Extreme rainstorm cascade disaster emergency decision-making method and system fusing multi-source data

Abstract

The invention discloses an extreme rainstorm cascade disaster emergency decision-making method and system fused with multi-source data, and the method comprises the steps: constructing a historical event knowledge graph and a current event dynamic evolution knowledge graph through integrating the multi-source data and using the integrated multi-source data; searching current and subsequent disaster risks and corresponding emergency decision-making schemes in the historical event knowledge graph according to static disaster characteristics of high similarity in attributes of the historical event knowledge graph and the current event dynamic evolution knowledge graph; and dynamically adjusting the emergency decision scheme and the historical event knowledge graph according to feedback information executed by the decision scheme. According to the method, the accuracy and the real-time performance of extreme rainstorm cascade disaster assessment and emergency decision making are improved, and meanwhile, the prediction capability and the adaptive capability are improved.

Description

Technical Field

[0001] The present invention relates to the field of prevention and control of extreme rainstorm cascade disaster risks, and particularly to an extreme rainstorm cascade disaster emergency decision-making method and system integrating multi-source data. Background Art

[0002] Extreme rainstorm events often easily induce secondary disasters such as floods, waterlogging, and debris flows, which in turn lead to derivative disasters, expanding the damage degree and influence scope of the extreme rainstorm events themselves, and having obvious cascade characteristics. Therefore, effectively curbing the amplification trend of disasters is beneficial to reducing economic losses and casualties.

[0003] Existing emergency decision-making methods are divided into three categories, including methods based on mathematical models, methods based on knowledge reasoning, and methods based on simulation. The deficiencies of methods based on mathematical models are that the assumptions are too idealistic, the real problems are simplified, it is difficult to reflect the complex dynamic evolution of disasters, and high requirements are imposed on the accuracy and integrity of data. The deficiencies of methods based on knowledge reasoning are that it is necessary to manually sort out a large amount of historical cases, with high costs, and the static knowledge base is difficult to respond to the changes in the disaster situation in real time and cope with complex disaster scenarios that have not been seen before. The deficiencies of methods based on simulation are that the modeling complexity is relatively high, it is difficult to quickly output results, and it is greatly restricted by computing and storage resources.

[0004] From a practical perspective, the existing emergency decision-making methods have deficiencies in dealing with extreme rainstorm cascade disasters. First, traditional emergency decision-making methods often rely on limited monitoring data and the experience judgment of decision-makers, and it is difficult to comprehensively and accurately evaluate the current situation of extreme rainstorm cascade disasters. Second, the existing emergency decision-making methods generally only give emergency disposal plans for the current disaster scenario, rarely considering the dynamic evolution characteristics and trends of cascade disasters, and lacking real-time performance and predictability, resulting in the decision being prone to lag. Summary of the Invention

[0005] In view of the above deficiencies in the prior art, the present invention provides an extreme rainstorm cascade disaster emergency decision-making method and system integrating multi-source data.

[0006] In order to achieve the above invention purpose, the technical solution adopted by the present invention is as follows: An extreme rainstorm cascade disaster emergency decision-making method integrating multi-source data, comprising the following steps: Step S1, collecting real-time data, forecast data, historical case data, emergency plan data, and accident investigation report data of extreme rainstorms, and preprocessing all the collected data; Step S2, constructing a historical event knowledge graph using historical case data, emergency plan data, and accident investigation report data; Step S3. Construct a dynamic evolution knowledge graph of the current event using real-time data and forecast data; Step S31. Extract static disaster feature entities, relationships between entities, and attributes from real-time data and forecast data; Step S32. Construct a dynamic evolution knowledge graph of the current event based on the historical event knowledge graph and the static disaster feature entities, relationships between entities, and attributes of real-time data and forecast data, and dynamically adjust the historical event knowledge graph; Step S4. Calculate the similarity in attributes between the static disaster feature entities of the dynamic evolution knowledge graph of the current event and the historical event knowledge graph; In the historical event knowledge graph, find static disaster feature entities and emergency response measure entities with high similarity to the dynamic evolution knowledge graph of the current event, and determine the existing disaster risks and corresponding emergency response measures currently and in the future; Step S5. Collect feedback information during the execution of emergency response measures, and adjust the emergency response measures in real time according to the feedback information; After the emergency response to the current disaster event is completed, revise the historical event knowledge graph in combination with the actual problems that occurred during the response process.

