Early warning methods and information release platforms applicable to emergencies

Abstract

This application relates to the field of meteorological disaster data analysis technology, and in particular provides an early warning method and information release platform suitable for emergencies. By deeply exploring the key elements in past emergency monitoring information and combining historical emergency response behavior data, the embodiment of this application aims to build an early warning and response mechanism that can automatically learn and self-optimize. The establishment of this mechanism will not only significantly improve the timeliness and accuracy of disaster warnings, but also ensure that when a disaster occurs, it can quickly generate and execute a targeted warning information sending plan and a data disaster recovery backup processing plan for specific areas, thereby minimizing casualties and property losses caused by the disaster.

Description

Technical Field

[0001] The present application relates to the technical field of meteorological disaster data analysis, and in particular to an early warning method and information release platform applicable to emergencies. Background Art

[0002] In today's society, emergencies such as meteorological and natural disasters occur frequently, posing serious threats to people's lives and property. Traditional disaster warning and emergency response systems often rely on manual judgment and empirical decision-making, resulting in problems such as delayed warnings and inaccurate response strategies. Therefore, how to use advanced technologies to improve the accuracy of disaster warnings and the efficiency of emergency responses has become a pressing technical challenge. Summary of the Invention

[0003] In order to improve the above problems, this application provides an early warning method and information release platform suitable for emergencies.

[0004] The present invention provides an early warning method for emergencies, which is applied to an information release platform. The method includes:

[0005] Obtain past emergency monitoring information and past emergency plan index collection;

[0006] Invoking a disaster monitoring element mining algorithm to perform disaster monitoring element mining on the past emergency event monitoring information to obtain a disaster monitoring element vector of the past emergency event monitoring information;

[0007] Calling the emergency plan knowledge embedding algorithm to perform emergency plan knowledge embedding on the historical emergency response behavior data corresponding to each past emergency plan index in the past emergency plan index set, and obtaining the historical emergency response behavior vector of each past emergency plan index;

[0008] Debugging the disaster emergency warning algorithm based on the disaster emergency matching vector obtained by feature mapping the disaster monitoring element vector and each of the historical emergency response behavior vectors to obtain a disaster emergency warning algorithm that meets the debugging termination requirements;

[0009] Based on the disaster monitoring element mining algorithm, the emergency plan knowledge embedding algorithm and the disaster emergency warning algorithm that meets the debugging termination requirements, an emergency plan generation algorithm is debugged to generate an emergency plan for the acquired emergency event monitoring information to be processed; the emergency plan generation includes generating a plan for the targeted sending of warning information to the target area, and generating a plan for data disaster recovery and backup processing to the target area.

[0010] Optionally, the disaster monitoring element mining algorithm includes several monitoring element mining branches, and the past emergency monitoring information includes real-time monitoring data of several dimensions; calling the disaster monitoring element mining algorithm to perform disaster monitoring element mining on the past emergency monitoring information to obtain the disaster monitoring element vector of the past emergency monitoring information includes:

[0011] Obtaining the plurality of monitoring element mining branches, each monitoring element mining branch corresponding to a dimension;

[0012] Through the several monitoring element mining branches, monitoring element identification is performed on the real-time monitoring data of corresponding dimensions of the past emergency monitoring information to obtain disaster monitoring element vectors corresponding to the corresponding dimensions of the past emergency monitoring information.

[0013] Optionally, the method further includes:

[0014] In the case that the past emergency event monitoring information includes real-time monitoring data in the meteorological dimension but does not include real-time monitoring data in the non-meteorological dimension, real-time monitoring data in the non-meteorological dimension is generated based on the real-time monitoring data in the meteorological dimension to obtain real-time monitoring data in several dimensions included in the past emergency event monitoring information, wherein the real-time monitoring data in the non-meteorological dimension includes at least one of real-time monitoring data in the personnel activity dimension, real-time monitoring data in the building structure dimension, and real-time monitoring data in the production dimension.

[0015] Optionally, obtaining the plurality of monitoring element mining branches includes:

[0016] Obtain the initial monitoring element mining branches corresponding to each dimension of the emergency reference monitoring information, where each initial monitoring element mining branch corresponds to one dimension;

[0017] By mining the initial monitoring elements corresponding to each dimension, the real-time monitoring data of the corresponding dimension of the emergency reference monitoring information are respectively identified with monitoring elements, thereby obtaining the intra-class disaster monitoring element vector corresponding to the corresponding dimension of the emergency reference monitoring information;

[0018] For each intra-class disaster monitoring element vector, feature mapping is performed between the intra-class disaster monitoring element vector and the historical emergency response behavior vectors indexed by each past emergency plan to obtain a corresponding disaster emergency matching vector, and the disaster emergency warning algorithm is debugged according to each of the disaster emergency matching vectors to obtain several disaster emergency warning algorithms;

[0019] Based on the disaster monitoring element mining algorithm, the emergency plan knowledge embedding algorithm and the disaster emergency warning algorithm, debugging and obtaining several emergency plan generation algorithms;

[0020] Based on the plan matching weights of the several emergency plan generation algorithms, several monitoring element mining branches are selected from the initial monitoring element mining branches corresponding to the various dimensions.

[0021] Optionally, the method further includes:

[0022] For each past emergency plan index in the past emergency plan index set, based on the request feature pair feature of the pre-recorded request feature vector and the historical emergency response behavior data, the historical emergency response behavior data corresponding to the past emergency plan index is obtained based on the past emergency plan index.

[0023] Optionally, the method further includes:

[0024] For each past emergency plan index in the set of past emergency plan indexes, generating guidance information for activating the AI ​​algorithm based on the past emergency plan index;

[0025] A generative adversarial network is called based on the guidance information, and historical emergency response behavior data corresponding to the past emergency plan index is determined through the generative adversarial network.

[0026] Optionally, the calling of the emergency plan knowledge embedding algorithm performs emergency plan knowledge embedding on the historical emergency response behavior data corresponding to each past emergency plan index in the past emergency plan index set to obtain the historical emergency response behavior vector of each past emergency plan index, including:

[0027] For each historical emergency response behavior data corresponding to a past emergency plan index, obtaining a response event stream corresponding to the historical emergency response behavior data;

[0028] The emergency plan knowledge embedding algorithm is called to embed the emergency plan knowledge into the response event stream, and obtain the historical emergency response behavior vector indexed by the past emergency plan.

[0029] Optionally, debugging the disaster emergency warning algorithm based on the disaster emergency matching vector obtained by feature mapping the disaster monitoring element vector and each of the historical emergency response behavior vectors to obtain a disaster emergency warning algorithm that meets the debugging termination requirements includes:

[0030] Inputting the disaster monitoring element vector and each of the historical emergency response behavior vectors into the disaster emergency warning algorithm based on the linkage feature focusing rule in the emergency plan generation algorithm, and outputting the disaster emergency matching vector corresponding to each past emergency plan index through the disaster emergency warning algorithm;

[0031] The disaster emergency matching vectors corresponding to the indexes of each past emergency plan are input into the first emergency warning subnet or the second emergency warning subnet, and the algorithm variables of the disaster emergency warning algorithm are optimized according to the warning training information to obtain a disaster emergency warning algorithm that meets the debugging termination requirements.

[0032] Optionally, the step of inputting the disaster monitoring element vector and each of the historical emergency response behavior vectors into a disaster emergency warning algorithm based on a linkage feature focusing rule in the emergency plan generation algorithm, and outputting a disaster emergency matching vector corresponding to each past emergency plan index through the disaster emergency warning algorithm, includes:

[0033] The disaster monitoring element vector is used as a request feature, and each of the historical emergency response behavior vectors is used as a response feature. The vectors are input into the disaster emergency warning algorithm based on the linkage feature focusing rule in the emergency plan generation algorithm. The request feature and each of the response features are feature matched by the disaster emergency warning algorithm to obtain the disaster emergency matching vector corresponding to each past emergency plan index.

[0034] Optionally, debugging the disaster emergency warning algorithm based on the disaster emergency matching vector obtained by feature mapping the disaster monitoring element vector and each of the historical emergency response behavior vectors to obtain a disaster emergency warning algorithm that meets the debugging termination requirements includes:

[0035] Acquire training annotations configured for the past emergency event monitoring information, where the training annotations configured for the past emergency event monitoring information belong to the past emergency plan index set;

[0036] Determining, based on the training annotations configured for the past emergency event monitoring information, a priori training indication of each past emergency plan index corresponding to the past emergency event monitoring information;

[0037] Calculating the matching training confidence of the past emergency event monitoring information corresponding to each past emergency plan index based on the disaster emergency matching vector obtained by feature mapping the disaster monitoring element vector and each historical emergency response behavior vector;

[0038] Based on the prior training indication and the matching training confidence, an algorithm training error is constructed, and with the convergence of the algorithm training error as the debugging goal, the algorithm variables of the disaster emergency warning algorithm are optimized to obtain a disaster emergency warning algorithm that meets the debugging termination requirements.

[0039] Optionally, the method further includes:

[0040] When a derived emergency plan index appears in the past emergency plan index set, obtaining historical emergency response behavior data corresponding to the derived emergency plan index;

[0041] Call the constructed emergency plan generation algorithm and the historical emergency response behavior data corresponding to each past emergency plan index in the past emergency plan index set including the derived emergency plan index, generate an emergency plan for the emergency event monitoring information to be processed, and obtain an emergency plan output result. The derived emergency plan index is an emergency plan index that has not appeared in the debugging process of the emergency plan generation algorithm.

[0042] Optionally, the method further includes:

[0043] Obtain monitoring information on pending emergencies;

[0044] Calling the disaster monitoring element mining algorithm in the emergency plan generation algorithm to mine disaster monitoring elements on the emergency event monitoring information to be processed to obtain a disaster monitoring element vector;

[0045] Calling the emergency plan knowledge embedding algorithm in the emergency plan generation algorithm to perform emergency plan knowledge embedding on the historical emergency response behavior data corresponding to each past emergency plan index in the past emergency plan index set, and obtaining the historical emergency response behavior vector of each past emergency plan index;

[0046] The disaster emergency warning algorithm in the emergency plan generation algorithm is called to perform feature mapping on the disaster monitoring element vector and the historical emergency response behavior vector indexed by each past emergency plan to obtain a disaster emergency matching vector, and the current emergency plan for the emergency event monitoring information to be processed is determined based on the disaster emergency matching vector.

