Metagroup intelligent information processing method and system

Abstract

The invention relates to the technical field of information processing, in particular to a meta-group intelligent information processing method and system, and the method comprises the steps: obtaining an incident place according to event information; determining alternative places according to a preset decision strategy on the basis of the incident place; obtaining monitoring network data corresponding to the alternative places, and performing hierarchical integration to obtain key format statements; processing the key format statement through a pre-training model to obtain a pre-trained key format statement; screening the pre-trained key format statements according to event information and a screening principle to obtain a data set; generating a resource allocation scheme according to the data set based on constraint conditions; and carrying out fitness optimization on the resource allocation scheme by utilizing a genetic algorithm to obtain a resource allocation scheme with the highest fitness. The problem of information island and the problem of low timeliness of information analysis when the existing security system processes the security task are solved.

Description

Technical Field

[0001] The present invention relates to the field of information processing technology, and in particular to a meta-swarm intelligent information processing method and system. Background Art

[0002] The security system has a strong regional multi-group characteristic, that is, there are multiple security units with basically the same structure and goals in different areas. These security units perform functions such as information collection, event handling, emergency response, patrol and control, and operate independently and coordinate actions.

[0003] Because security work often requires collaboration across multiple regions and departments, information sharing and resource allocation between different units are crucial for improving security efficiency. In practice, each security unit employs different strategies and methods to address issues within its oversight area, tailored to the specific needs of the area they oversee. Security systems and human resources rely on regional or local databases and information management platforms, creating barriers to information sharing across regions and departments. The lack of unified standards and interfaces hinders information flow, creating "information silos" and hindering the efficiency of cross-regional collaboration and resource allocation.

[0004] In addition, since the current analysis method for security-related data is still mainly based on manual analysis, for some security activities with high timeliness requirements, the existing security system finds it difficult to quickly obtain information support that matches the security activities even when it aggregates data and information from multiple security units.

[0005] Therefore, how to achieve effective collaboration and information integration among different security units becomes the key to improving the overall response capability of the security system. Summary of the Invention

[0006] To solve one of the above problems, the present invention provides a meta-swarm intelligent information processing method and system.

[0007] To achieve the above objectives, the present invention provides, on one hand, a meta-swarm intelligent information processing method, comprising: obtaining event information corresponding to a security task, and obtaining a corresponding incident location based on the event information; determining an alternative location based on the incident location according to a preset decision-making strategy; obtaining monitoring network data corresponding to the alternative location, hierarchically integrating the event information and the monitoring network data to obtain a key format sentence, wherein the key format sentence is presented in a text modality in a preset format; processing the key format sentence using a pre-trained model to obtain a pre-trained key format sentence; filtering the pre-trained key format sentence based on the event information and a filtering principle to obtain a data set, wherein the filtering principle includes at least one of the following: a time correlation principle, a spatial correlation principle, and an event similarity principle; generating a manpower deployment plan based on the data set based on constraints, wherein the constraints include at least one of the following: a maximum manpower constraint for a single security task, a total manpower exhaustion constraint, and a maximum available manpower constraint for a single security officer; and optimizing the fitness of the manpower deployment plan using a genetic algorithm to obtain a manpower deployment plan with the highest fitness.

[0008] On the other hand, the present invention provides a meta-group intelligent information processing system, comprising: an event acquisition module for acquiring event information corresponding to a security task, and acquiring a corresponding incident location according to the event information; an intelligent decision-making module for determining an alternative location based on the incident location according to a preset decision-making strategy; an integration processing module for acquiring monitoring network data corresponding to the alternative location, hierarchically integrating the event information and the monitoring network data, and obtaining a key format sentence, wherein the key format sentence is presented in a text mode of a preset format; a pre-training module for processing the key format sentence through a pre-training model, and obtaining a pre-trained key format sentence. format statements; a screening module for screening the pre-trained key format statements according to the event information and screening principles to obtain a data set, wherein the screening principles include at least one of the following: time correlation principle, spatial correlation principle and event similarity principle; a manpower deployment module for generating a manpower deployment plan based on the data set based on constraint conditions, wherein the constraint conditions include at least one of the following: maximum manpower constraint for a single security task, total manpower exhaustion constraint and maximum available manpower constraint for a single security personnel; a meta-population optimization module for optimizing the fitness of the manpower deployment plan using a genetic algorithm to obtain a manpower deployment plan with the highest fitness.