[0007] Furthermore, in step S1, preprocess all the collected data, including the following steps: (1) Manually delete noise data; (2) Use the Pandas library in Python to delete duplicate data and missing data in structured text data, and unify the format of the same type of data; (3) Use the re library in Python in combination with regular expressions to delete advertisement links, garbled characters, special symbols, and extra spaces in unstructured text data.

[0008] Furthermore, step S2 includes the following steps: Step S21. Use the natural language processing library in Python, in combination with the Word2vec model and regular expressions, to extract static disaster feature entities and emergency response measure entities from pre-historical case data, emergency plan data, and accident investigation report data; Step S22. Adopt a rule-based method to identify the relationships between entities; Step S23. Extract the attributes of entities through regular expressions; Step S24. Construct a historical event knowledge graph with each entity as a node, the relationships between entities as edges, and the attributes of entities and their relationships as the attributes of nodes and edges respectively; Step S25. Use Neo4j to store the constructed historical event knowledge graph.

[0009] Further, the static disaster feature entities and emergency response measure entities in step S21 specifically include: The static disaster feature entities include time information entities, location information entities, disaster entities, disaster-affected entities, and disaster-forming environment entities; the emergency response measure entities include organizational entities, emergency resource entities, and emergency operation entities.

[0010] Further, step S32 includes the following steps: Back up the historical event knowledge graph to obtain a backup knowledge graph; Compare the static disaster entities, relationships, and attributes of the real-time data and forecast data with the backup knowledge graph; If there is overlapping content between the two, retain the corresponding nodes, edges, and attributes of the overlapping content in the backup knowledge graph; If there is content in the backup knowledge graph that does not appear in the static disaster entities, relationships, and attributes of the real-time data and forecast data, delete the corresponding nodes, edges, and attributes of the non-appearing content in the backup knowledge graph; If there is content in the static disaster entities, relationships, and attributes of the real-time data and forecast data that does not appear in the backup knowledge graph, add the corresponding nodes, edges, and attributes of the non-appearing content to both the backup knowledge graph and the historical event knowledge graph; After the above processing of the backup knowledge graph, generate the current event dynamic evolution knowledge graph.

[0011] Further, step S4 includes the following steps: S41. Calculate the similarity of the attributes of the static feature entities that overlap between the current event dynamic evolution knowledge graph and the historical event knowledge graph, including: Calculate the similarity of the static feature entities that overlap between the current event dynamic evolution knowledge graph and the historical event knowledge graph in the th attribute ; For numerical data:

[0012] where is the th attribute value of the static feature entities that overlap between the current event dynamic evolution knowledge graph and the historical event knowledge graph, is the th attribute value of the static feature entities that overlap between the current event dynamic evolution knowledge graph and the historical event knowledge graph; For interval data:

[0013] where and respectively represent the upper and lower limits of the th attribute value of the overlapping static feature entities in the current event dynamic evolution knowledge graph. and respectively represent the upper and lower limits of the th attribute value of the overlapping static feature entities in the historical event knowledge graph; For symbolic data:

[0014] According to the weights of the attributes, calculate the similarity of the overlapping static feature entities between the current event dynamic evolution knowledge graph and the historical event knowledge graph in terms of attributes:

[0015] where is the weight of the th attribute, .

[0016] S42. Set the attribute similarity threshold, and based on the attribute similarity threshold, search in the historical event knowledge graph for static disaster feature entities with high similarity to the current event dynamic evolution knowledge graph, and determine the existing and subsequent disaster risks; S43. In the historical event knowledge graph, retrieve the emergency response measure entities corresponding to the highly similar static disaster entities, as well as the relationships and attributes between the entities, and determine the emergency response measures corresponding to the existing and subsequent disaster risks.