[0047] Optionally, the method further includes:

[0048] When a derived emergency plan index appears in the past emergency plan index set, obtaining historical emergency response behavior data corresponding to the derived emergency plan index, wherein the derived emergency plan index is an emergency plan index that has not appeared during the debugging process of the emergency plan generation algorithm;

[0049] Performing emergency plan knowledge embedding on the historical emergency response behavior data corresponding to the derived emergency plan index to obtain a historical emergency response behavior vector corresponding to the derived emergency plan index;

[0050] The calling of the disaster emergency warning algorithm in the emergency plan generation algorithm, performing feature mapping on the disaster monitoring element vector and the historical emergency response behavior vectors indexed by each past emergency plan to obtain a disaster emergency matching vector, and determining a current emergency plan for the pending emergency event monitoring information based on the disaster emergency matching vector, includes:

[0051] Through the disaster emergency warning algorithm in the emergency plan generation algorithm, feature mapping is performed on the disaster monitoring element vector and the historical emergency response behavior vectors of the past emergency plan indexes and the historical emergency response behavior vectors of the derived emergency plan indexes to obtain a disaster emergency matching vector, and the current emergency plan for the emergency event monitoring information to be processed is determined based on the disaster emergency matching vector.

[0052] An embodiment of the present application provides an information publishing platform, comprising 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 performs the above-mentioned method.

[0053] An embodiment of the present application provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program implements the above method when executed.

[0054] In the embodiments of this application, by deeply exploring key elements in past emergency monitoring information and combining it with historical emergency response behavior data, this embodiment of the application aims to build an early warning and response mechanism that can automatically learn and self-optimize. The establishment of this mechanism will not only significantly improve the timeliness and accuracy of disaster warnings, but also ensure that when a disaster occurs, early warning information targeted delivery plans and data disaster recovery and backup processing plans for specific areas can be quickly generated and executed, thereby minimizing casualties and property losses caused by disasters.

[0055] It can be seen that the embodiments of the present application are proposed to address the shortcomings of existing technologies in disaster warning and emergency response, in order to contribute new solutions to social security and stability through intelligent means. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 A flowchart of an early warning method for emergencies provided in an embodiment of the present application.

[0057] Figure 2 A schematic diagram of the structure of an information publishing platform provided in an embodiment of the present application. DETAILED DESCRIPTION

[0058] In order to better understand the above technical solution, the technical solution 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 solution of the present application, rather than limitations on the technical solution 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.

[0059] Figure 1A warning method for emergencies is shown, which is applied to an information release platform. The method includes the following steps 110 to 150.

[0060] Step 110: Obtain past emergency monitoring information and past emergency plan index sets.

[0061] The information release platform first retrieves past emergency monitoring information and emergency response plan indexes from relevant databases. This information can include historical meteorological data, geological activity records, flood level records, and more. This information is crucial for subsequent analysis and algorithm training.

[0062] Step 120: Calling a disaster monitoring element mining algorithm to perform disaster monitoring element mining on the past emergency monitoring information to obtain a disaster monitoring element vector of the past emergency monitoring information.

[0063] The information release platform then uses a disaster monitoring element mining algorithm to conduct in-depth data mining on the previously acquired emergency monitoring information. This process primarily extracts key disaster-related elements, such as wind speed, rainfall, and temperature fluctuations, from a large amount of historical data and converts this information into disaster monitoring element vectors. This step aims to extract key information that reflects disaster characteristics from complex and changing data.

[0064] Step 130: Call the emergency plan knowledge embedding algorithm to embed the emergency plan knowledge into the historical emergency response behavior data corresponding to each past emergency plan index in the past emergency plan index set, and obtain the historical emergency response behavior vector of each past emergency plan index.

[0065] The information release platform then uses an emergency plan knowledge embedding algorithm to process each historical emergency response action data point in the index set of past emergency plans. This algorithm converts key information and response actions from emergency plans into vector form, facilitating subsequent data analysis and comparison. In this way, the information release platform can learn effective strategies and response patterns from historical emergency plans.

[0066] Step 140: Debug the disaster emergency warning algorithm based on the disaster emergency matching vector obtained by feature mapping the disaster monitoring element vector and each of the historical emergency response behavior vectors to obtain a disaster emergency warning algorithm that meets the debugging termination requirements.

[0067] The information release platform uses feature mapping between disaster monitoring element vectors and historical emergency response behavior vectors to generate disaster emergency matching vectors. These matching vectors reflect the appropriate emergency response strategies for different disaster scenarios. Based on these matching vectors, the information release platform debugs the disaster emergency warning algorithm, continuously adjusting algorithm parameters until preset debugging termination requirements are met. This step ensures the accuracy and effectiveness of the warning algorithm.

[0068] Step 150: Based on the disaster monitoring element mining algorithm, the emergency plan knowledge embedding algorithm and the disaster emergency warning algorithm that meets the debugging termination requirements, an emergency plan generation algorithm is debugged to generate an emergency plan for the acquired emergency event monitoring information to be processed; the emergency plan generation includes generating a plan for the targeted sending of warning information to the target area, and generating a plan for data disaster recovery and backup processing to the target area.

[0069] After completing the above steps, the information release platform will further debug and optimize the emergency plan generation algorithm by combining the disaster monitoring element mining algorithm, the emergency plan knowledge embedding algorithm, and the debugged disaster emergency warning algorithm. This emergency plan generation algorithm can quickly generate targeted emergency plans based on real-time emergency monitoring information. These plans can include sending warnings to specific areas and implementing data disaster recovery backups. In this way, the information release platform can respond quickly to disasters, provide effective emergency measures, and mitigate their impact.

[0070] It can be seen that the information release platform can achieve rapid response and effective early warning to meteorological disasters through this series of technical means and algorithm processing, and provide the public with safer and more reliable information services.

[0071] Based on the above content, each step involved in step 110 to step 150 is described in detail below.

[0072] In step 110, past emergency monitoring information refers to data and information collected by various monitoring devices and systems during past emergencies such as meteorological and geological disasters. This data includes, but is not limited to, meteorological data such as wind speed, wind direction, rainfall, temperature, and humidity, as well as seismic wave data and flood level records. This information is crucial for understanding the causes, processes, and impacts of disasters, and serves as a crucial reference for subsequent disaster warnings and emergency response planning.

[0073] The Past Emergency Plan Index Collection contains an index of historical emergency plans developed for various emergencies. Each emergency plan index corresponds to a complete set of emergency response procedures and measures, including but not limited to evacuation routes, relief supply distribution, and medical assistance plans. These plans may have been implemented and verified in past incidents, making them valuable references for guiding future emergency response efforts. The index collection makes it easy to find and access these plans, enabling a swift response should similar incidents occur again.

[0074] In the meteorological disaster scenario, step 110 is the starting point of the entire early warning and emergency response process. In this step, the information release platform retrieves past emergency monitoring information and a collection of past emergency plan indexes from its connected databases. Past emergency monitoring information is the foundation of an effective early warning system. This information includes not only historical meteorological data such as wind speed, rainfall, and temperature, but also key data such as geological activity records and flood levels. This data, accumulated and recorded over long periods of time, provides a rich data source for the information release platform, enabling it to analyze and compare patterns and trends in disaster occurrence. Simultaneously, the information release platform also retrieves a collection of past emergency plan indexes. These plan indexes were developed and implemented in response to various emergencies in the past and contain specific measures and procedures for addressing different disaster scenarios. By obtaining these plan indexes, the information release platform can understand how similar events were historically handled, which strategies were effective, and which require improvement. This not only helps improve the accuracy of early warnings but also optimizes the emergency response process, ensuring swift and effective action when disasters occur. It can be seen that "obtaining past emergency monitoring information and past emergency plan index sets" in step 110 is a key link in the entire early warning and emergency response process. It provides a solid foundation for subsequent data analysis, early warning algorithm debugging and emergency plan generation.

[0075] In step 120, a disaster monitoring element mining algorithm is used to extract key elements related to disasters from a large amount of emergency monitoring information. This algorithm can be based on machine learning and deep learning technologies, especially neural network algorithms such as convolutional neural networks (CNN) or long short-term memory networks (LSTM). For example, a CNN model for meteorological disasters can identify patterns related to specific disasters (such as heavy rain, typhoons, etc.) by learning spatial features in historical meteorological data. These models are capable of processing multi-dimensional data, including time series data, image data, etc., thereby mining disaster monitoring elements hidden in complex data.

[0076] Disaster monitoring feature mining involves using the aforementioned algorithms to conduct in-depth analysis of emergency monitoring information to extract key information and features closely related to the occurrence and development of disasters. This process involves multiple steps, including data preprocessing, feature extraction, and pattern recognition. The goal is to filter out valuable information for disaster warning and emergency response from massive amounts of monitoring data.

[0077] Disaster monitoring feature vectors refer to a set of numerical vectors reflecting disaster characteristics, obtained through algorithmic processing during the disaster monitoring feature mining process. These vectors can be composed of a series of numerical values, each representing a specific disaster monitoring feature or characteristic. For example, in meteorological disaster monitoring, a disaster monitoring feature vector can include numerical data for multiple dimensions such as wind speed, rainfall, and air pressure, such as [wind speed: 10m / s, rainfall: 50mm, air pressure: 1010hPa]. These vectors not only provide a quantitative description of the disaster but also serve as important inputs for subsequent early warning algorithms and emergency plan generation algorithms.

[0078] In step 120 of the meteorological disaster application scenario, the information release platform utilizes advanced disaster monitoring factor mining algorithms to conduct in-depth data processing and analysis of past emergency monitoring information. This step is a core component of building an effective early warning system. First, the information release platform retrieves a large amount of past emergency monitoring information from its database. This information includes historical meteorological data, geological activity records, flood level records, and more. This data is both diverse and massive, requiring efficient algorithms for mining. Next, the platform utilizes a disaster monitoring factor mining algorithm. This algorithm can be based on complex neural network models, such as convolutional neural networks or recurrent neural networks, capable of processing multi-dimensional data and extracting key features. The algorithm meticulously analyzes each piece of monitoring information, identifying various disaster-related factors, such as a sudden increase in wind speed or abnormal changes in rainfall. Through this mining process, the algorithm transforms the raw monitoring information into disaster monitoring factor vectors. These vectors accurately describe the key characteristics of the disaster in numerical form, providing strong data support for subsequent early warning algorithms. For example, a typical disaster monitoring factor vector might include a series of numerical values ​​representing different disaster factors, such as wind speed, rainfall, and temperature. In general, disaster monitoring factor mining in step 120 is a key step in building a meteorological disaster early warning system. Through this step, the information release platform can extract valuable information from massive amounts of monitoring data, laying a solid foundation for subsequent early warning and emergency response work.

