorological state weighted one-hot encoding. This method can comprehensively consider the accuracy and reliability of the encoding and provide a scientific basis for subsequent weighted processing. This not only improves the accuracy and credibility of data processing, but also provides more solid and accurate data support for various application scenarios, demonstrating significant technological progress and practical application value.

[0207] It is worth mentioning that, through the feature focusing strategy, the target transition meteorological state unique-hot coding and the target meteorological state weighted unique-hot coding are processed to obtain the confidence coefficients corresponding to the target transition meteorological state unique-hot coding and the target meteorological state weighted unique-hot coding, respectively, including: constructing a multidimensional feature evaluation model, which integrates meteorological principles, historical data analysis and machine learning technology; using the model to extract dynamic features of the target transition meteorological state unique-hot coding, identify the coherence and stability of its transformation trend and the similarity with the historical transformation pattern, and assign a transition confidence coefficient to it accordingly; at the same time, static feature analysis is performed on the target meteorological state weighted unique-hot coding to evaluate the rationality of the weights of each dimension, the consistency with the current real-time meteorological data and its representativeness in the long-term meteorological data, and a state confidence coefficient is assigned to it according to the analysis results; finally, combining the analysis results of dynamic and static features, the final confidence coefficients of the target transition meteorological state unique-hot coding and the target meteorological state weighted unique-hot coding are comprehensively determined.

[0208] In detail, the data storage system adopts a sophisticated and comprehensive feature focusing strategy when determining the confidence coefficients of the target transition meteorological state one-hot encoding and the target meteorological state weighted one-hot encoding. This process involves building a multidimensional feature evaluation model, the design of which integrates in-depth meteorological principles, rich historical data analysis experience, and advanced machine learning technology.

[0209] First, the data storage system uses this multidimensional feature evaluation model to dynamically extract features from the one-hot encoding of the target transitional meteorological state. This step focuses on the transition trend of the transitional meteorological state, including its coherence, stability, and similarity to historical transition patterns. For example, the system analyzes whether the transitional state shows a smooth and continuous trend of change, and whether this trend is consistent with the meteorological transition patterns observed in the past. Based on these analyses, the system assigns a transition confidence coefficient to the one-hot encoding of the target transitional meteorological state, which reflects the credibility and stability of the transitional state.

[0210] At the same time, the data storage system will also perform static feature analysis on the weighted unique hot coding of the target meteorological state. This step focuses on evaluating the rationality of the weights of each dimension of the weighted unique hot coding, as well as the consistency with the current real-time meteorological data and its representativeness in long-term meteorological data. For example, the system will check whether the weight distribution in the weighted unique hot coding is scientific and conforms to the basic principles of meteorology, and will compare it with real-time meteorological observation data to verify its accuracy. In addition, the system will also examine the frequency and stability of the code in historical data to evaluate its representativeness. Based on the analysis results of these static features, the system will assign a state confidence coefficient to the weighted unique hot coding of the target meteorological state, which reflects the accuracy and reliability of the code.

[0211] Finally, the data storage system will combine the analysis results of dynamic and static features to comprehensively determine the final confidence coefficients of the target transition meteorological state one-hot encoding and the target meteorological state weighted one-hot encoding. This process ensures the scientificity and comprehensiveness of the confidence coefficients, providing a solid foundation for subsequent data processing and decision support.

[0212] For example, the data storage system is processing a data set containing multiple meteorological parameters such as temperature, humidity, and wind speed. When using the multidimensional feature evaluation model for analysis, the system found that the transition trend of the one-hot encoding of a target transition meteorological state is highly similar to the historical pattern of cold air moving south, and the changes are coherent and stable, so it is assigned a higher transition confidence coefficient. At the same time, for the weighted one-hot encoding of a target meteorological state, the system found that its weight distribution is reasonable, highly consistent with the real-time observation data, and highly representative in the historical data, so it is also assigned a higher state confidence coefficient.

[0213] Through this method, the data storage system can more accurately evaluate and process meteorological data, improving the scientificity and reliability of data processing. This not only helps to improve the accuracy of meteorological forecasts, but also provides more solid and reliable data support for various application scenarios.