[0017] An extreme rainstorm cascading disaster emergency decision-making system that integrates multi-source data, used to execute the above-mentioned extreme rainstorm cascading disaster emergency decision-making method based on multi-source data, includes: Data acquisition module: used to collect data related to extreme rainstorm cascading disasters and preprocess the data; Knowledge graph construction module: used to construct a historical event knowledge graph based on historical case data, emergency plan data, and accident investigation reports; construct a current event dynamic evolution knowledge graph based on the historical event knowledge graph, real-time data, and forecast data; Decision generation module: responsible for finding static disaster entities with high similarity in attributes between the current event dynamic evolution knowledge graph and the static disaster features in the historical event knowledge graph, and retrieving the emergency response measures for the existing and subsequent disasters in the historical event knowledge graph according to the highly similar static disaster entities; Decision execution monitoring module: responsible for monitoring the execution of the decision-making plan, collecting feedback information, and transmitting the feedback information to other modules for analysis and processing; User interaction module: responsible for providing a visual operation interface for decision-makers, displaying disaster information and emergency decision-making plans, and supporting manual modification.

[0018] The present invention has the following beneficial effects: By integrating multi-source data, the present invention constructs a historical event knowledge graph and a current event dynamic evolution knowledge graph using the integrated multi-source data. According to the static disaster characteristics with high similarity in attributes between the historical event knowledge graph and the current event dynamic evolution knowledge graph, it retrieves the current and subsequent disaster risks and corresponding emergency decision-making plans in the historical event knowledge graph, and dynamically adjusts the emergency decision-making plans and the historical event knowledge graph according to the feedback information of the implementation of the decision-making plans. The present invention integrates heterogeneous and fragmented multi-source data of extreme rainstorm cascade disasters, comprehensively reflects the static disaster characteristics and dynamic evolution process of extreme rainstorm cascade disasters, and improves the accuracy of emergency decision-making for extreme rainstorm cascade disasters; constructs a current event dynamic evolution knowledge graph based on the historical event knowledge graph, real-time data, and forecast data, improving the timeliness and predictability of disaster situation judgment and emergency decision-making for extreme rainstorm cascade disasters; continuously adjusts the historical event knowledge graph through real-time data, forecast data, and emergency decision-making feedback information each time, improving the adaptive ability of emergency decision-making for extreme rainstorm cascade disasters. Brief Description of the Drawings

[0019] Figure 1 It is the overall flowchart of this method. Specific Embodiments

[0020] The following describes the specific embodiments of the present invention to facilitate those skilled in the art of this technology to understand the present invention. However, it should be clear that the present invention is not limited to the scope of the specific embodiments. For those of ordinary skill in the art of this technology, as long as various changes are within the spirit and scope of the present invention defined and determined by the appended claims, these changes are obvious, and all inventions and creations using the concept of the present invention are within the scope of protection.

[0021] As Figure 1 shown, the present invention provides an emergency decision-making method for extreme rainstorm cascade disasters that integrates multi-source data, including the following steps: Step S1: Collect real-time data, forecast data, historical case data, emergency plan data, and accident investigation report data of extreme rainstorms, and preprocess all the collected data; Collect data from different sources, such as real-time data, forecast data, historical case data, emergency plan data, accident investigation report data, etc. The first type of data is real-time data and forecast data, specifically including the actual precipitation data and forecast data of the meteorological management department, the river and reservoir water level data of the water resources management department, the meteorological risk warning data of geological disasters of the natural resources management department, the real-time road conditions and road condition prediction data of Baidu Maps, and news reports describing the occurrence and development process of current extreme rainstorm events. The second type of data is case data, collecting text data such as news reports describing the occurrence and development process of historical extreme rainstorm events. The third type is emergency plan data, collecting emergency plans for flood disasters and geological disasters in each province, provincial capitals, and flood-prone cities. The fourth type is accident investigation reports, collecting accident investigation reports on rainstorms, floods, and geological disasters; Preprocess all the collected data, including the following steps: (1) Manually delete noise data, including news texts lacking key information such as occurrence time, location, and description of the occurrence and development process of the event, emergency plans irrelevant to flood disasters and geological disasters, and accident investigation reports irrelevant to rainstorms, floods, and geological disasters; (2) Use the Pandas library in Python to process duplicate data and missing data in structured texts. Specifically, use the duplicated function and drop_duplicates function to find and delete duplicate data respectively, and use the isnull function and dropna function to find and delete data with missing values respectively. In addition, unify the data format. Use the datetime library to unify the date format into the form of "year-month-day hour:minute", for example, "2021-7-20 14:00". Unify the number of decimal places reserved, etc.; (3) In Python, use the re library combined with predefined regular expressions to delete advertisement links, garbled characters, special symbols, and extra spaces in unstructured texts. Their corresponding rules are https?: / / [^\s]+\.(com|cn|org|net|edu)[^\s], garbled characters and special symbols other than [^\w\s\u4e00-\u9fa5,。!?、:;‘’“”《》【】()…—\-], and \s+.