[0079] In step 140, the emergency plan knowledge embedding algorithm is an advanced machine learning technology used to convert the complex knowledge and emergency response behaviors in the emergency plan into a numerical form that can be understood by computers, namely a vector. This algorithm often uses neural network models in deep learning, such as Word2Vec, BERT or Transformer, to capture the semantic information and contextual relationships in the emergency plan text. Taking Word2Vec as an example, the model learns the association between words by training a large amount of emergency plan text data, and finally maps each word or phrase to a point in a high-dimensional vector space, so that semantically similar words are close in distance in the vector space. In this way, the key information and response strategies in the emergency plan can be expressed quantitatively, which is convenient for subsequent matching, analysis and optimization.

[0080] Historical emergency response behavior data refers to records of the emergency response measures and actions taken by relevant departments or organizations during past emergencies. This data can include response times, resource deployment, rescue measures, personnel placement, and other aspects, and is a crucial basis for evaluating and optimizing emergency response plans. By analyzing this historical data, effective emergency response strategies and models can be identified, providing valuable experience for future disaster response.

[0081] Emergency plan knowledge embedding involves using an emergency plan knowledge embedding algorithm to convert key information and response behaviors into vector form. This step aims to transform the textual content of the emergency plan into computer-operable numerical data, facilitating automated matching, analysis, and optimization. Knowledge embedding allows for more precise capture of semantic information and contextual relationships within the emergency plan, improving the intelligence of emergency response.

[0082] Historical emergency response behavior vectors are numerical feature vectors derived from processing historical emergency response behavior data using an emergency plan knowledge embedding algorithm. These vectors precisely describe the key characteristics and strategies of historical emergency response behaviors as a series of numerical values. For example, a historical emergency response behavior vector might include multiple dimensions, such as [response time: 0.8, resource allocation efficiency: 0.9, rescue measure effectiveness: 0.7, etc.], where the value of each dimension represents the performance or importance of that aspect in the historical emergency response. These vectors not only provide data support for subsequent early warning and emergency plan generation, but also help identify shortcomings and areas for improvement in emergency response.

[0083] In step 130 of the meteorological disaster application scenario, the information release platform invokes the emergency plan knowledge embedding algorithm to process the historical emergency response behavior data corresponding to each past emergency plan index in the past emergency plan index set. This step is a key component in building an intelligent emergency plan system, aiming to convert historical emergency response behaviors into a computer-readable format for deeper analysis and optimization. First, the information release platform retrieves a set of past emergency plan indexes from the database. These indexes are associated with specific historical emergency response behavior data. This data details key information such as emergency measures, resource allocation, and rescue operations taken in different disaster scenarios. Next, the platform processes this historical data using the emergency plan knowledge embedding algorithm. This algorithm deeply mines the semantic information and contextual relationships within the emergency plan text, transforming complex emergency knowledge and response behaviors into points in a high-dimensional vector space. This process not only preserves the core information of the original data but also ensures that similar emergency response strategies appear close together in the vector space, facilitating subsequent matching and comparison. Through this processing step, the information release platform obtains the historical emergency response behavior vector for each past emergency plan index. These vectors accurately describe the key characteristics and strategies of historical emergency responses in numerical form, providing strong data support for subsequent disaster emergency warnings and the generation of emergency plans. For example, when faced with a similar meteorological disaster, the platform can quickly identify the most similar historical cases and effective emergency response strategies by comparing the current situation with the vector representation of historical data, thereby guiding real-time disaster response efforts.

[0084] In step 140, feature mapping can involve mapping and associating disaster monitoring element vectors with historical emergency response behavior vectors. This mapping helps reveal the relationship between different disaster characteristics and emergency response behaviors, thereby providing data support for disaster warning and emergency plan development. For example, feature mapping can reveal a strong correlation between certain specific meteorological conditions and specific emergency response strategies.

[0085] A disaster emergency response matching vector is a numerical feature vector derived through feature mapping. It reflects the degree of match between disaster monitoring elements and historical emergency response behaviors. This vector can be composed of a series of values, each representing the degree of match between a specific disaster characteristic or emergency response behavior and historical data. For example, a disaster emergency response matching vector might be in the form [0.85, 0.7, 0.9, ...], where each value represents the degree of match between the corresponding characteristic. Such a vector helps quickly identify the similarity between the current disaster situation and historical cases and select appropriate emergency response strategies accordingly.

[0086] A disaster emergency warning algorithm that meets the commissioning termination requirements is one that, after multiple rounds of commissioning and optimization, meets pre-set performance standards. This algorithm can accurately predict the likelihood and severity of a disaster based on disaster monitoring data and issue timely warnings. Commissioning termination requirements may include multiple metrics such as warning accuracy, false alarm rate, and missed alarm rate, ensuring the algorithm's high reliability and effectiveness in practical applications.

[0087] In step 140, the information release platform uses the disaster monitoring element vectors obtained in the previous step to perform feature mapping with the historical emergency response behavior vectors to generate a disaster emergency matching vector. The core of this process is to reveal the inherent connection between disaster characteristics and emergency response. First, the platform uses advanced mathematical models and algorithms, such as support vector machines (SVMs) or neural networks in machine learning, to perform a detailed comparison and analysis of the disaster monitoring element vectors and the historical emergency response behavior vectors. These models can capture complex patterns and nonlinear relationships in the data, thereby more accurately generating disaster emergency matching vectors. Next, the platform debugs the disaster emergency warning algorithm based on the generated disaster emergency matching vectors. The goal of debugging is to ensure that the warning algorithm can provide accurate and timely warnings in various disaster situations. To this end, the platform uses a large amount of historical data for simulation testing, continuously adjusting the algorithm's parameters and model structure until its performance meets the preset debugging termination requirements. These requirements may include high accuracy, low false alarm and false negative rates, etc., to ensure the reliability and effectiveness of the algorithm in actual operation. During the debugging process, the platform also closely monitors the algorithm's real-time performance and feedback to promptly identify problems and optimize them. Ultimately, after repeated debugging and improvements, the information release platform will develop a disaster emergency warning algorithm that meets the debugging termination requirements. This algorithm will be able to provide the public with accurate and timely disaster warning information in practical applications, helping people better cope with the challenges of meteorological disasters.

[0088] In step 150, the emergency plan generation algorithm is an advanced data processing and analysis algorithm that combines machine learning and artificial intelligence technologies. It aims to quickly and automatically generate targeted emergency plans based on real-time emergency monitoring information. This algorithm can be built on complex neural network models, such as convolutional neural networks (CNNs) or long short-term memory networks (LSTMs), to capture and analyze patterns and trends in time series data. Taking a specific neural network algorithm as an example, the emergency plan generation algorithm may employ a sequence-to-sequence (Seq2Seq) model based on deep learning. This model can learn the mapping relationship from the input disaster monitoring data sequence to the output emergency plan sequence. During the training process, the algorithm makes extensive use of historical data, including past emergency monitoring information and corresponding successful emergency plans, to learn how to generate the most effective emergency response strategy based on the current disaster situation. Furthermore, the algorithm may incorporate elements of reinforcement learning, enabling the algorithm to self-optimize through continuous trial and error, improving the quality and effectiveness of the generated emergency plans. In this way, the emergency plan generation algorithm can provide decision makers with scientific and reasonable emergency plans when disasters occur, thereby minimizing the losses caused by the disaster.

[0089] Specifically, the information release platform will further debug and optimize the emergency plan generation algorithm based on the disaster monitoring element mining algorithm, emergency plan knowledge embedding algorithm, and the disaster emergency warning algorithm that meets the debugging termination requirements developed and optimized in the previous steps. First, the platform will integrate the outputs of the aforementioned algorithms—namely, the disaster monitoring element vectors, the historical emergency response behavior vectors, and the predictions of the disaster emergency warning algorithm—as input data for the emergency plan generation algorithm. This data will provide the generation algorithm with a comprehensive reference of disaster scenarios and historical emergency response strategies. Next, the platform will utilize deep learning techniques, such as sequence-to-sequence (Seq2Seq) models or generative adversarial networks (GANs), to train and optimize the emergency plan generation algorithm. These techniques help the algorithm learn disaster characteristics and emergency response patterns extracted from historical data and generate targeted emergency plans based on real-time emergency monitoring information. During the debugging process, the platform will continuously adjust the algorithm's parameters and structure to improve the accuracy and effectiveness of the generated emergency plans. This includes, but is not limited to, adjusting parameters such as the number of neural network layers, nodes, and learning rate, as well as optimizing the model's loss function and regularization term. Ultimately, after repeated debugging and refinement, the information release platform will develop an efficient emergency plan generation algorithm. Based on real-time emergency monitoring information, this algorithm can rapidly generate plans for targeted early warning information distribution and data disaster recovery and backup in targeted areas. These plans will help decision-makers respond quickly to disasters and provide effective emergency measures, thereby minimizing their impact. It's worth noting that optimizing the emergency plan generation algorithm is an ongoing process. As new historical data accumulates and algorithmic technology continues to advance, the information release platform needs to regularly update and optimize the algorithm to ensure it remains at the forefront of the industry and provides the public with safer and more reliable information services.

[0090] The embodiment of the present application achieves efficient early warning and emergency response to emergencies by comprehensively applying advanced disaster monitoring factor mining algorithms and emergency plan knowledge embedding algorithms. Specifically, by acquiring and analyzing past emergency monitoring information and emergency plan index sets, the embodiment of the present application can accurately mine the key elements of disaster monitoring and convert these elements into vector data with representational significance. Furthermore, by embedding knowledge of historical response behaviors in emergency plans to form emergency response behavior vectors, the system can learn and optimize strategies for responding to different disaster scenarios.

[0091] More importantly, this embodiment of the application uses feature mapping technology to combine disaster monitoring element vectors with historical emergency response behavior vectors to generate disaster emergency matching vectors, providing a scientific basis for debugging disaster emergency warning algorithms. A carefully debugged warning algorithm can ensure that the warning mechanism is triggered quickly and accurately when an emergency occurs.

[0092] Ultimately, by integrating the aforementioned algorithms, the present embodiment has developed an efficient emergency plan generation algorithm. This algorithm not only generates targeted warning information delivery plans for specific regions, ensuring that relevant people receive critical warning information immediately, but also automatically generates data protection and emergency backup plans based on the set data disaster recovery and backup strategies, thereby effectively reducing personal injury and property losses caused by emergencies. Overall, the implementation of the present embodiment will greatly enhance the intelligent level of disaster warning and emergency response, providing solid technical support for social security and stability.