[0009] The beneficial effect of the present invention is reflected in that the meta-group intelligent information processing method and system provided by the present invention solves the information island problem and the low timeliness of information analysis that exist in existing security systems when processing security tasks. The present invention screens out alternative locations based on the location of the incident, obtains data from the monitoring network of the alternative locations, and modally unifies the data. The data is quickly screened as supporting data for security activities, and at the same time, a manpower deployment plan is generated using the supporting data. The manpower deployment plan is quickly optimized and evaluated through a genetic algorithm to select the optimal manpower deployment plan. In addition, when screening alternative locations, a radiation location map with the location of the incident as the root node is first constructed, and the radiation locations are scored using the constructed possibility model. The radiation locations with the highest scores are used as alternative locations to quickly select the most relevant alternative locations and arrange security activities reasonably and quickly. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 A schematic diagram of the topological network structure of the security system provided in Example 1 of the present invention; Figure 2 This is a flow chart of the method for processing meta-swarm intelligent information provided in Example 1 of the present invention; Figure 3 This is a schematic diagram of the structure of the meta-group intelligent information processing system provided in Example 1 of the present invention. DETAILED DESCRIPTION

[0011] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0012] Example 1 This embodiment provides a method for processing meta-swarm intelligent information, which is applied to the security system to process security tasks. The topological network structure of the security system can be set up as a three-level node, such as Figure 1 As shown, the system includes unit nodes, routing nodes, and root nodes. Specifically, a security unit can serve as a unit node, the routing node can be the parent unit of a basic security unit, and the root node can be the parent unit of a basic security unit. Implementations can be adjusted based on the specific security system architecture.

[0013] like Figure 2 As shown, the meta-swarm intelligent information processing method of this embodiment may include the following steps: Step S101, obtain the event information corresponding to the security task, and obtain the corresponding location of the incident based on the event information; specifically, the security task is usually obtained through the unit node, and can be obtained through various forms and channels such as on-site, network, pictures, videos, and voice. The event information should at least include the location of the incident. Generally speaking, the event information should also include some basic information of the event, such as time, place, people, specific facts, etc.

[0014] In an optional embodiment, after step S101 and before step S106, the method further includes: evaluating the urgency of the security task to obtain the urgency level of the security task; searching and matching in the security database according to the security task to obtain a reference manpower allocation; specifically, after obtaining the security task, the staff will evaluate the urgency of the event (for example, urgent, emergency, and normal), and then match it with previous events in the event database to provide a reference value for the completion time of similar previous events. , Reference value of number of participants This information can then be used as a reference for similar events, such as the duration and number of participants. According to the security system architecture of the present invention, event acquisition and event assessment steps can generally be completed at a unit node (e.g., a basic security unit at the incident site).

[0015] Step S102 determines an alternative location based on the incident location and a preset decision-making strategy. Specifically, after receiving event information and conducting a preliminary assessment, the unit node can report the information to the routing node or root node, which then determines the alternative location based on the overall monitoring network deployment. This step of determining the alternative location in S102 can be performed by an intelligent decision-making module within the routing node or root node, or by another independent intelligent decision-making module.

[0016] In an optional implementation, step S102 may include the following sub-steps: Step S1021 counts the number of locations with the incident location as the center. These locations are defined as locations within a preset distance from the incident location where a monitoring system is installed. Specifically, these locations are locations where the target of the incident (generally, specific individuals involved in the security incident) may be located. The preset distance can be determined based on the target's mode of transportation. Since the primary focus is on collecting data from the monitoring network, locations covered by the monitoring network are primarily considered.

[0017] Step S1022: Construct a location map based on the incident location and radial locations. The location map contains point data and edge data. The point data includes the incident location and radial locations, and the edge data includes edges that directly or indirectly connect the incident location to the radial locations. Specifically, the edge data represents the routes between the radial locations. The probability of the event target appearing at different locations during different time periods after the incident is proportional to the length of the edge. The radial locations and routes between the locations can be integrated and analyzed using the location map.