[0022] Step S2: Use historical case data, emergency plan data, and accident investigation report data to construct a historical event knowledge graph, including the following steps: Step S21: Use the natural language processing library in Python, combined with the Word2vec model and regular expressions, to extract static disaster feature entities and emergency response measure entities from pre-historical case data, emergency plan data, and accident investigation report data; Two major categories of entities are extracted. The first major category of entities is static disaster characteristic entities, including basic information entities reflecting time and location, disaster entities such as "rainstorm", "flood", "landslide", "mudslide", etc., disaster-affected body entities related to casualties, damage to building facilities, and damage to critical infrastructure, and disaster-forming environment entities covering social environment, natural environment, and economic environment; The second major category of entities is emergency response measure entities, including organizational entities such as "fire brigade", emergency resource entities such as "cotton-padded quilt", "tent", etc., and emergency action entities related to personnel evacuation, engineering emergency rescue, medical rescue, etc.; Specifically, (1) Match news texts according to the regular expression rules of time shown in Table 1, and extract time information entities such as the disaster occurrence time, the time when emergency response measures are taken, and the time of disaster evolution and escalation points. (2) Construct a regular expression based on the database containing Chinese provincial and city names to match the disaster occurrence location. After extracting the place names, unify the place names into the form of "a certain province / autonomous region + a certain city". (3) Refer to the "Classification and Codes of Natural Disasters" (GB / T 28921-2012) to determine the seed trigger words, including "typhoon", "rainstorm", "flood", "landslide", "mudslide", "collapse", "ground subsidence", "ground settlement", "ground fissure". Then, train a Word2vec model (the dimension of word vectors is 100, the window size is 5, the minimum word frequency is 1, and the number of threads is 4) on the corpus containing all news texts, and extract the top 10 words most similar to various sub-trigger words to obtain the disaster type trigger word list, as shown in Table 2. According to the disaster type trigger word list, identify different disaster entities in the historical case data, laying a foundation for subsequent identification of the relationships between disaster entities. (4) Refer to the "Classification and Codes of Disaster-Affected Bodies of Natural Disasters" (GB / T 32572-2016) to determine the seed trigger words, including "person", "household property", "public property". Then, train a Word2vec model (the dimension of word vectors is 100, the window size is 5, the minimum word frequency is 1, and the number of threads is 4) on the corpus containing all news texts, and extract the top 20 words most similar to various sub-trigger words to obtain the disaster-affected body type trigger word list, as shown in Table 3. According to the disaster-affected body type trigger word list, identify different disaster-affected body entities in the historical case data, laying a foundation for subsequent identification of the relationships between disaster entities and disaster-affected body entities. (5) The extraction methods of disaster-forming environment entities, organizational entities, emergency resource entities, and emergency action entities are similar to those in (3) and (4). First, establish trigger word lists, as shown in Tables 4 and 5 respectively, and then identify entities based on the trigger word lists.