[0093] In some preferred embodiments, the disaster monitoring element mining algorithm includes several monitoring element mining branches, and the past emergency event monitoring information includes real-time monitoring data of several dimensions; calling the disaster monitoring element mining algorithm to perform disaster monitoring element mining on the past emergency event monitoring information to obtain the disaster monitoring element vector of the past emergency event monitoring information includes: obtaining the several monitoring element mining branches, each monitoring element mining branch corresponds to a dimension; through the several monitoring element mining branches, the real-time monitoring data of the corresponding dimensions of the past emergency event monitoring information are respectively identified with monitoring elements to obtain the disaster monitoring element vector of the corresponding dimension of the past emergency event monitoring information.

[0094] In this embodiment, the information release platform utilizes a sophisticated disaster monitoring factor mining algorithm to perform disaster warning and emergency response tasks. This algorithm is not a single structure, but rather consists of multiple monitoring factor mining branches, each focused on processing specific dimensions of data from past emergency monitoring information.

[0095] Specifically, past emergency monitoring information is a multi-dimensional dataset that can include real-time monitoring data on wind speed, rainfall, temperature, air pressure, humidity, etc. This data is crucial for fully understanding the nature and impact of disaster events.

[0096] The information publishing platform first obtains these monitoring factor mining branches and ensures that each branch corresponds to a specific data dimension. For example, one branch may be responsible for analyzing wind speed data, while another branch may focus on interpreting rainfall data.

[0097] The platform then uses these specialized monitoring factor mining branches to conduct in-depth factor identification of various dimensions of past emergency monitoring information. During this process, each branch uses its specialized analysis model to extract key disaster-related information from complex data.

[0098] Ultimately, each monitoring element mining branch outputs a disaster monitoring element vector for its corresponding dimension. These vectors not only capture the core characteristics of the data but also reflect the potential connection between the data and the disaster. The information release platform integrates these vectors to form a comprehensive, multi-dimensional set of disaster monitoring element vectors, providing a rich data foundation for subsequent emergency response plan development and disaster warning.

[0099] In this way, the information release platform can more accurately capture subtle changes in disaster events, thereby more precisely predicting their potential impact. This not only improves the accuracy of early warnings but also provides a more scientific basis for decision-making in emergency response. In this way, the information release platform significantly enhances the effectiveness of disaster early warning and emergency response. Specifically, using a disaster monitoring element mining algorithm that includes multiple monitoring element mining branches, the platform can perform refined processing of data from different dimensions of past emergency monitoring information, thereby more accurately identifying key disaster-related elements. This not only enriches the data dimensions of disaster monitoring but also improves data utilization efficiency and the accuracy of early warnings. Furthermore, by integrating disaster monitoring element vectors from various dimensions, the information release platform provides more comprehensive and scientific data support for the development of subsequent emergency plans and disaster early warnings, helping to reduce casualties and property losses caused by disasters and improving overall social safety.

[0100] In some other possible embodiments, the method further includes: when the past emergency event monitoring information includes real-time monitoring data in the meteorological dimension and does not include real-time monitoring data in the non-meteorological dimension, generating real-time monitoring data in the non-meteorological dimension based on the real-time monitoring data in the meteorological dimension, and obtaining real-time monitoring data in several dimensions included in the past emergency event monitoring information, wherein the real-time monitoring data in the non-meteorological dimension includes at least one of real-time monitoring data in the personnel activity dimension, real-time monitoring data in the building structure dimension, and real-time monitoring data in the production dimension.

[0101] When processing past emergency monitoring information on information release platforms, we sometimes encounter situations where existing monitoring data focuses primarily on meteorological dimensions, such as temperature, humidity, and wind speed, but lacks real-time monitoring data on non-meteorological dimensions, such as human activity, building stability, and the operating status of production equipment. This data is equally important for comprehensively assessing disaster risks and developing emergency response plans.

[0102] To address this data gap, the information release platform has adopted an innovative data generation method. When only real-time meteorological data is available, the platform combines this data with historical data, model predictions, and machine learning algorithms to generate real-time non-meteorological data.

[0103] Specifically, the platform first analyzes historical correlations between meteorological data and various non-meteorological data. For example, strong winds may be associated with decreased structural stability, while extreme low temperatures may affect the normal operation of production equipment. Based on these correlations, the platform builds a series of data transformation and prediction models.

[0104] The information release platform then uses these models to infer corresponding non-meteorological real-time monitoring data based on current meteorological real-time monitoring data. This inferred data can include key indicators such as the density of human activity, stress distribution of building structures, and temperature and pressure of production equipment.

[0105] In this way, the information release platform can build a more comprehensive, multi-dimensional disaster monitoring dataset. This dataset not only includes traditional meteorological data but also incorporates non-meteorological data closely related to disaster risk. This provides a richer and more accurate information foundation for subsequent disaster warnings and emergency response planning.

[0106] This significantly enhances the information dissemination platform's data processing and disaster warning capabilities. While only providing real-time meteorological monitoring data, the platform can intelligently generate non-meteorological data, thereby constructing a more comprehensive and detailed dataset. This not only enriches the types of disaster monitoring data but also improves the correlation between data and the accuracy of warnings. Furthermore, this multi-dimensional data provides strong support for the development of more precise and effective emergency response plans, helping to reduce disaster risks, protect people's lives and property, and enhance society's overall disaster prevention and mitigation capabilities.

[0107] Under some optional technical ideas, the obtaining of the several monitoring element mining branches includes: obtaining the initial monitoring element mining branches corresponding to the emergency reference monitoring information and each dimension, each initial monitoring element mining branch corresponding to a dimension; through the initial monitoring element mining branches corresponding to each dimension, the real-time monitoring data of the corresponding dimension of the emergency reference monitoring information are respectively identified to obtain the intra-class disaster monitoring element vector corresponding to the corresponding dimension of the emergency reference monitoring information; for each intra-class disaster monitoring element vector, feature mapping is performed based on the intra-class disaster monitoring element vector and the historical emergency response behavior vector indexed by each past emergency plan to obtain the corresponding disaster emergency matching vector, and the disaster emergency warning algorithm is debugged according to each of the disaster emergency matching vectors to obtain several disaster emergency warning algorithms; based on the disaster monitoring element mining algorithm, the emergency plan knowledge embedding algorithm and the disaster emergency warning algorithm, several emergency plan generation algorithms are debugged; based on the plan matching weights of the several emergency plan generation algorithms, several monitoring element mining branches are selected from the initial monitoring element mining branches corresponding to each dimension.

[0108] When building its disaster monitoring and emergency response system, the information publishing platform adopted a highly flexible and intelligent approach to optimize the selection process for its monitoring factor mining branches. This process is key to ensuring the system can accurately and efficiently respond to various emergencies.

[0109] First, the platform acquires a set of reference monitoring information for emergencies. This information includes real-time monitoring data from multiple dimensions, such as meteorology, geology, and the environment. Simultaneously, the platform also maintains a set of initial monitoring factor mining branches corresponding to these dimensions. These branches are pre-designed, each dedicated to identifying and analyzing monitoring factors in a specific data dimension.

[0110] Next, the information release platform uses these initial monitoring elements to mine branches and conduct in-depth processing of the real-time monitoring data for the corresponding dimensions in the emergency reference monitoring information. Through this step, the platform extracts key monitoring elements from each dimension of data and converts them into within-class disaster monitoring element vectors. These vectors not only reflect the core characteristics of the data but also provide important reference information for subsequent emergency response.

[0111] To further enhance the system's emergency response capabilities, the information release platform also performs feature mapping between these intra-class disaster monitoring element vectors and historical emergency response behavior vectors indexed by previous emergency response plans. Through this mapping process, the platform generates corresponding disaster emergency matching vectors, which reveal the inherent connections between different disaster monitoring elements and emergency response behaviors.

[0112] Based on these disaster emergency matching vectors, the information release platform will debug the disaster emergency warning algorithm to obtain several emergency warning algorithms for different disaster scenarios. These algorithms can quickly and accurately issue warning signals when a disaster occurs, providing relevant personnel with valuable response time.

[0113] The platform then combines the disaster monitoring element mining algorithm, the emergency plan knowledge embedding algorithm, and the debugged disaster emergency warning algorithm to further debug and generate several emergency plan generation algorithms. These algorithms can dynamically generate the most effective emergency plans based on real-time monitoring data and disaster conditions.

[0114] Finally, the information release platform carefully selects several monitoring element mining branches from the initial set of monitoring element mining branches based on the plan matching weights of these emergency plan generation algorithms. These branches will play a key role in subsequent disaster monitoring and emergency response processes, ensuring that the platform can respond to various emergencies quickly and accurately.

[0115] In this way, the information release platform has significantly enhanced the intelligence of its disaster monitoring and emergency response capabilities. By flexibly selecting and optimizing monitoring factor mining branches, the platform can more accurately capture and analyze real-time monitoring data from different dimensions, thereby more accurately predicting and assessing disaster risks. Furthermore, by combining emergency plan knowledge embedding algorithms with disaster emergency warning algorithms, the platform can dynamically generate efficient emergency plans and respond quickly to disasters. This not only improves the accuracy and timeliness of disaster warnings, but also provides relevant personnel with more scientific and effective emergency response guidance, helping to reduce disaster losses and protect people's lives and property.

[0116] In some other possible embodiments, the method further includes: for each past emergency plan index in the past emergency plan index set, according to a request feature pair feature of a pre-recorded request feature vector and the historical emergency response behavior data, based on the past emergency plan index, obtaining the historical emergency response behavior data corresponding to the past emergency plan index.

[0117] When handling emergency response plan-related tasks, the information release platform will adopt a series of refined data processing measures. In this specific embodiment, the platform focuses on how to extract and utilize historical emergency response behavior data from past emergency response plan execution experiences to optimize future emergency response strategies.

[0118] First, the information release platform maintains a collection of past emergency response plan indexes. Each index in this collection corresponds to an emergency response plan that has been implemented historically. These indexes are not just simple numbers or labels; they are also associated with rich emergency response context information.

[0119] Next, for each past emergency plan index in the collection, the platform performs a critical data extraction step. The core of this step is to accurately locate and retrieve the historical emergency response behavior data corresponding to that index by using pre-recorded request feature vectors and request feature pairs of historical emergency response behavior data.

[0120] In the embodiments of this application, a "request feature vector" is a technical term that refers to a numerical vector describing the key attributes of an emergency request. These attributes may include the time, location, disaster type, and scale of the request. A "request feature pair from historical emergency response data" refers to a data pair in historical emergency response data that matches or correlates with a specific request feature vector.