[0214] In summary, the data storage system determines the confidence coefficients of the target transition meteorological state one-hot encoding and the target meteorological state weighted one-hot encoding by adopting a feature focusing strategy and a multidimensional feature evaluation model. This method comprehensively considers the analysis results of dynamic and static features, providing a scientific basis for subsequent weighted processing.

[0215] In another exemplary embodiment, the method further includes: when the u-th round structured attribute vector identification does not belong to the Y-th round structured attribute vector identification, the first transition meteorological state one-hot encoding is determined as the one-hot encoding to be processed of the v+1th algorithm branch of the structured storage analysis algorithm.

[0216] In this embodiment, the method executed by the data storage system covers more case processing, especially the branch logic when processing structured attribute vector recognition. When the data storage system proceeds to the uth round of structured attribute vector recognition and finds that the recognition result of this round does not belong to any of the Y rounds of structured attribute vector recognition that have been performed before, the system will adopt a specific strategy to determine the one-hot encoding to be processed for the next algorithm branch.

[0217] In detail, if the structured attribute vector recognition of round u is unique, that is, it does not repeat the recognition results of the previous round Y, then the data storage system will consider using the first transition meteorological state one-hot encoding. This encoding may represent the transition from one meteorological state to another, contains important dynamic information, and is very valuable for analyzing the changing trend of meteorological data and predicting future states.

[0218] In this case, the data storage system will determine the first transitional meteorological state unique hot code as the to-be-processed unique hot code of the v+1th algorithm branch of the structured storage analysis algorithm. This means that in the subsequent branch processing, the algorithm will mainly focus on the information contained in this transitional state in order to more accurately capture and predict changes in meteorological conditions.

[0219] For example, a data storage system is monitoring the weather conditions in a region and has performed multiple rounds of structured attribute vector recognition. In a certain round of recognition (i.e., round u), the system finds that the current weather state is different from the recognition results of all previous rounds (i.e., round Y), which may mean that the weather conditions in the region are changing significantly. At this time, the system will choose to use the first transition weather state one-hot encoding, which reflects the transition of the weather state from the previous state to the current state. Then, the system uses this one-hot encoding as the input of the v+1th algorithm branch to further analyze the possible impact and consequences of this weather state transition.

[0220] In this way, the data storage system can handle different meteorological conditions more flexibly, especially those with sudden or significant meteorological changes. This not only improves the system's adaptability to complex meteorological conditions, but also provides the possibility for more accurate meteorological forecasting and analysis.

[0221] In this way, when processing structured attribute vector recognition, the data storage system can flexibly select the one-hot encoding to be processed according to the actual situation. In particular, when encountering new and unique meteorological conditions, the system can use the first transition meteorological state one-hot encoding to deeply analyze the changing trend of meteorological conditions, thereby improving the accuracy and foresight of meteorological data analysis. This flexibility and accuracy not only enhances the practical value of the system, but also provides a more comprehensive and in-depth data foundation for meteorological forecasting and decision support.

[0222] In some independent embodiments, the method further includes: based on the target structured meteorological attribute feature set, performing structured storage on the target multi-mode meteorological monitoring data.

[0223] In detail, based on the target structured meteorological attribute feature set, the target multi-mode meteorological monitoring data is structured and stored, including: building a flexible and scalable data storage architecture that can adapt to meteorological monitoring data of different modes; defining a series of data storage templates according to the target structured meteorological attribute feature set, each template corresponding to a specific meteorological attribute or feature; classifying and mapping the target multi-mode meteorological monitoring data into the corresponding data storage template according to its attributes or features; designing a reasonable data structure and indexing mechanism within each template to ensure efficient retrieval and update of data; at the same time, implementing data compression and encryption strategies to optimize storage space and ensure data security; finally, ensuring the accuracy and integrity of stored data through automated data verification and error correction mechanisms. Through this method, the structured, efficient and secure storage of target multi-mode meteorological monitoring data can be achieved.

[0224] In an embodiment of the data storage system, the process of structured storage of target multi-mode meteorological monitoring data based on the target structured meteorological attribute feature set is highly refined and systematic.