[0018] In step S1023, a probability assessment model is established to evaluate the probability scores of all radiation locations based on the time of occurrence of the security task, the location map, and the target's movement speed. Specifically, the time of occurrence of the event can be used to determine the time period after the incident. The corresponding radiation location can be found in the location map based on the specific time period after the incident. The movement speed of the event target can also be taken into consideration to establish a probability assessment model for comprehensive evaluation.

[0019] Step S1024, determine the radiation locations with the highest probability scores in the preset number as alternative locations. Since the target of the event changes physically over time after the event occurs, in addition to obtaining the monitoring network data of the location where the event occurred, it is also possible to obtain the monitoring network data of the possible locations of the event target, and to establish alternative locations for information linkage and integration. The number of alternative locations can be determined based on the specific circumstances of the event (such as the total number of radiation locations, the specific time period after the event), or based on the number preset by the system. The set of established alternative locations can be expressed as ,in is the root node.

[0020] A specific implementation of the determination step S102 is given below: first, count the locations with monitoring systems around the incident location, and then generate a graph of these locations based on the shortest path distance with the incident location as the root node: , in, A collection of location nodes with a monitoring system , Indicates the location of the incident; Edge Set , This represents the edge connecting the incident location and other locations along the shortest path (not a straight-line distance). The probability of the event target appearing at different locations within different time periods after the incident is proportional to the length of the edge. For example, the probability of the event target appearing at a location directly connected to the incident location is highest within 1 hour after the incident, and the probability of the event target appearing at a location with a secondary connection to the incident location is highest after 1 hour. The following probability assessment model can be used to define the probability of the event target appearing as follows: , in, Indicates the probability score of the radiation location; is the straight-line distance calculation function, Represents calculation and the distance between them; The function is expressed as , The meaning of the function is to output the value on the left of the symbol "|" when the conditions listed on the right of the symbol "|" are met. Indicates the distance threshold. Specifically, it can be obtained by multiplying the time by the preset moving speed, which can be obtained based on the description of the vehicle in the event information. Assign each radiation location and take the radiation location with higher score as the candidate location set , and output the set of candidate locations. The default is the root node. In the specific implementation process, the root node can be preset as the set of candidate locations. ; You can also include the root node in the radiation location by default, and give the root node a higher probability score by default during calculation.

[0021] Step S103: Acquire monitoring network data corresponding to the candidate locations, hierarchically integrate the event information and monitoring network data, and obtain key format statements, where the key format statements are presented in a pre-formatted text mode. Specifically, in the security topology, each candidate location belongs to a specific node. Based on the acquired set of candidate locations, data from the monitoring networks corresponding to these nodes can be acquired and hierarchically integrated through a data integration platform. The monitoring network can include devices and platforms such as smart surveillance cameras and social media, responsible for collecting various security-related data. The data from the monitoring network may generally include text, voice, images, video, and other types. To facilitate subsequent use of this information, the information can be integrated into textual data.

[0022] In one optional embodiment, the key format statement includes at least one of the following key information: time, location, target, target characteristics, and target behavior. Key information should be capable of identifying a specific event within the monitored network data. Key information such as time, location, target, target characteristics, and target behavior can be used to roughly characterize the event within the monitored network data.

[0023] In an optional embodiment, hierarchical integration of event information and monitoring network data includes: performing secondary integration of event information and each monitoring network data to obtain text modal data, wherein the secondary integration refers to converting the modalities of event information and monitoring network data into text modalities; performing primary integration of text modal data, wherein the primary integration refers to extracting key information from event information and monitoring network data, and using multiple components of the key information to form key format statements in Chinese word order. Specifically, if the monitoring network data and event information are stored in text modalities, the key information can be directly extracted without conversion; if the monitoring network information or event information is stored in non-text modalities, information conversion is required before extracting the key information.