[0121] By comparing the current request feature vector with the request feature pairs stored in historical data, the information release platform can accurately locate historical emergency response behavior data under similar circumstances. This data includes valuable information such as the response strategy at the time, resource allocation, and response effectiveness evaluation.

[0122] Once this historical emergency response data is acquired, the information release platform can conduct in-depth analysis and learning. This not only helps the platform better understand the optimal emergency response strategies under different disaster scenarios, but also provides strong data support for the formulation and adjustment of future emergency plans.

[0123] This allows the information release platform to efficiently leverage historical emergency response data, improving the intelligence and accuracy of its response. By comparing request feature vectors with historical data, the platform can accurately locate and capture emergency response experience from similar scenarios, providing valuable reference for future emergency responses. This not only helps shorten response times and improve response efficiency, but also ensures optimal resource allocation, minimizing disaster losses. Overall, this solution significantly enhances the information release platform's capabilities and flexibility in disaster response.

[0124] In other possible embodiments, the method further includes: for each past emergency plan index in the past emergency plan index set, generating guidance information for activating the AI ​​algorithm based on the past emergency plan index; calling a generative adversarial network based on the guidance information, and determining the historical emergency response behavior data corresponding to the past emergency plan index through the generative adversarial network.

[0125] When processing emergency plan-related data, the information release platform adopts an innovative method to obtain and utilize historical emergency response behavior data in order to further improve the intelligence and accuracy of data processing.

[0126] First, for each past emergency plan index in the past emergency plan index collection, the information release platform generates specific guidance information based on this index. This guidance information is tailored to the AI ​​algorithm, designed to activate and guide it to more accurately process relevant data. Specifically, the guidance information may include identification of emergency plan types, extraction of key features of emergency response behaviors, and specific requirements for data screening and processing.

[0127] Next, the information publishing platform uses this guidance to invoke a generative adversarial network (GAN). A GAN is a deep learning model composed of two components: a generator and a discriminator. It uses adversarial training to improve data generation and discrimination capabilities. In this scenario, the GAN is used to determine the historical emergency response behavior data corresponding to the index of a past emergency plan.

[0128] Specifically, the generator attempts to generate emergency response behavior data related to a specific emergency plan based on guidance information, while the discriminator is responsible for determining whether this data truly reflects historical emergency response behavior. Through continuous adversarial training, the generative adversarial network gradually "learns" how to generate more realistic and accurate historical emergency response behavior data.

[0129] Ultimately, the information release platform will use the historical emergency response behavior data identified through the generative adversarial network for subsequent emergency plan development, optimization, and evaluation. This data not only contains a wealth of emergency response experience but has also been further refined and optimized through deep learning models, making it more valuable and practical.

[0130] By introducing generative adversarial networks and customized guidance, the information release platform is able to more intelligently and accurately acquire and utilize historical emergency response behavior data. This approach not only improves the automation level of data processing but also enhances the authenticity and reliability of the data. Furthermore, through the training and optimization of deep learning models, the platform can more effectively extract valuable emergency response experience and knowledge from historical data, providing more scientific and accurate decision-making support for future disaster response. Overall, this technical solution has significantly enhanced the intelligence and practical capabilities of the information release platform in the field of disaster response.

[0131] It is worth mentioning that calling a generative adversarial network based on the guidance information and determining the historical emergency response behavior data corresponding to the past emergency plan index through the generative adversarial network includes: constructing a conditional generative adversarial network according to the guidance information, wherein the generator of the conditional generative adversarial network receives the guidance information as a conditional input and generates candidate emergency response behavior data associated with the past emergency plan index; passing the generated candidate emergency response behavior data to the discriminator of the conditional generative adversarial network, and providing the discriminator with the real historical emergency response behavior data corresponding to the past emergency plan index as a reference; comparing and discriminating the candidate emergency response behavior data and the real historical emergency response behavior data through the discriminator, and outputting a discrimination result, which indicates the similarity between the candidate data and the real data; adjusting the parameters of the generator based on the feedback of the discriminator, and optimizing the performance of the generative adversarial network through multiple iterative training until the generated candidate emergency response behavior data reaches a preset authenticity threshold; after the generative adversarial network training is completed, using the trained generator to generate corresponding high-authenticity historical emergency response behavior data according to the guidance information of the past emergency plan index.

[0132] In some other embodiments, the emergency plan knowledge embedding algorithm is called to perform emergency plan knowledge embedding on the historical emergency response behavior data corresponding to each past emergency plan index in the past emergency plan index set to obtain the historical emergency response behavior vector of each past emergency plan index, including: for the historical emergency response behavior data corresponding to each past emergency plan index, obtaining the response event stream corresponding to the historical emergency response behavior data; calling the emergency plan knowledge embedding algorithm to perform emergency plan knowledge embedding on the response event stream to obtain the historical emergency response behavior vector of the past emergency plan index.

[0133] During its operation, the information release platform employs an advanced algorithm called "Emergency Plan Knowledge Embedding" to more efficiently process and utilize historical emergency response data. This algorithm converts complex emergency response data into vector form, facilitating subsequent data analysis and pattern recognition.

[0134] Specifically, once the information publishing platform obtains a collection of past emergency plan indexes, it processes the historical emergency response behavior data corresponding to each past emergency plan index in the collection. First, the platform extracts the response event stream from the historical emergency response behavior data corresponding to each past emergency plan index. This "response event stream" refers to a sequential record of a series of events and actions that occurred during the emergency response process, encompassing a wealth of emergency response information and knowledge.

[0135] Next, the information release platform invokes the emergency plan knowledge embedding algorithm to embed these response event streams. The core of this algorithm lies in its ability to map each event or action in the response event stream to a point (i.e., a vector) in a high-dimensional vector space, while preserving the correlation and semantic information between events. In this way, emergency response behavior data, which was originally complex and difficult to directly compare, is transformed into a series of computable vectors.

[0136] After processing the emergency plan knowledge embedding algorithm, the information release platform will obtain the historical emergency response behavior vector corresponding to each past emergency plan index. These vectors not only contain the key information of the original emergency response behavior, but also exist in a form that is easier for computers to process and analyze.

[0137] By introducing an emergency plan knowledge embedding algorithm, the information release platform can more effectively process and utilize historical emergency response behavior data. This algorithm converts complex emergency response data into vector form, simplifying the data representation while preserving the inter-data correlation and semantic information. This enables the platform to more accurately analyze and identify emergency response patterns, providing strong data support for the development and optimization of future emergency plans. Overall, this technical solution has significantly improved the information release platform's capabilities and efficiency in processing emergency response data.

[0138] In some independent embodiments, an emergency plan knowledge embedding algorithm is called to embed emergency plan knowledge into the response event stream to obtain the historical emergency response behavior vector of the past emergency plan index, including: constructing an emergency plan knowledge embedding model including a multi-layer neural network structure, which learns the inherent laws and characteristics of emergency response behavior data through training; passing the response event stream as input data to the emergency plan knowledge embedding model, wherein each event is represented as an initial vector in the model; performing nonlinear transformation and feature extraction on the initial vector through the neural network layer in the emergency plan knowledge embedding model, and gradually encoding the semantic information and contextual relationship in the event stream into the vector; at the last layer of the model, outputting a fully transformed and encoded vector, which is the historical emergency response behavior vector of the past emergency plan index, which integrates the key features and contextual information of the emergency response behavior and can be used for subsequent analysis, comparison, retrieval and other operations.

[0139] Specifically, the information release platform employs a sophisticated technical approach to process historical emergency response behavior data corresponding to past emergency plan indexes. The core of this approach is to invoke an emergency plan knowledge embedding algorithm to deeply process the response event stream, thereby generating historical emergency response behavior vectors rich in information and features. First, the information release platform constructs an emergency plan knowledge embedding model comprised of a multi-layer neural network structure. This model is not a simple structure; instead, it learns the inherent patterns and characteristics of emergency response behavior data through extensive training data. During training, the model gradually understands the complex relationships and patterns within emergency response. Next, the platform passes the response event stream as input to this embedding model. Here, "response event stream" refers to a chronological series of emergency response events. Each event is represented in the model as an initial vector, which serves as the starting point for the model to understand and process the data. Subsequently, the platform applies nonlinear transformations and feature extraction to these initial vectors through the neural network layers within the emergency plan knowledge embedding model. This step is crucial for data processing, as it gradually encodes the semantic information and contextual relationships within the event stream into the vectors. This means that as the data flows through the neural network, each vector incorporates more information and features, becoming richer and more accurate. Finally, at the final layer of the model, the platform outputs fully transformed and encoded vectors. These vectors not only contain key information about the original emergency response event but also incorporate rich contextual relationships and semantic features. These vectors, known as "historical emergency response behavior vectors indexed by past emergency plans," can be used for subsequent data analysis, comparison, retrieval, and pattern recognition.

[0140] By constructing an emergency plan knowledge embedding model with a multi-layer neural network structure, the information release platform can deeply explore the inherent patterns and characteristics of emergency response behavior data. This technical solution not only improves the accuracy and efficiency of data processing, but also provides strong data support for subsequent emergency response analysis and decision-making. Through nonlinear transformation and feature extraction, the platform can more accurately capture key information and contextual relationships in emergency response events, thereby generating feature-rich historical emergency response behavior vectors. These vectors can be used in a variety of complex data analysis and pattern recognition tasks in the future, significantly enhancing the intelligence level of the information release platform in the field of emergency management and response.

[0141] Under some preferred design ideas, the disaster emergency warning algorithm is debugged based on the disaster emergency matching vector obtained by feature mapping the disaster monitoring element vector and each historical emergency response behavior vector to obtain a disaster emergency warning algorithm that meets the debugging termination requirements, including: inputting the disaster monitoring element vector and each historical emergency response behavior vector into the disaster emergency warning algorithm based on the linkage feature focusing rule in the emergency plan generation algorithm, and outputting the disaster emergency matching vector corresponding to each past emergency plan index through the disaster emergency warning algorithm; inputting the disaster emergency matching vector corresponding to each past emergency plan index into the first emergency warning subnet or the second emergency warning subnet, optimizing the algorithm variables of the disaster emergency warning algorithm according to the warning training information, and obtaining the disaster emergency warning algorithm that meets the debugging termination requirements.

[0142] The information release platform follows a series of refined design principles when building and optimizing its disaster emergency warning algorithm. During this process, the platform leverages data from disaster monitoring element vectors and historical emergency response behavior vectors, generating disaster emergency matching vectors through feature mapping to further debug the warning algorithm.