[0225] First, the data storage system will build a flexible and scalable data storage architecture. The design of this architecture takes into account the diversity and complexity of meteorological monitoring data, so it can adapt to data of different modalities, such as temperature, humidity, wind speed, air pressure, etc. This flexibility ensures that no matter how the data type changes, the storage architecture can be effectively adapted.

[0226] Next, the system will define a series of data storage templates based on the target structured meteorological attribute feature set. Each template is designed based on a specific meteorological attribute or feature, such as temperature template, humidity template, etc. These templates not only provide structured storage space for data, but also make data classification and retrieval more convenient.

[0227] Then, the target multi-mode meteorological monitoring data will be classified and mapped to the corresponding data storage template according to its own attributes or characteristics. For example, temperature data will be stored in the temperature template, and wind speed data will be stored in the wind speed template. This classified storage method helps to organize and manage data, and also improves data accessibility.

[0228] Within each template, the data storage system will design a reasonable data structure and index mechanism. These mechanisms can ensure efficient retrieval and update of data. For example, by establishing an index, the system can quickly locate a specific data record, thereby improving the efficiency of data processing.

[0229] In addition, the system will also implement data compression and encryption strategies. Data compression can reduce storage space usage and storage costs, while data encryption can ensure data security and prevent data leakage or illegal access.

[0230] Finally, data storage systems ensure the accuracy and integrity of stored data through automated data validation and error correction mechanisms, which can automatically detect and correct errors or inconsistencies in the data, thereby improving the quality of the data.

[0231] For example, the data storage system is processing a meteorological monitoring data set that contains three modalities: temperature, humidity, and wind speed. The system first defines the corresponding data storage templates based on these three modalities. Then, it maps the temperature data to the temperature template, the humidity data to the humidity template, and the wind speed data to the wind speed template. Within each template, the system indexes the data for quick retrieval and update. At the same time, the system also compresses and encrypts the data to save storage space and ensure data security. Finally, through automated data verification and error correction mechanisms, the system ensures that the stored data is accurate and complete.

[0232] In summary, the data storage system achieves structured, efficient, and secure storage of target multi-mode meteorological monitoring data by building a flexible and scalable storage architecture, defining data storage templates, classifying and mapping data, designing data structures and indexing mechanisms, implementing data compression and encryption strategies, and automated data verification and error correction mechanisms. This approach not only improves the efficiency and quality of data processing, but also provides a solid foundation for subsequent data analysis and decision support.

[0233] Further, Figure 2 Schematic diagram of a data storage system 200 provided in an embodiment of the present application. Figure 2 The data storage system 200 shown includes a processor 210, which can call and run a computer program from a memory to implement the method in the embodiment of the present application.

[0234] Alternatively, if Figure 2 As shown, the data storage system 200 may further include a memory 230. The processor 210 may call and run a computer program from the memory 230 to implement the method in the embodiment of the present application.

[0235] The memory 230 may be a separate device independent of the processor 210 , or may be integrated into the processor 210 .

[0236] Alternatively, if Figure 2 As shown, the data storage system 200 may further include a transceiver 220, and the processor 210 may control the transceiver 220 to interact with other devices, specifically, to send information or data to other devices, or to receive information or data sent by other devices.

[0237] Optionally, the data storage system 200 may implement the corresponding processes corresponding to the storage engine or components in the storage engine (such as a processing module) or a device deployed with a storage engine in each method of the embodiments of the present application, which will not be described here for the sake of brevity.

[0238] It should be understood that the processor of the embodiment of the present application may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method embodiment can be completed by the hardware integrated logic circuit or software instructions in the processor. The above processor can be a general processor, a digital signal processor (Digital Signal Processor, DSP), an application specific integrated circuit (Application Specific Integrated Circuit, ASIC), a field programmable gate array (Field Programmable Gate Array, FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. The methods, steps and logic block diagrams disclosed in the embodiments of the present application can be implemented or executed. The general processor can be a microprocessor or the processor can also be any conventional processor. The steps of the method disclosed in the embodiment of the present application can be directly embodied as a hardware decoding processor to perform, or the hardware and software modules in the decoding processor are combined and performed. The software module can be located in a mature storage medium in the field such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory or an electrically erasable programmable memory, a register, etc. The storage medium is located in the memory, and the processor reads the information in the memory and completes the steps of the above method in combination with its hardware.