[0024] In a specific implementation, a data integration platform can be established to complete the first-level integration and the second-level integration. The first-level integration can be set at the root node, and the second-level integration can be set at the routing node. The data after the second-level integration is uploaded to the root node for the first-level integration. After the integration is completed, the data is backed up and the data after the routing node integration is transmitted back. The integrated data includes the original event information and the data collected by the unit node. The second-level integration can refer to the modality unification after the routing node receives the data collected by the unit node. Common modalities of security data include RGB image modality, audio modality, text modality, etc. Since data of different modalities have great differences in storage format and analysis method, it is necessary to extract key information from these multimodal data first. The key information concerned by the present invention includes: time, location, task, target, target characteristics, target behavior, etc. For RGB image modality data, the image frame can be analyzed by combining the target detection algorithm with the human behavior recognition algorithm to obtain the target, target characteristics and target behavior, etc., and output in the form of this article; for audio data, the corresponding text data can be output by an audio-to-text model such as Whisper, and the text modality data does not need to be additionally converted during the second-level integration. The first-level integration is responsible for integrating the second-level integration data transmitted by the routing node. As can be seen from the above, the modalities of all data in the second-level integration are unified into textual modalities. What the first-level integration needs to do is to sort out key information such as time, location, target, target characteristics, target behavior, etc. from the text data, and construct them into preset format sentences based on Chinese grammar and reading habits. For example, the key format sentence in the embodiment of the present invention can adopt the following format: "with ( Target Features )of( Target )At( time )exist( Place ) was carried out ( Target behavior )", where "target characteristics", "target", "time", "location", and "target behavior" in the brackets are variables of this format statement. Through this format statement, the connotation of the events recorded in the monitored network data can be briefly summarized.

[0025] Step S104, processing the key format statement through the pre-trained model to obtain the pre-trained key format statement; specifically, the key format statement can be processed by the routing node, for example, it can be embedded through the pre-trained model, and the pre-processing model can be a Transformer (a deep learning model architecture based on the self-attention mechanism), Bert (Bidirectional Encoder Representations from Transformers, a pre-trained language representation model based on the Transformer architecture), etc., so as to achieve pre-processing of the key information of the event information and the key information of the remaining data.

[0026] Step S105, filter the pre-trained key format sentences according to the event information and the filtering principle to obtain a data set, wherein the filtering principle includes at least one of the following: time correlation principle, space correlation principle and event similarity principle; in an optional embodiment, the time correlation principle means: retain the pre-trained key format sentences with overlapping time; the space correlation principle means: retain the pre-trained key format sentences with overlapping location; the event similarity principle means: retain the pre-trained key format sentences with similarity higher than the threshold for target, target behavior and target feature. Specifically, the monitoring network data of the alternative locations can be filtered based on the event information, and the information with a high correlation with the event information can be retained. The specific filtering principles can include the following interpretations: (1) certain retention of the data that coincides with the time of the event involved in the event information; (2) certain retention of the data that coincides with the location of the event involved in the event information; (3) comparison of the similarity between the target behavior, target and target feature of the event information and the monitoring network data of each alternative location, and the data corresponding to the alternative location that is greater than the threshold is retained, and the threshold value can be set according to the specific event situation. The data that meets the conditions is retained using the filtering principle to obtain a data set. , the dataset DThis data can serve as supporting data for security. Using screening principles, key statements related to security tasks are identified. Events occurring at candidate locations corresponding to these key statements may be related to the actual event, necessitating the dispatch of security personnel to investigate these relevant candidate locations and the corresponding events. This screening eliminates irrelevant key statements and, consequently, candidate locations, reducing subsequent security workload and improving security efficiency. The more key statements contained in the filtered dataset, the more candidate locations related to the event, potentially requiring more manpower.

[0027] In step S106, a manpower allocation plan is generated based on the data set based on constraints. Constraints include at least one of the following: a maximum manpower constraint for a single security task, a total manpower utilization constraint, and a maximum available manpower constraint for a single security officer. Specifically, once the data set is determined, information about the incident at the incident location and related events at other alternative locations can be obtained from the data set, allowing for manpower allocation. Certain constraints must be adhered to during manpower allocation. The maximum manpower constraint for a single security task limits the total manpower allocated to a single security task, ensuring a balanced distribution of security manpower and preventing excessive manpower allocated to a single task from impacting the normal operation of other tasks. The total manpower utilization constraint indicates that the security system has a limited manpower capacity, requiring that manpower allocations meet the upper limit of the total manpower capacity. The maximum available manpower constraint for a single security officer ensures that the manpower allocated to a single security officer (i.e., the individual security officer's working hours) does not exceed reasonable limits to prevent fatigue.