[0143] First, the information release platform uses the disaster monitoring element vector and the historical emergency response action vectors as input data and passes them to the disaster emergency warning algorithm within the emergency plan generation algorithm. This algorithm follows specific linkage feature focusing rules to identify and extract key features from the input data.

[0144] After receiving this data, the disaster emergency warning algorithm performs a complex feature mapping operation, matching and integrating the disaster monitoring elements with the characteristics of historical emergency response behavior vectors. This generates disaster response matching vectors corresponding to each past emergency plan index. These vectors not only reflect the specific circumstances of the disaster but also incorporate historical emergency response experience from similar disasters.

[0145] Next, the information release platform inputs the generated disaster emergency matching vector into the first or second emergency warning subnet. These two subnets are key components of the disaster emergency warning algorithm, responsible for training and optimizing warnings based on the input matching vector.

[0146] Within the subnet, the platform optimizes and adjusts the algorithm variables of the disaster emergency warning algorithm based on pre-set warning training information. This process is iterative, and the platform will continuously try different variable combinations to find the best warning effect.

[0147] Ultimately, when the performance of the disaster emergency warning algorithm reaches the preset debugging termination requirements, the information release platform stops the debugging process and saves the optimized algorithm parameters. At this point, the platform has obtained a disaster emergency warning algorithm that meets the requirements and can quickly and accurately generate warning information when future disasters occur.

[0148] By optimizing the disaster emergency warning algorithm based on the feature mapping results of disaster monitoring element vectors and historical emergency response behavior vectors, the information release platform can significantly improve the accuracy and response speed of the warning algorithm. This optimization method not only integrates real-time disaster monitoring data and historical emergency response experience, but also continuously adjusts algorithm variables through iterative training, making the warning algorithm more adaptable to complex and changing disaster situations. Ultimately, this technical solution will enhance the information release platform's disaster warning capabilities and provide the public with more timely and accurate disaster warning information.

[0149] In the next step, the disaster monitoring element vector and each of the historical emergency response behavior vectors are input into the disaster emergency warning algorithm based on the linkage feature focusing rule in the emergency plan generation algorithm, and the disaster emergency matching vector corresponding to each past emergency plan index is output through the disaster emergency warning algorithm, including: taking the disaster monitoring element vector as the request feature and each of the historical emergency response behavior vectors as the response feature, inputting them into the disaster emergency warning algorithm based on the linkage feature focusing rule in the emergency plan generation algorithm, and performing feature matching on the request feature and each of the response features through the disaster emergency warning algorithm to obtain the disaster emergency matching vector corresponding to each past emergency plan index.

[0150] During the disaster emergency warning process, the information release platform takes a series of precise actions to ensure the accuracy and timeliness of the warning. During this process, the platform uses the disaster monitoring element vector and the historical emergency response behavior vector to generate a disaster emergency matching vector using the disaster emergency warning algorithm.

[0151] First, the information release platform uses the disaster monitoring element vector as a request feature. These request features reflect the current disaster's real-time status, including but not limited to key information such as disaster type, intensity, and impact area. These features are crucial for triggering emergency responses and developing contingency plans.

[0152] At the same time, the platform also uses each historical emergency response behavior vector as a response feature. Response features contain past disaster response experience and knowledge, and they represent effective emergency response behaviors and strategies in different disaster scenarios.

[0153] Next, these request and response features are fed into the disaster emergency warning algorithm within the emergency plan generation algorithm. This algorithm, designed based on the principle of linked feature focus, intelligently identifies correlations and similarities between request and response features.

[0154] The disaster emergency warning algorithm performs a detailed feature matching operation on the request features and each response feature. During this process, the algorithm analyzes the fit between the current disaster situation and the emergency response behaviors of similar disasters in history, thereby determining which historical experience is most valuable for the current disaster response.

[0155] Ultimately, after feature matching, the disaster emergency warning algorithm outputs disaster emergency response matching vectors corresponding to each past emergency plan index. These vectors not only comprehensively consider the actual situation of the current disaster but also incorporate effective response strategies for similar disasters in history, providing valuable data support for subsequent emergency response decisions.

[0156] By using disaster monitoring element vectors as request features and historical emergency response behavior vectors as response features, and inputting these into a disaster emergency warning algorithm based on linked feature focusing rules for feature matching, the information release platform can efficiently generate disaster emergency matching vectors. This technical solution not only improves the accuracy of warnings but also effectively utilizes historical emergency response experience, enabling the information release platform to make more scientific and timely response decisions when faced with disasters, thereby minimizing the losses caused by disasters.

[0157] In some independent embodiments, the request feature is feature matched with each response feature by the disaster emergency warning algorithm to obtain a disaster emergency matching vector corresponding to each past emergency plan index, including: constructing a feature matching module in the disaster emergency warning algorithm, which uses deep learning technology to capture the complex relationship between the request feature and the response feature; inputting the request feature, i.e., the disaster monitoring element vector, into the feature matching module and encoding it to generate a request feature vector; then, inputting each response feature, i.e., the historical emergency response behavior vector, into the feature matching module in turn and performing a matching calculation with the request feature vector one by one; during the matching calculation process, the feature matching module evaluates the similarity between the request feature vector and each response feature vector, and this similarity calculation can be based on cosine similarity, Euclidean distance or other advanced vector similarity measurement methods; based on the similarity calculation results, a disaster emergency matching vector is assigned to each past emergency plan index, which numerically reflects the degree of matching between the current disaster situation and the past emergency plan, thereby providing a means for the information release platform to quantitatively evaluate the applicability of the emergency plan.

[0158] Specifically, a key technology used by the information release platform in performing disaster warning tasks is to perform feature matching between disaster monitoring data and historical emergency response data using a disaster emergency warning algorithm, thereby quickly identifying the emergency response plan that best matches the current disaster situation. During this process, the platform implements a series of refined operations. First, the information release platform builds a dedicated feature matching module into the disaster emergency warning algorithm. This module leverages deep learning technology to deeply explore the deep connections and complex relationships between request features (i.e., disaster monitoring feature vectors) and response features (i.e., historical emergency response behavior vectors). The application of deep learning technology makes the feature matching process more intelligent and precise. Next, the platform inputs the disaster monitoring feature vectors, also known as request features, into the feature matching module. Here, the request features are encoded into a high-dimensional vector representation, the request feature vector. This vector comprehensively and accurately captures all aspects of the current disaster. Subsequently, the information release platform sequentially feeds each historical emergency response behavior vector, also known as the response features, into the feature matching module. These response features are also converted into vector form and matched one-to-one with the request feature vector. During the matching calculation process, the feature matching module uses a specific algorithm to evaluate the similarity between the request feature vector and each response feature vector. This similarity calculation can rely on a variety of methods, including but not limited to advanced vector similarity measurement techniques such as cosine similarity and Euclidean distance. These techniques can accurately quantify the proximity between two vectors. Finally, based on the similarity calculation results, the information release platform assigns a disaster emergency response matching vector to each past emergency plan index. This vector not only numerically reflects the degree of match between the current disaster situation and past emergency plans, but more importantly, it provides the platform with a scientific and objective quantitative assessment method, allowing it to quickly find the most appropriate emergency plan when faced with a disaster.

[0159] By building a feature matching module and leveraging deep learning technology to match request and response features, the information release platform can efficiently generate a corresponding disaster response matching vector for each past emergency plan index. This not only improves the accuracy of emergency plan selection but also significantly shortens emergency response decision-making time. Furthermore, by quantitatively assessing the applicability of emergency plans, the platform enables more scientific and orderly emergency management and resource allocation when disasters occur, thereby minimizing losses and protecting public life and property.

[0160] In other preferred embodiments, the disaster emergency warning algorithm is debugged based on the disaster emergency matching vector obtained by feature mapping the disaster monitoring element vector and each of the historical emergency response behavior vectors to obtain a disaster emergency warning algorithm that meets the debugging termination requirements, including: obtaining training annotations configured for the past emergency event monitoring information, the training annotations configured for the past emergency event monitoring information belong to the past emergency plan index set; determining the prior training indications of the past emergency event monitoring information corresponding to each past emergency plan index based on the training annotations configured for the past emergency event monitoring information; calculating the matching training confidence of the past emergency event monitoring information corresponding to each past emergency plan index based on the disaster emergency matching vector obtained by feature mapping the disaster monitoring element vector and each of the historical emergency response behavior vectors; constructing an algorithm training error based on the prior training indications and the matching training confidence, and optimizing the algorithm variables of the disaster emergency warning algorithm with the convergence of the algorithm training error as the debugging goal to obtain a disaster emergency warning algorithm that meets the debugging termination requirements.

[0161] To improve the accuracy and response speed of the disaster emergency warning algorithm, the information release platform will implement a series of optimization measures. A key step is to debug the disaster emergency warning algorithm based on the feature mapping results between disaster monitoring element vectors and historical emergency response behavior vectors.

[0162] First, the information publishing platform obtains training annotations for past emergency monitoring information. These training annotations are associated with a collection of past emergency response plan indexes and provide the platform with additional explanations and clarifications about the emergency monitoring information. These annotations can include information on the type of disaster, severity, impact area, and effectiveness of the emergency response.

[0163] Next, the platform uses these training annotations to determine prior training indicators for each past emergency response plan index based on past emergency monitoring information. These prior training indicators, derived from historical data and expert knowledge, provide the algorithm with preliminary judgments about the applicability of emergency response plans. These indicators help the algorithm converge more quickly to the correct solution in subsequent calculations.

[0164] The information release platform then calculates the confidence level of the matching training between the past emergency monitoring information and the index of each historical emergency plan based on the feature mapping results of the disaster monitoring element vector and the historical emergency response behavior vector. This confidence level reflects the degree of match between the current disaster situation and the emergency response plans for similar disasters in history. By comparing the confidence levels of different emergency response plans, the platform can determine which plans are most likely to be effective in the current disaster.

[0165] Finally, the platform constructs an algorithm training error based on prior training indicators and matching training confidence. With convergence of this training error as the debugging goal, it optimizes the algorithmic variables of the disaster emergency warning algorithm. By continuously adjusting the algorithmic variables, the platform can gradually reduce the training error, thereby improving the accuracy and response speed of the warning algorithm. When the training error converges below a preset threshold, the platform deems the disaster emergency warning algorithm to have met the debugging termination requirements.