[0239] It can be understood that the memory in the embodiments of the present application can be a volatile memory or a non-volatile memory, or can include both volatile and non-volatile memories. Among them, the non-volatile memory can be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), or a flash memory. The volatile memory can be a random access memory (RAM), which is used as an external cache. By way of example and not limitation, many forms of RAM are available, such as static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link dynamic random access memory (SLDRAM), and direct RAM bus random access memory (DR RAM). It should be noted that the memory of the systems and methods described herein is intended to include, but is not limited to, these and any other suitable types of memory.

[0240] It should be understood that the above-mentioned memory is exemplary but not restrictive. For example, the memory in the embodiments of the present application may also be static random access memory (static RAM, SRAM), dynamic random access memory (dynamic RAM, DRAM), synchronous dynamic random access memory (synchronous DRAM, SDRAM), double data rate synchronous dynamic random access memory (double data rate SDRAM, DDR SDRAM), enhanced synchronous dynamic random access memory (enhanced SDRAM, ESDRAM), synchronous link dynamic random access memory (synch link DRAM, SLDRAM) and direct memory bus random access memory (Direct Rambus RAM, DR RAM), etc. That is to say, the memory in the embodiments of the present application is intended to include but not limited to these and any other suitable types of memory.

[0241] Based on the above, a computer-readable storage medium is provided, on which a computer program is stored, and the computer program implements the above method when running.

[0242] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various modifications and variations.

Claims

1. A method for storing meteorological multi-type structured data, characterized in that: Applied to a data storage system, the method comprises: Acquire target multi-mode meteorological monitoring data to be stored, and a meteorological type label corresponding to the target multi-mode meteorological monitoring data, wherein the meteorological type label is used to describe a structured meteorological attribute feature set of the target multi-mode meteorological monitoring data; Performing data splitting on the target multi-mode meteorological monitoring data to obtain a meteorological monitoring data set of the target multi-mode meteorological monitoring data, and performing enhancement processing on the meteorological monitoring data set to obtain an enhanced meteorological monitoring data set; According to the meteorological type label, the enhanced meteorological monitoring data set is subjected to X rounds of structured attribute vector recognition through a structured storage analysis algorithm to obtain a target structured meteorological attribute feature set of the target multi-mode meteorological monitoring data; Among them, during the Y-round structured attribute vector identification in the X-round structured attribute vector identification, the transition meteorological state one-hot coding of the structured storage analysis algorithm is mapped to the meteorological state conduction network obtained in the target multi-mode meteorological monitoring data, and the meteorological state one-hot coding of at least Z network units in the meteorological state conduction network is vector-integrated with the meteorological state one-hot coding of the corresponding neighboring network units, as the object to be processed of the next algorithm branch of the structured storage analysis algorithm, the Z network units include the network units corresponding to the modal identification when the target multi-mode meteorological monitoring data is split, and both X and Z are positive integers, and Y is a positive integer not greater than X.

2. The method according to claim 1, characterized in that According to the meteorological type label, the enhanced meteorological monitoring data set is subjected to X rounds of structured attribute vector recognition through a structured storage analysis algorithm to obtain a target structured meteorological attribute feature set of the target multi-mode meteorological monitoring data, including: In the u-th round of structured attribute vector recognition, a first transitional meteorological state one-hot encoding generated by a v-th algorithm branch of the structured storage analysis algorithm is obtained, wherein the first transitional meteorological state one-hot encoding is obtained based on the enhanced meteorological monitoring data set and the meteorological type label, wherein v is a positive integer, and u is a positive integer not greater than X; When the u-th round of structured attribute vector identification belongs to the Y-th round of structured attribute vector identification, the first transitional meteorological state one-hot encoding is mapped to the target multi-mode meteorological monitoring data to obtain the meteorological state transmission network, and the meteorological state one-hot encodings of the Z network units in the meteorological state transmission network are weighted with the meteorological state one-hot encodings of the corresponding neighboring network units to obtain a first meteorological state weighted one-hot encoding; According to the weighted one-hot encoding of the first meteorological state, determining the one-hot encoding to be processed of the v+1th algorithm branch of the structured storage analysis algorithm, and continuing the recognition processing to obtain the uth round of intermediate structured meteorological attribute feature set of the target multi-mode meteorological monitoring data; According to the meteorological type label, the structured attribute vector of the u+1th round is identified on the uth round intermediate structured meteorological attribute feature set by using the structured storage analysis algorithm to obtain the target structured meteorological attribute feature set; The method also includes: when the u-th round of structured attribute vector identification does not belong to the Y-th round of structured attribute vector identification, the first transitional meteorological state one-hot encoding is determined as the one-hot encoding to be processed of the v+1th algorithm branch of the structured storage analysis algorithm.