[0023] Table 1 Regular Expression Rules of Time

[0024] Table 2 Natural Disaster Type Trigger Word List

[0025] Table 3 List of Trigger Words for Types of Disaster-Bearing Bodies in Natural Disasters

[0026] Table 4 List of Trigger Words for Disaster-Inducing Environments in Natural Disasters

[0027] Table 5 List of Trigger Words for Entities of Emergency Disposal Measures

[0028] Step S22: Use a rule-based method to identify the relationships between entities; Use a rule-based method to formulate a series of rules to identify the relationships between entities; if the text mentions that "heavy rain triggers floods", then it can be determined that there is a "trigger" relationship between "heavy rain" and "floods"; for "traffic police conduct safety hazard inspections, issue safety warnings, clear blocked roads, and open rescue green channels", it can be determined that there is an "execute" relationship between "traffic police" and "conduct safety hazard inspections, issue safety warnings, clear blocked roads, and open rescue green channels".

[0029] Step S23: Extract the attributes of entities through regular expressions; For example, the attributes of "heavy rain" include rainfall and rainfall duration, and the attribute of emergency resources is quantity, etc.

[0030] Step S24: Construct a historical event knowledge graph with each entity as a node, the relationships between entities as edges, and the attributes of entities and their relationships as the attributes of nodes and edges respectively; For example, taking "heavy rain", "flood", "mudslide", "traffic interruption" as nodes and "trigger", "cause" as edges can reflect an evolutionary path of extreme rainstorm cascading disasters. The relationships between "waterlogging" and entities such as "pumping stations", "sandbags", "drainage vehicles", and between "drainage emergency rescue teams" and entities such as "pumping stations", "sandbags", "drainage vehicles" are "disposal measures" and "use" respectively, which can reflect specific emergency disposal measures.

[0031] Step S25: Use Neo4j to store the constructed historical event knowledge graph.

[0032] Step S3: Use real-time data and forecast data to construct a dynamic evolution knowledge graph of the current event; Step S31: Extract static disaster feature entities, the relationships between entities, and their attributes from real-time data and forecast data; Step S32: Construct a dynamic evolution knowledge graph of the current event based on the static disaster feature entities, the relationships and attributes between entities in the historical event knowledge graph, real-time data, and forecast data, and dynamically adjust the historical event knowledge graph; Back up the historical event knowledge graph to obtain a backup knowledge graph; Compare the static disaster entities, relationships, and attributes in the real-time data and forecast data with the backup knowledge graph; If there is overlapping content between the two, retain the corresponding nodes, edges, and attributes of the overlapping content in the backup knowledge graph; If there is content in the backup knowledge graph that does not appear in the static disaster entities, relationships, and attributes of the real-time data and forecast data, delete the corresponding nodes, edges, and attributes of the unappeared content in the backup knowledge graph; If there is content in the static disaster entities, relationships, and attributes of the real-time data and forecast data that does not appear in the backup knowledge graph, add the corresponding nodes, edges, and attributes of the unappeared content to both the backup knowledge graph and the historical event knowledge graph; After the above processing of the backup knowledge graph, generate a dynamic evolution knowledge graph of the current event.

[0033] Step S4: Calculate the similarity in attributes between the static disaster feature entities in the dynamic evolution knowledge graph of the current event and the historical event knowledge graph; in the historical event knowledge graph, search for static disaster feature entities and emergency response measure entities with high similarity to the dynamic evolution knowledge graph of the current event, and determine the existing disaster risks and corresponding emergency response measures at present and in the future, which specifically include the following steps: S41: Calculate the similarity in attributes between the overlapping static feature entities in the dynamic evolution knowledge graph of the current event and the historical event knowledge graph, including: Calculate the similarity of the overlapping static feature entities in the dynamic evolution knowledge graph of the current event and the historical event knowledge graph in the th attribute ; For numerical data:

[0034] Among them, is the th attribute value of the overlapping static feature entity in the dynamic evolution knowledge graph of the current event, is the th attribute value of the overlapping static feature entity in the dynamic evolution knowledge graph of the current event,; For interval data:

[0035] Among them, and are respectively the upper and lower limits of the th attribute value of the overlapping static feature entities of the current event's dynamic evolution knowledge graph. and are respectively the upper and lower limits of the th attribute value of the overlapping static feature entities of the historical event knowledge graph; For symbolic data:

[0036] According to the weight of the attribute, calculate the similarity of the overlapping static feature entities between the current event's dynamic evolution knowledge graph and the historical event knowledge graph in terms of attributes:

[0037] where is the weight of the th attribute, .