[0166] In this way, by obtaining training annotations from past emergency monitoring information, determining prior training instructions, calculating matching training confidence, and constructing and optimizing disaster emergency warning algorithms based on this information, the information release platform can significantly improve the accuracy and response speed of the warning algorithm. This optimization method not only takes into account historical data and expert knowledge, but also accurately matches the relationship between the current disaster and past emergency response plans through feature mapping and confidence calculation. Ultimately, this technical solution will enable the information release platform to generate warning information more quickly and accurately when faced with disasters, thereby effectively improving the efficiency and effectiveness of emergency response.

[0167] In some preferred embodiments, the method also includes: when a derived emergency plan index appears in the past emergency plan index set, obtaining historical emergency response behavior data corresponding to the derived emergency plan index; calling the constructed emergency plan generation algorithm and the historical emergency response behavior data corresponding to each past emergency plan index in the past emergency plan index set including the derived emergency plan index, performing emergency plan generation on the emergency event monitoring information to be processed, and obtaining an emergency plan output result, wherein the derived emergency plan index is an emergency plan index that has not appeared during the debugging process of the emergency plan generation algorithm.

[0168] When executing emergency plan generation tasks, information release platforms face a variety of complex and ever-changing emergencies. In some cases, new emergency plan indexes, known as derived emergency plan indexes, may appear. When this occurs, the information release platform implements a series of countermeasures to ensure the effective generation of emergency plans.

[0169] First, once a derivative emergency plan index is detected within the set of past emergency plan indexes, the information publishing platform immediately takes action to retrieve the historical emergency response behavior data corresponding to that derivative emergency plan index. This data represents valuable experience accumulated from the platform's previous handling of similar incidents and is crucial for understanding and responding to new derivative emergency plans.

[0170] Next, the information release platform invokes the established emergency plan generation algorithm and takes as input the historical emergency response behavior data corresponding to all past emergency plan indices, including the derived emergency plan index. The goal of this step is to leverage historical data and advanced algorithms to conduct in-depth analysis of the pending emergency monitoring information and generate effective emergency plans.

[0171] During this process, the emergency plan generation algorithm considers various factors, including but not limited to the type, severity, and scope of the emergency, as well as the effectiveness of historical emergency responses. Through complex calculations and analysis, the algorithm generates a detailed emergency plan output tailored to the current emergency.

[0172] It's worth noting that the derived emergency plan index here was not present during the debugging of the emergency plan generation algorithm. This means the platform can quickly adapt and respond effectively to new challenges. This flexibility and adaptability are essential capabilities for information release platforms when handling emergencies.

[0173] By acquiring historical emergency response behavior data corresponding to the derived emergency plan index and invoking the emergency plan generation algorithm to generate emergency plan outputs tailored to the current emergency, the information release platform demonstrates significant flexibility and adaptability. This capability enables the platform to respond quickly and develop effective response strategies in the face of unprecedented challenges. This not only improves the efficiency and accuracy of the information release platform's emergency response capabilities but also provides strong technical support for protecting public life and property.

[0174] In other embodiments, the method also includes: obtaining monitoring information of pending emergency events; calling the disaster monitoring element mining algorithm in the emergency plan generation algorithm to perform disaster monitoring element mining on the pending emergency event monitoring information to obtain a disaster monitoring element vector; calling the emergency plan knowledge embedding algorithm in the emergency plan generation algorithm to perform emergency plan knowledge embedding on the historical emergency response behavior data corresponding to each past emergency plan index in the past emergency plan index set to obtain a historical emergency response behavior vector of each past emergency plan index; calling the disaster emergency warning algorithm in the emergency plan generation algorithm to perform feature mapping on the disaster monitoring element vector and the historical emergency response behavior vector of each past emergency plan index to obtain a disaster emergency matching vector, and determining the current emergency plan for the pending emergency event monitoring information based on the disaster emergency matching vector.

[0175] When handling emergencies, the information release platform needs to quickly and accurately generate corresponding emergency plans. To achieve this goal, the platform will perform a series of refined operations.

[0176] First, the information release platform obtains the monitoring information of the emergency to be processed. This information may come from various sensors, monitoring systems, or other information sources, and contains key data such as the type, location, time, and scope of impact of the emergency.

[0177] Next, the platform invokes the disaster monitoring element mining algorithm within the emergency plan generation algorithm to conduct an in-depth analysis of the emergency monitoring information. This step aims to extract core disaster-related elements, such as disaster type, intensity, and development trend, from the massive amount of monitoring information and convert this information into a disaster monitoring element vector. This vector concisely and comprehensively describes the key characteristics of the current emergency.

[0178] The information release platform then uses the emergency plan knowledge embedding algorithm within the emergency plan generation algorithm to process the historical emergency response behavior data corresponding to each past emergency plan index in the past emergency plan index set. This process aims to convert this historical data into a computer-friendly format, known as historical emergency response behavior vectors. These vectors not only contain the specific content of the emergency plan but also reflect the actual effectiveness and feedback during its implementation.

[0179] Next, the platform invokes the Disaster Emergency Warning Algorithm within the emergency plan generation algorithm. This algorithm's core function is to perform feature mapping between disaster monitoring element vectors and historical emergency response behavior vectors indexed by previous emergency plans. Through complex mathematical calculations and pattern recognition techniques, the algorithm identifies an emergency plan that best matches the current disaster situation. The result of this matching process is a Disaster Emergency Matching Vector, which quantitatively represents the degree of compatibility between different emergency plans and the current disaster.

[0180] Finally, based on the results of the disaster emergency matching vector, the information release platform determines the current emergency response plan for the pending emergency monitoring information. This plan is not only based on historical data and expert knowledge, but also fully considers the actual situation and development trends of the current disaster, thus ensuring the targeted and effective response plan.

[0181] In this way, by acquiring monitoring information on pending emergencies, mining disaster monitoring elements, embedding emergency plan knowledge, and performing feature mapping and matching, the information release platform can quickly and accurately determine the emergency plan that best matches the current emergency. This process not only improves the automation level of emergency plan generation but also significantly reduces the time cost of human intervention and decision-making, thereby ensuring the timeliness and effectiveness of emergency response. Furthermore, this technical solution fully leverages historical data and expert knowledge, improving the scientific nature and reliability of emergency plans and providing strong technical support for protecting public life and property.

[0182] In some further embodiments, the method further includes: when a derived emergency plan index appears in the past emergency plan index set, obtaining historical emergency response behavior data corresponding to the derived emergency plan index, the derived emergency plan index being an emergency plan index that has not appeared during the debugging process of the emergency plan generation algorithm; performing emergency plan knowledge embedding on the historical emergency response behavior data corresponding to the derived emergency plan index to obtain a historical emergency response behavior vector corresponding to the derived emergency plan index.

[0183] Then the disaster emergency warning algorithm in the emergency plan generation algorithm is called to perform feature mapping on the disaster monitoring element vector and the historical emergency response behavior vectors of the past emergency plan indexes to obtain a disaster emergency matching vector, and determine the current emergency plan for the pending emergency event monitoring information based on the disaster emergency matching vector, including: through the disaster emergency warning algorithm in the emergency plan generation algorithm, the disaster monitoring element vector and the historical emergency response behavior vectors of the past emergency plan indexes and the historical emergency response behavior vectors of the derived emergency plan indexes to obtain a disaster emergency matching vector, and determine the current emergency plan for the pending emergency event monitoring information based on the disaster emergency matching vector.

[0184] When handling emergencies, information publishing platforms often face new and unexpected emergency situations, which may require the platform to quickly adapt and develop corresponding emergency plans. In these situations, derived emergency plan indexes may appear, that is, emergency plan indexes that were not found during the initial debugging of the emergency plan generation algorithm.

[0185] When such a derivative emergency plan index appears in the set of past emergency plan indexes, the information release platform takes immediate action. First, the platform retrieves the historical emergency response behavior data corresponding to this derivative emergency plan index. This data records the emergency response actions and outcomes in similar past incidents and is a valuable reference for understanding and responding to current emergencies.

[0186] The information release platform then processes this historical emergency response behavior data using emergency plan knowledge embedding technology. This process aims to convert this raw data into a format more easily processed and analyzed by computers: a historical emergency response behavior vector. This vector representation captures the key characteristics and response patterns of emergency plans, facilitating subsequent feature mapping and matching.

[0187] After completing these preparatory steps, the information release platform will further invoke the disaster emergency warning algorithm within the emergency plan generation algorithm. This algorithm not only performs feature mapping between the disaster monitoring element vectors and the historical emergency response behavior vectors of each previous emergency plan index, but also considers the historical emergency response behavior vectors of the newly acquired derived emergency plan indexes. Through comprehensive comparison and analysis, the algorithm can identify an emergency plan that best matches the current disaster situation.

[0188] Specifically, the disaster emergency warning algorithm calculates the similarity or matching degree between the disaster monitoring element vector and each historical emergency response action vector, thereby generating a disaster emergency matching vector. This vector reflects the degree of fit between different emergency response plans and the current disaster situation. Ultimately, based on the results of this disaster emergency matching vector, the information release platform can determine the current emergency response plan for the pending emergency monitoring information.

[0189] By introducing a derived emergency plan index and its corresponding historical emergency response behavior data, the information release platform possesses greater flexibility and adaptability when handling emergencies. This technical solution enables the platform to quickly identify matching emergency plans when faced with new and unknown emergency situations, thereby improving the efficiency and accuracy of emergency response. Furthermore, by comprehensively considering historical data and current disaster situations, the platform is able to develop more scientific and reasonable emergency plans, providing more solid technical support for protecting public life and property.

[0190] Furthermore, Figure 2 This is a structural diagram of an information publishing platform 200 provided in an embodiment of the present application. Figure 2 The information publishing platform 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.

[0191] Alternatively, as Figure 2 As shown, the information publishing platform 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.

[0192] The memory 230 may be a separate device independent of the processor 210 , or may be integrated into the processor 210 .

[0193] Alternatively, as Figure 2 As shown, the information publishing platform 200 may further include a transceiver 220. The processor 210 may control the transceiver 220 to interact with other devices. Specifically, the transceiver 220 may send information or data to other devices, or receive information or data sent by other devices.

[0194] Optionally, the information publishing platform 200 can implement the corresponding processes corresponding to the storage engine or components in the storage engine (such as a processing module) or the device deployed with the storage engine in each method of the embodiments of the present application. For the sake of brevity, they will not be repeated here.

[0195] It should be understood that the processor of the embodiments of the present application may be an integrated circuit chip with signal processing capabilities. During implementation, each step of the above method embodiment can be completed by hardware integrated logic circuits in the processor or software instructions. The above processor can be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. The various methods, steps, and logic block diagrams disclosed in the embodiments of the present application can be implemented or executed. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in the embodiments of the present application can be directly implemented as a hardware decoding processor, or can be implemented by a combination of hardware and software modules in the decoding processor. The software module can be located in a storage medium mature in the art, such as random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, 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.