[0028] In one optional embodiment, generating a manpower allocation plan based on a dataset based on constraints includes: generating a manpower allocation plan based on the dataset, the urgency level, and a reference manpower allocation based on the constraints. Specifically, in a previous embodiment, upon receiving event information, the event information can be assessed for urgency, searched and matched against a historical security database, and referenced against historical event records. Therefore, in this embodiment, the manpower allocation plan should also consider the urgency level and the reference manpower allocation, thereby rationally allocating manpower to security tasks and avoiding insufficient or wasted human resources.

[0029] In a specific implementation, step S106 can be achieved as follows: after the above information analysis is completed, the routing node performs manpower allocation optimization and sets an optimization goal. The optimization goal is to maximize the manpower allocation obtained by high-priority tasks and ensure the rational use of manpower. The optimization objective function Z is: , in, i Indicates the task number.j Indicates the security manpower serial number, It's a task Priority, Assigned to the task of manpower.

[0030] The constraints can be set by the following formula: (1) The manpower allocation for each task cannot exceed the maximum manpower required, but must meet the minimum requirements: , in, Indicates the minimum manpower requirement, A is the number of alternative locations, is the average distance between alternative locations, is the amount of manpower required per unit distance; Indicates the The maximum number of manpower required for a task is equal to , is the proportion of working hours in a day, T The reference value of the event completion time (in days) for similar events. N is a reference value for the number of participants (security personnel) in similar incidents. This shows that this constraint sets the upper and lower limits for the manpower required for a single task. The upper limit is determined by the completion time and number of participants in previous similar incidents, while the lower limit is determined by the number of alternative locations and their distance.

[0031] (2) The total manpower required for all tasks cannot exceed the available human resources: , in, The total number of manpower.

[0032] (3) The working hours of each security personnel shall not exceed their maximum working hours. , that is, each security personnel The total manpower assigned across all tasks cannot exceed their work hour limit: , in, equal .

[0033] Step S107 , using a genetic algorithm to optimize the fitness of the human resource deployment plan, and obtaining a human resource deployment plan with the highest fitness; specifically, by optimizing the human resource deployment plan, the efficiency of the human resource deployment plan can be further improved.

[0034] In an optional embodiment, step S107 may include the following steps: converting each labor deployment plan into a matrix; using a genetic algorithm to converge the fitness value of each matrix; and selecting the labor deployment plan corresponding to the matrix with the highest fitness value as the labor deployment plan with the highest fitness. By using the genetic algorithm to evaluate fitness, each labor deployment plan can be treated as an individual to form a population, and the individuals can be selected through a survival of the fittest process to find the optimal individual solution.

[0035] In a specific embodiment, after the aforementioned constraint conditions are constructed, the problem can be optimized using a genetic algorithm. Step S107 can be implemented by the following specific steps: (1) Let the manpower deployment plan be individuals, and each individual represents a possible manpower deployment plan. An individual can be represented by a length of An array or matrix representing the number of manpower allocated to each security personnel on each task, individual Expressed as: ; (2) The fitness function is used to evaluate the quality of individuals. It can be expressed as the optimization objective function The value of , that is: , ; (3) Assume that there is individuals, that is, the population is , then the individual Probability of being selected for: ， in, represents the total fitness of the population, Represents an individual Adaptability; (4) Calculate the cumulative probability : , Generate a random number , choose to meet The individual is selected as the individual; (5) Select an intersection point , select two individuals from the selected individuals to divide the human resources allocation plan into two parts, and then exchange the two parts. Assume that the individuals are and , then the subpopulation after crossover is: , ; (6) Then use the gene mutation operation to randomly change an individual's Human resource allocation , for example, randomly selecting tasks and security personnel , then change the manpower allocation value The value of , so that it falls within the legal range. The current value is Within the range, after mutation Can be A new value in the range, the probability of mutation is a small value, usually between 0.01 and 0.1. For example, if , then each individual has a 10% probability of mutation; (7) Repeat steps (2) to (6) until the value of Z converges, and take the human resource allocation plan with the highest individual fitness as the optimal solution.