[0038] S42. Set the attribute similarity threshold, and based on the attribute similarity threshold, search in the historical event knowledge graph for static disaster feature entities with high similarity to the current event's dynamic evolution knowledge graph to determine the current and subsequent existing disaster risks; S43. In the historical event knowledge graph, retrieve the corresponding emergency response measure entities, their relationships, and attributes of the static disaster entities with high similarity to determine the emergency response measures corresponding to the current and subsequent disaster risks.

[0039] If it is determined according to the historical event knowledge graph that the current event has caused a power failure, then it can be preliminarily judged based on the dynamic evolution path that there is a power failure risk currently, and real-time data can be continuously combined to reason about the possible disasters or accidents caused by the current disaster. For example, if the knowledge graph shows that "heavy rain" will cause "landslide", and currently it is monitored that the rainfall in a certain area is large, and at the same time the geological conditions in this area are unstable, it can be inferred that a landslide disaster may occur in this area. Further, retrieve the corresponding emergency response measures such as evacuating people, setting up warning lines, issuing warnings, and closing roads.

[0040] Step S5. Collect the feedback information during the execution of the emergency response measures, and adjust the emergency response measures in real time according to the feedback information; after the emergency response to the current disaster event is completed, revise the historical event knowledge graph in combination with the actual problems that occurred during the handling process.

[0041] The generated decision-making plan is issued to relevant emergency departments for implementation, and feedback information during the plan implementation process of each department is collected through a real-time monitoring system; based on the feedback information, the decision-making plan is adjusted to form a closed-loop decision-making mechanism, continuously improving the accuracy and effectiveness of emergency decision-making; specifically, adjustments can be made from the following aspects: (1) Precise disaster situation assessment. Since limited data can be obtained in the initial stage of a disaster, there may be deviations in the disaster situation assessment. When the feedback information shows that the original plan's assessment of the impact degree and scope of the disaster situation is inaccurate, it is necessary to correct the assessment results and re-give emergency response measures. For example, if the original plan judges that a certain area is only slightly affected by waterlogging, but the actual feedback information indicates that due to the blockage of the drainage system in this area, the degree of waterlogging far exceeds expectations. At this time, it is necessary to re-assess the affected degree of this area, adjust the allocation of rescue personnel and materials accordingly, and add emergency response measures such as "unclogging the drainage system in this area". (2) Adjustment of the allocation of rescue personnel and materials. If the feedback information shows that there is a shortage of personnel and materials in some severely affected areas, while there is an excess of personnel and materials in some less affected areas, then re-adjust the resource allocation plan and transfer some personnel and materials from the excess areas to the nearby shortage areas. (3) Improvement of the emergency response process. If there are communication barriers and situations of shifting responsibilities among departments during the cooperation process, then clearly define the specific responsibilities and task divisions of each department in responding to extreme rainstorm cascade disasters in a timely manner. If it is found that some measures are ineffective, such as the drainage measures for urban waterlogging in the original plan fail to effectively relieve the waterlogging problem, an expert team should be organized to analyze the reasons and make improvements. If the drainage pipe network planning is unreasonable, professional opinions can be referred to and the drainage strategy can be adjusted, such as adding temporary drainage points and optimizing the drainage route. After the emergency response and rescue of the current incident are completed, in combination with the characteristics of this incident and the problems encountered during the disposal process, the existing emergency plan is revised. Add or adjust specific measures for dealing with different types of cascade disasters, such as formulating more detailed response strategies for secondary disasters such as landslides and mudslides caused by rainstorms, and adding them to the knowledge graph for the convenience of emergency disposal of subsequent similar incidents.