3. The method according to claim 2, characterized in that The obtaining of the first transition meteorological state one-hot encoding generated by the vth algorithm branch of the structured storage analysis algorithm includes: Obtaining the u-1th round of intermediate structured meteorological attribute feature set of the target multi-mode meteorological monitoring data; According to the meteorological type label and the u-1th round intermediate structured meteorological attribute feature set, the first transition meteorological state one-hot encoding generated by the vth algorithm branch is obtained.

4. The method according to claim 3, characterized in that The step of obtaining the u-1th round of intermediate structured meteorological attribute feature set of the target multi-mode meteorological monitoring data includes: In response to u being equal to 1, determining the enhanced meteorological monitoring data set as the u-1th round intermediate structured meteorological attribute feature set; In response to u being greater than 1, the structured attribute vector of the u-1th round is identified on the enhanced meteorological monitoring data set through the structured storage analysis algorithm according to the meteorological type label to obtain the u-1th round intermediate structured meteorological attribute feature set.

5. The method according to claim 3, characterized in that: The step of obtaining the first transitional meteorological state unique-hot encoding generated by the vth algorithm branch based on the meteorological type label and the u-1th round intermediate structured meteorological attribute feature set includes: When the vth algorithm branch is the first algorithm branch of the structured storage analysis algorithm, the meteorological type label and the u-1th round intermediate structured meteorological attribute feature set are processed by the vth algorithm branch to obtain the first transition meteorological state one-hot encoding; When the vth algorithm branch is the transition algorithm branch of the structured storage analysis algorithm, the second transition meteorological state one-hot code generated by the v-1th algorithm branch of the structured storage analysis algorithm is obtained, and the one-hot code to be processed of the vth algorithm branch is determined based on the second transition meteorological state one-hot code, and the one-hot code to be processed of the vth algorithm branch is processed by the vth algorithm branch to obtain the one-hot code of the first transition meteorological state.

6. The method according to claim 5, characterized in that The step of processing the meteorological type label and the u-1th round intermediate structured meteorological attribute feature set through the vth algorithm branch to obtain the first transitional meteorological state one-hot encoding includes: Mining the meteorological type one-hot encoding of the meteorological type label; Mining the linear meteorological state one-hot encoding of the u-1th round intermediate structured meteorological attribute feature set; The meteorological type one-hot encoding of the meteorological type label and the linear meteorological state one-hot encoding of the u-1th round intermediate structured meteorological attribute feature set are processed through the vth algorithm branch to obtain the first transition meteorological state one-hot encoding.

7. The method according to claim 5, characterized in that The determining, according to the second transitional meteorological state one-hot encoding, the to-be-processed one-hot encoding of the v-th algorithm branch comprises: Determine the second transitional meteorological state one-hot encoding as the one-hot encoding to be processed of the vth algorithm branch; Alternatively, the second transition meteorological state one-hot code is mapped to the target multi-mode meteorological monitoring data to obtain the meteorological state transmission network, and the meteorological state one-hot codes of the Z network units in the meteorological state transmission network are weighted with the meteorological state one-hot codes of the corresponding neighboring network units to obtain the second meteorological state weighted one-hot code, and the one-hot code to be processed of the vth algorithm branch is determined based on the second meteorological state weighted one-hot code.