[0036] The meta-swarm intelligent information processing method provided by this embodiment solves the information island problem and the low timeliness of information analysis that exist in existing security systems when processing security tasks. This embodiment selects alternative locations based on the location of the incident, obtains data from the monitoring network of the alternative locations, and modally unifies the data. The data is quickly screened as supporting data for the security task, and the supporting data is used to generate a manpower deployment plan. The manpower deployment plan is quickly optimized and evaluated through a genetic algorithm to select the optimal manpower deployment plan. In addition, when screening alternative locations, a radiation location map with the location of the incident as the root node is first constructed, and the radiation locations are scored using the constructed possibility model. The radiation locations with the highest scores are selected as alternative locations to quickly select the most relevant alternative locations and arrange security activities reasonably and quickly.

[0037] This embodiment also provides a meta-swarm intelligent information processing system for executing the aforementioned meta-swarm intelligent information processing method. The specific technical features of the system have been described in detail in the method, and will not be repeated here. Only the architecture of the system will be described. Figure 3 As shown, the system includes multiple functional modules, which can be implemented by independent chips, devices or equipment, or multiple functional modules can be integrated into one chip, device or equipment. The system includes the following functional modules: Event acquisition module 301 is used to obtain event information corresponding to security tasks and, based on this information, to determine the corresponding incident location. Specifically, security tasks are typically acquired through unit nodes, meaning event acquisition module 301 is typically located within the unit node. This information can be acquired through various channels, including on-site, online, through images, videos, and audio. Event information should at least include the incident location and, generally, also include basic information about the event, such as time, location, people involved, and specific facts.

[0038] Intelligent decision-making module 302 is used to determine alternative locations based on the incident location and a preset decision-making strategy. Specifically, after receiving event information and conducting a preliminary assessment, the unit node can report the information to the routing node or root node, which then determines the alternative location based on the overall monitoring network deployment. Intelligent decision-making module 302 can be located within the routing node or root node, or in another independent device.

[0039] In an optional embodiment, the location of the incident is used as a basis to determine alternative locations according to a preset decision-making strategy, including: counting radiation locations with the location of the incident as the center, wherein the radiation location refers to a location with a monitoring system within a preset distance from the location of the incident; constructing a location map based on the location of the incident and the radiation locations, the location map contains point data and edge data, the point data includes the location of the incident and the radiation locations, and the edge data includes edges directly or indirectly connected between the location of the incident and the radiation locations; establishing a possibility assessment model to evaluate the possibility scores of all radiation locations based on the occurrence time of the security task, the location map and the target movement speed; determining a preset number of radiation locations with the highest possibility scores as alternative locations. Specifically, since after the incident occurs, the target of the incident undergoes physical location changes with time, therefore, in addition to obtaining the monitoring network data of the location of the incident, it is also possible to obtain the monitoring network data of the possible locations of the target of the incident, and to establish alternative locations for information linkage and integration. The established set of alternative locations can be expressed as ,in is the root node.

[0040] Integration processing module 303 is used to obtain monitoring network data corresponding to candidate locations, hierarchically integrate event information and monitoring network data, and obtain key format statements, where the key format statements are presented in a pre-formatted text mode. Specifically, in the security topology, each candidate location belongs to a specific node. Based on the obtained set of candidate locations, monitoring network data corresponding to the nodes of these candidate locations can be obtained and hierarchically integrated through the data integration platform. The monitoring network can include devices and platforms such as smart surveillance cameras and social media, responsible for collecting various security-related data. The data from the monitoring network may generally include text, voice, images, video, and other types. To facilitate subsequent use of this information, the information can be integrated into textual data.

[0041] In one optional embodiment, the key format statement includes at least one of the following key information: time, location, target, target characteristics, and target behavior. Key information should be capable of identifying a specific event within the monitored network data. Key information such as time, location, target, target characteristics, and target behavior can be used to roughly characterize the event within the monitored network data.