[0042] The present invention provides an extreme rainstorm cascade disaster emergency decision-making system integrating multi-source data, including: A data acquisition module: used to collect data related to extreme rainstorm cascade disasters and preprocess the data; A knowledge graph construction module: used to construct a historical event knowledge graph based on historical case data, emergency plan data, and accident investigation reports; construct a current event dynamic evolution knowledge graph based on the historical event knowledge graph, real-time data, and forecast data; Decision-making generation module: responsible for finding static disaster entities with high similarity in attributes between the static disaster features of the current event's dynamic evolution knowledge graph and the historical event knowledge graph, and retrieving emergency response measures for the current and subsequent disasters in the historical event knowledge graph; Decision-making execution monitoring module: responsible for monitoring the execution of the decision-making plan, collecting feedback information, and transmitting the feedback information to other modules for analysis and processing; User interaction module: responsible for providing a visual operation interface for decision-makers, displaying disaster information and emergency decision-making plans, and supporting manual modification.

[0043] The present invention is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, and the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, such that the instructions executed by the processor of the computer or other programmable data processing devices generate means for implementing the functions specified in Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0044] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, such that the instructions stored in the computer-readable memory generate a manufactured article including instruction means that implement the functions specified in Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0045] These computer program instructions can also be loaded onto a computer or other programmable data processing device, such that a series of operation steps are executed on the computer or other programmable device to generate computer-implemented processing, so that the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0046] In the present invention, specific embodiments are used to illustrate the principles and implementation manners of the present invention. The description of the above embodiments is only used to help understand the method and its core idea of the present invention; at the same time, for those of ordinary skill in the art, based on the idea of the present invention, there will be changes in the specific implementation manners and application scopes. In summary, the content of this specification should not be construed as a limitation to the present invention.

[0047] Those of ordinary skill in the art will realize that the embodiments described herein are for helping the reader understand the principles of the present invention, and it should be understood that the protection scope of the present invention is not limited to such specific statements and embodiments. Those of ordinary skill in the art can make various other specific deformations and combinations that do not depart from the essence of the present invention according to these technical revelations disclosed by the present invention, and these deformations and combinations are still within the protection scope of the present invention.

Claims

1. An extreme rainstorm cascading disaster emergency decision-making method integrating multi-source data, characterized in that, It includes the following steps: Step S1: Collect real-time data, forecast data, historical case data, emergency plan data, and accident investigation report data of extreme rainstorms, and preprocess all the collected data; Step S2: Use historical case data, emergency plan data, and accident investigation report data to construct a historical event knowledge graph; Step S3: Based on the historical event knowledge graph, use real-time data and forecast data to construct a current event dynamic evolution knowledge graph; Step S31: Extract static disaster feature entities, relationships between entities, and attributes in real-time data and forecast data; Step S32: According to the historical event knowledge graph and the static disaster feature entities, relationships between entities, and attributes in real-time data and forecast data, construct a current event dynamic evolution knowledge graph, and dynamically adjust the historical event knowledge graph; Step S4: Calculate the similarity of the attributes of the static disaster feature entities between the current event dynamic evolution knowledge graph and the historical event knowledge graph; In the historical event knowledge graph, find the static disaster feature entities and emergency response measure entities with high similarity to the current event dynamic evolution knowledge graph, and determine the existing disaster risks and corresponding emergency response measures currently and in the future; Step S5: Collect feedback information during the execution of the emergency response measures, and adjust the emergency response measures in real-time according to the feedback information; After the emergency response to the current disaster event is completed, revise the historical event knowledge graph in combination with the actual problems that occurred during the response process.

2. The extreme rainstorm cascading disaster emergency decision-making method for integrating multi-source data according to claim 1, wherein The preprocessing of all the collected data in step S1 includes the following steps: (1) Manually delete noise data; (2) Use the Pandas library in Python to delete duplicate data and missing data in structured text data, and unify the formats of the same type of data; (3) Use the re library in Python in combination with regular expressions to delete advertisement links, garbled characters, special symbols, and extra spaces in unstructured text data.

3. The extreme rainstorm cascading disaster emergency decision-making method for integrating multi-source data according to claim 1, characterized in that, Step S2 includes the following steps: Step S21: Use the natural language processing library in Python, in combination with the Word2vec model and regular expressions, to extract static disaster feature entities and emergency response measure entities from pre-historical case data, emergency plan data, and accident investigation report data; Step S22: Adopt a rule-based method to identify the relationships between entities; Step S23: Extract the attributes of entities through regular expressions; Step S24: Use each entity as a node, the relationship between entities as an edge, and the attributes of the entity and its relationship as the attributes of the node and the edge respectively to construct a historical event knowledge graph; Step S25: Use Neo4j to store the constructed historical event knowledge graph.