8. The method according to claim 2 or 7, characterized in that: The meteorological state one-hot encodings of the Z network units in the meteorological state transmission network are weighted with the meteorological state one-hot encodings of the corresponding neighboring network units to obtain a weighted one-hot encoding of the target meteorological state, including: For each of the Z network units, weighting the unique hot coding of the meteorological state of the network unit and the unique hot coding of the meteorological state of the neighboring network units of the network unit as the current unique hot coding of the meteorological state of the network unit, and obtaining a meteorological state conduction network after vector integration; According to the meteorological state conduction network after the vector integration, the weighted one-hot encoding of the target meteorological state is obtained, and the weighted one-hot encoding of the target meteorological state is the weighted one-hot encoding of the first meteorological state or the weighted one-hot encoding of the second meteorological state; The step of obtaining the weighted one-hot encoding of the target meteorological state based on the meteorological state transmission network after the vector integration includes: performing feature encoding on the meteorological state transmission network after the vector integration based on the meteorological monitoring data set to obtain the weighted one-hot encoding of the target meteorological state; The step of performing feature encoding on the meteorological state conduction network after vector integration according to the meteorological monitoring data set to obtain the weighted one-hot encoding of the target meteorological state comprises: mapping the meteorological state conduction network after vector integration into the meteorological monitoring data set according to the feature encoding rule between the meteorological monitoring data set and the target multi-mode meteorological monitoring data to obtain a third meteorological state weighted one-hot encoding; obtaining the weighted one-hot encoding of the target meteorological state according to the third meteorological state weighted one-hot encoding; The method of obtaining the weighted unique hot coding of the target meteorological state based on the weighted unique hot coding of the third meteorological state includes: adjusting the dimension of the weighted unique hot coding of the third meteorological state based on the dimension information of the target transition meteorological state unique hot coding to obtain the weighted unique hot coding of the target meteorological state; wherein the dimension of the weighted unique hot coding of the target meteorological state is consistent with the dimension of the target transition meteorological state unique hot coding, and when the target transition meteorological state unique hot coding is the first transition meteorological state unique hot coding, the weighted unique hot coding of the target meteorological state is the weighted unique hot coding of the first meteorological state, and when the target transition meteorological state unique hot coding is the second transition meteorological state unique hot coding, the weighted unique hot coding of the target meteorological state is the weighted unique hot coding of the second meteorological state.

9. The method according to claim 2 or 7, characterized in that: Determining the to-be-processed one-hot coding of the target algorithm branch according to the weighted one-hot coding of the target meteorological state includes: taking the weighted one-hot coding of the target meteorological state as the to-be-processed one-hot coding of the target algorithm branch; or weighting the one-hot coding of the target transition meteorological state and the weighted one-hot coding of the target meteorological state to obtain the one-hot coding of the target algorithm branch to be processed; wherein, when the target algorithm branch is the v+1th algorithm branch, the weighted one-hot coding of the target meteorological state is the weighted one-hot coding of the first meteorological state, and the one-hot coding of the target transition meteorological state is the one-hot coding of the first transition meteorological state; when the target algorithm branch is the vth algorithm branch, the weighted one-hot coding of the target meteorological state is the weighted one-hot coding of the second meteorological state, and the one-hot coding of the target transition meteorological state is the one-hot coding of the second transition meteorological state; The step of weighting the target transition meteorological state unique hot code and the target meteorological state weighted unique hot code to obtain the target algorithm branch's to-be-processed unique hot code comprises: determining the confidence coefficients corresponding to the target transition meteorological state unique hot code and the target meteorological state weighted unique hot code respectively; weighting the target transition meteorological state unique hot code and the target meteorological state weighted unique hot code according to the confidence coefficients corresponding to the target transition meteorological state unique hot code and the target meteorological state weighted unique hot code respectively to obtain the target algorithm branch's to-be-processed object; The method of determining the confidence coefficients respectively corresponding to the target transition meteorological state one-hot encoding and the target meteorological state weighted one-hot encoding includes: processing the target transition meteorological state one-hot encoding and the target meteorological state weighted one-hot encoding through a feature focusing strategy to obtain the confidence coefficients respectively corresponding to the target transition meteorological state one-hot encoding and the target meteorological state weighted one-hot encoding.

10. A data storage system, characterized in that: The method comprises at least one processor and a memory; the memory stores computer-executable instructions; the at least one processor executes the computer-executable instructions stored in the memory, so that the at least one processor executes the method according to any one of claims 1 to 9.

Images