[0196] It is understood that the memory in the embodiments of the present application may be a volatile memory or a non-volatile memory, or may include both volatile and non-volatile memories. Among them, the non-volatile memory may 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 may 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.

[0197] It should be understood that the above-mentioned memories are exemplary but not restrictive. For example, the memories in the embodiments of the present application may also be 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 RAM (DR RAM), etc. In other words, the memories in the embodiments of the present application are intended to include, but are not limited to, these and any other suitable types of memories.

[0198] 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.

[0199] The above are merely examples of the present application and are not intended to limit the present application. For those skilled in the art, the present application may be subject to various modifications and variations.

Claims

1. A warning method for emergencies, characterized in that: Applied to an information publishing platform, the method includes: Obtain past emergency monitoring information and past emergency plan index collection; Calling a disaster monitoring element mining algorithm to perform disaster monitoring element mining on the past emergency monitoring information to obtain a disaster monitoring element vector of the past emergency monitoring information; the disaster monitoring element mining algorithm includes a plurality of monitoring element mining branches, and the past emergency monitoring information includes real-time monitoring data of a plurality of dimensions; calling the disaster monitoring element mining algorithm to perform disaster monitoring element mining on the past emergency monitoring information to obtain a disaster monitoring element vector of the past emergency monitoring information includes: obtaining the plurality of monitoring element mining branches, each monitoring element mining branch corresponding to a dimension; using the plurality of monitoring element mining branches, respectively performing monitoring element identification on the real-time monitoring data of corresponding dimensions of the past emergency monitoring information to obtain the disaster monitoring element vector of the corresponding dimension of the past emergency monitoring information; Calling the emergency plan knowledge embedding algorithm to respectively embed the historical emergency response behavior data corresponding to each past emergency plan index in the past emergency plan index set to obtain the historical emergency response behavior vector of each past emergency plan index, including: for each past emergency plan index corresponding to the historical emergency response behavior data, obtaining the response event stream corresponding to the historical emergency response behavior data; calling the emergency plan knowledge embedding algorithm to embed the emergency plan knowledge on the response event stream to obtain the historical emergency response behavior vector of the past emergency plan index; Debugging the disaster emergency warning algorithm based on the disaster emergency matching vector obtained by feature mapping the disaster monitoring element vector and each of the historical emergency response behavior vectors to obtain a disaster emergency warning algorithm that meets the debugging termination requirements; the disaster emergency matching vector reflects the degree of matching between the disaster monitoring element and the historical emergency response behavior; Based on the disaster monitoring element mining algorithm, the emergency plan knowledge embedding algorithm and the disaster emergency warning algorithm that meets the debugging termination requirements, an emergency plan generation algorithm is debugged to generate an emergency plan for the acquired emergency event monitoring information to be processed; the emergency plan generation includes generating a plan for the targeted sending of warning information to the target area, and generating a plan for data disaster recovery and backup processing to the target area.

2. The method according to claim 1, characterized in that The method further includes: if the past emergency event monitoring information includes real-time monitoring data in a meteorological dimension but does not include real-time monitoring data in a non-meteorological dimension, generating real-time monitoring data in a non-meteorological dimension based on the real-time monitoring data in the meteorological dimension, to obtain real-time monitoring data in several dimensions included in the past emergency event monitoring information, wherein the real-time monitoring data in the non-meteorological dimension includes at least one of real-time monitoring data in a personnel activity dimension, real-time monitoring data in a building structure dimension, and real-time monitoring data in a production dimension; The obtaining of the plurality of monitoring element mining branches includes: Obtain the initial monitoring element mining branches corresponding to each dimension of the emergency reference monitoring information, where each initial monitoring element mining branch corresponds to one dimension; By mining the initial monitoring elements corresponding to each dimension, the real-time monitoring data of the corresponding dimension of the emergency reference monitoring information are respectively identified with monitoring elements, thereby obtaining the intra-class disaster monitoring element vector corresponding to the corresponding dimension of the emergency reference monitoring information; For each intra-class disaster monitoring element vector, feature mapping is performed between the intra-class disaster monitoring element vector and the historical emergency response behavior vectors indexed by each past emergency plan to obtain a corresponding disaster emergency matching vector, and the disaster emergency warning algorithm is debugged according to each of the disaster emergency matching vectors to obtain several disaster emergency warning algorithms; Based on the disaster monitoring element mining algorithm, the emergency plan knowledge embedding algorithm and the disaster emergency warning algorithm, debugging and obtaining several emergency plan generation algorithms; Based on the plan matching weights of the several emergency plan generation algorithms, several monitoring element mining branches are selected from the initial monitoring element mining branches corresponding to the various dimensions.

3. The method according to claim 1, characterized in that The method further comprises: For each past emergency plan index in the past emergency plan index set, based on the request feature pair feature of the pre-recorded request feature vector and the historical emergency response behavior data, the historical emergency response behavior data corresponding to the past emergency plan index is obtained based on the past emergency plan index.

4. The method according to claim 1, wherein The method further comprises: For each past emergency plan index in the set of past emergency plan indexes, generating guidance information for activating the AI ​​algorithm based on the past emergency plan index; A generative adversarial network is called based on the guidance information, and historical emergency response behavior data corresponding to the past emergency plan index is determined through the generative adversarial network.

5. The method according to claim 1, wherein The method of debugging the disaster emergency warning algorithm based on the disaster emergency matching vector obtained by feature mapping the disaster monitoring element vector and each of the historical emergency response behavior vectors to obtain a disaster emergency warning algorithm that meets the debugging termination requirements includes: Inputting the disaster monitoring element vector and each of the historical emergency response behavior vectors into the disaster emergency warning algorithm based on the linkage feature focusing rule in the emergency plan generation algorithm, and outputting the disaster emergency matching vector corresponding to each past emergency plan index through the disaster emergency warning algorithm; Inputting the disaster emergency matching vector corresponding to each past emergency plan index into the first emergency warning subnet or the second emergency warning subnet, optimizing the algorithm variables of the disaster emergency warning algorithm according to the warning training information, and obtaining a disaster emergency warning algorithm that meets the debugging termination requirements; The step of inputting the disaster monitoring element vector and each of the historical emergency response behavior vectors into the disaster emergency warning algorithm based on the linkage feature focusing rule in the emergency plan generation algorithm, and outputting the disaster emergency matching vector corresponding to each of the past emergency plan indexes through the disaster emergency warning algorithm, includes: The disaster monitoring element vector is used as a request feature, and each of the historical emergency response behavior vectors is used as a response feature. The vectors are input into the disaster emergency warning algorithm based on the linkage feature focusing rule in the emergency plan generation algorithm. The request feature and each of the response features are feature matched by the disaster emergency warning algorithm to obtain the disaster emergency matching vector corresponding to each past emergency plan index.

6. The method according to claim 1, wherein The method of debugging the disaster emergency warning algorithm based on the disaster emergency matching vector obtained by feature mapping the disaster monitoring element vector and each of the historical emergency response behavior vectors to obtain a disaster emergency warning algorithm that meets the debugging termination requirements includes: Acquire training annotations configured for the past emergency event monitoring information, where the training annotations configured for the past emergency event monitoring information belong to the past emergency plan index set; Determining, based on the training annotations configured for the past emergency event monitoring information, a priori training indication of each past emergency plan index corresponding to the past emergency event monitoring information; Calculating the matching training confidence of the past emergency event monitoring information corresponding to each past emergency plan index based on the disaster emergency matching vector obtained by feature mapping the disaster monitoring element vector and each historical emergency response behavior vector; Based on the prior training indication and the matching training confidence, an algorithm training error is constructed, and with the convergence of the algorithm training error as the debugging goal, the algorithm variables of the disaster emergency warning algorithm are optimized to obtain a disaster emergency warning algorithm that meets the debugging termination requirements.

7. The method according to any one of claims 1 to 6, characterized in that The method further comprises: When a derived emergency plan index appears in the past emergency plan index set, obtaining historical emergency response behavior data corresponding to the derived emergency plan index; Call the constructed emergency plan generation algorithm and the historical emergency response behavior data corresponding to each past emergency plan index in the past emergency plan index set including the derived emergency plan index, generate an emergency plan for the emergency event monitoring information to be processed, and obtain an emergency plan output result. The derived emergency plan index is an emergency plan index that has not appeared in the debugging process of the emergency plan generation algorithm.

8. The method according to claim 1, characterized in that The method further comprises: Obtain monitoring information on pending emergencies; Calling the disaster monitoring element mining algorithm in the emergency plan generation algorithm to mine disaster monitoring elements on the emergency event monitoring information to be processed to obtain a disaster monitoring element vector; Calling the emergency plan knowledge embedding algorithm in the emergency plan generation algorithm to perform emergency plan knowledge embedding on the historical emergency response behavior data corresponding to each past emergency plan index in the past emergency plan index set, and obtaining the historical emergency response behavior vector of each past emergency plan index; Calling the disaster emergency warning algorithm in the emergency plan generation algorithm, performing feature mapping on the disaster monitoring element vector and the historical emergency response behavior vectors indexed by each past emergency plan to obtain a disaster emergency matching vector, and determining the current emergency plan for the pending emergency event monitoring information based on the disaster emergency matching vector; The method further comprises: When a derived emergency plan index appears in the past emergency plan index set, obtaining historical emergency response behavior data corresponding to the derived emergency plan index, wherein the derived emergency plan index is an emergency plan index that has not appeared during the debugging process of the emergency plan generation algorithm; Performing emergency plan knowledge embedding on the historical emergency response behavior data corresponding to the derived emergency plan index to obtain a historical emergency response behavior vector corresponding to the derived emergency plan index; The calling of the disaster emergency warning algorithm in the emergency plan generation algorithm, performing feature mapping on the disaster monitoring element vector and the historical emergency response behavior vectors indexed by each past emergency plan to obtain a disaster emergency matching vector, and determining a current emergency plan for the pending emergency event monitoring information based on the disaster emergency matching vector, includes: Through the disaster emergency warning algorithm in the emergency plan generation algorithm, feature mapping is performed on the disaster monitoring element vector and the historical emergency response behavior vectors of the past emergency plan indexes and the historical emergency response behavior vectors of the derived emergency plan indexes to obtain a disaster emergency matching vector, and the current emergency plan for the emergency event monitoring information to be processed is determined based on the disaster emergency matching vector.

9. An information publishing platform, 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 8.

Images