4. The extreme rainstorm cascading disaster emergency decision-making method for fusing multi-source data according to claim 3, characterized in that, The static disaster feature entities and emergency response measure entities in step S21 specifically include: The static disaster feature entities include time information entities, location information entities, disaster entities, disaster-bearing entities, and disaster-forming environment entities; The emergency response measure entities include organization entities, emergency resource entities, and emergency action entities.

5. The extreme rainstorm cascading disaster emergency decision-making method for integrating multi-source data according to claim 1, characterized in that Step S32 includes the following steps: Back up the historical event knowledge graph to obtain a backup knowledge graph; Compare the static disaster entities, relationships, and attributes of real-time data and forecast data with the backup knowledge graph; If there is overlapping content between the two, retain the corresponding nodes, edges, and attributes of the overlapping content in the backup knowledge graph; If there is content in the backup knowledge graph that does not appear in the static disaster entities, relationships, and attributes of real-time data and forecast data, delete the corresponding nodes, edges, and attributes of the unappeared content in the backup knowledge graph; If there is content in the static disaster entities, relationships, and attributes of real-time data and forecast data that does not appear in the backup knowledge graph, add the corresponding nodes, edges, and attributes of the unappeared content to both the backup knowledge graph and the historical event knowledge graph; After the above processing of the backup knowledge graph, generate the current event dynamic evolution knowledge graph.

6. The extreme rainstorm cascading disaster emergency decision-making method for integrating multi-source data according to claim 1, characterized in that The step S4 includes the following steps: S41. Calculate the similarity of the attributes of the static feature entities that overlap between the current event dynamic evolution knowledge graph and the historical event knowledge graph, including: Calculate the similarity of the overlapping static feature entities between the current event's dynamically evolving knowledge graph and the historical event's knowledge graph on the th attribute ; For numerical data: Among them, is the th attribute value of the static feature entity that coincides with the current event's dynamic evolution knowledge graph, is the th attribute value,; For interval data: Among them, and are respectively the upper and lower limits of the th attribute value of the coincident static feature entities of the current event dynamic evolution knowledge graph. and are respectively the upper and lower limits of the th attribute value of the coincident static feature entities of the historical event knowledge graph; For symbolic data: Calculate the similarity of the attributes of the static feature entities that overlap between the current event dynamic evolution knowledge graph and the historical event knowledge graph according to the weight of the attributes: Among them, is the weight of the th attribute, . S42. Set the attribute similarity threshold, and based on the attribute similarity threshold, find the static disaster feature entities with high similarity to the current event dynamic evolution knowledge graph in the historical event knowledge graph, and determine the existing and subsequent disaster risks; S43. In the historical event knowledge graph, retrieve the emergency response measure entities corresponding to the static disaster entities with high similarity and the relationships and attributes between the entities, and determine the emergency response measures corresponding to the existing and subsequent disaster risks.

7. An extreme rainstorm cascading disaster emergency decision-making system integrating multi-source data, characterized in that, Used to execute the extreme rainstorm cascading disaster emergency decision-making method for fusing multi-source data as described in claim 1, including: Data acquisition module: used to collect data related to extreme rainstorm cascading disasters and preprocess the data; Knowledge graph construction module: used to construct a historical event knowledge graph based on historical case data, emergency plan data, and accident investigation reports; construct a current event dynamic evolution knowledge graph based on the historical event knowledge graph, real-time data, and forecast data; Decision generation module: responsible for finding the static disaster entities with high similarity in attributes between the current event dynamic evolution knowledge graph and the historical event knowledge graph, and retrieving the emergency response measures for the current and subsequent disasters in the historical event knowledge graph according to the static disaster entities with high similarity; Decision execution monitoring module: responsible for monitoring the execution of the decision-making plan, collecting feedback information, and transmitting the feedback information to other modules for analysis and processing; User interaction module: responsible for providing a visual operation interface for decision-makers, displaying disaster information and emergency decision-making plans, and supporting manual modification.

Images