[0042] In an optional embodiment, hierarchical integration of event information and monitoring network data includes: performing secondary integration of event information and each monitoring network data to obtain text modal data, wherein the secondary integration refers to converting the modalities of event information and monitoring network data into text modalities; performing primary integration of text modal data, wherein the primary integration refers to extracting key information from event information and monitoring network data, and using multiple components of the key information to form key format statements in Chinese word order. Specifically, if the monitoring network data and event information are stored in text modalities, the key information can be directly extracted without conversion; if the monitoring network information or event information is stored in non-text modalities, information conversion is required before extracting the key information.

[0043] The pre-training module 304 is used to process the key format statements through the pre-training model to obtain the pre-trained key format statements. Specifically, the key format statements can be processed by the routing node. For example, they can be embedded through the pre-training model. The pre-processing model can be a Transformer (a deep learning model architecture based on the self-attention mechanism), Bert (Bidirectional Encoder Representations from Transformers, a pre-trained language representation model based on the Transformer architecture), and other models, thereby achieving pre-processing of the key information of the event information and the key information of the remaining data.

[0044] The screening module 305 is used to screen the pre-trained key format sentences according to the event information and the screening principle to obtain a data set, wherein the screening principle includes at least one of the following: time correlation principle, space correlation principle and event similarity principle; in an optional embodiment, the time correlation principle means: retaining the pre-trained key format sentences with overlapping time; the space correlation principle means: retaining the pre-trained key format sentences with overlapping location; the event similarity principle means: retaining the pre-trained key format sentences with similarity of target, target behavior and target feature higher than the threshold. Specifically, the monitoring network data of the alternative location can be screened based on the event information, and the information with a high correlation with the event information can be retained. The specific screening principles can include the following interpretations: (1) certain retention of the information that coincides with the time of the event involved in the event information; (2) certain retention of the information that coincides with the location of the event involved in the event information; (3) comparison of the similarity of the target behavior, target and target feature of the event information and the monitoring network data of each alternative location, and the data corresponding to the alternative location that is greater than the threshold is retained, and the threshold value can be set according to the specific event situation. Using a screening principle, data that meets the criteria is retained, resulting in a dataset. This dataset, D, serves as supporting data for the security task. This screening principle identifies key format statements relevant to the security task. These key format statements correspond to events occurring at candidate locations that are likely to be related to the actual event, necessitating the dispatch of security personnel to investigate these relevant candidate locations and the corresponding events. This screening eliminates irrelevant key format statements, thereby eliminating irrelevant candidate locations, reducing subsequent security workload and improving security efficiency. The more key format statements contained in the filtered dataset, the more candidate locations relevant to the event, potentially requiring more personnel.

[0045] The manpower allocation module 306 is configured to generate a manpower allocation plan based on the data set based on constraints. Constraints include at least one of the following: a maximum manpower constraint for a single security task, a total manpower utilization constraint, and a maximum available manpower constraint for a single security officer. Specifically, once the data set is determined, information about the incident at the incident location and related events at other alternative locations can be obtained from the data set, allowing for manpower allocation. Certain constraints must be adhered to during manpower allocation. The maximum manpower constraint for a single security task limits the total manpower allocated to a single security task, ensuring a balanced distribution of security manpower and preventing excessive manpower allocated to a single task from impacting the normal operation of other tasks. The total manpower utilization constraint indicates that the security system has a limited manpower capacity, requiring that manpower allocations meet the upper limit of the total manpower capacity. The maximum available manpower constraint for a single security officer ensures that the manpower allocated to a single security officer (i.e., the individual security officer's working hours) does not exceed reasonable limits to prevent fatigue.

[0046] The meta-population optimization module 307 is used to optimize the fitness of the human resource deployment plan using a genetic algorithm to obtain the human resource deployment plan with the highest fitness. Specifically, by optimizing the human resource deployment plan, the efficiency of the human resource deployment plan can be further improved.

[0047] In one optional implementation, a genetic algorithm is used to optimize the fitness of human resource deployment plans, including: converting each human resource deployment plan into a matrix; using the genetic algorithm to converge the fitness value of each matrix; and selecting the human resource deployment plan corresponding to the matrix with the highest fitness value as the human resource deployment plan with the highest fitness. By evaluating fitness through the genetic algorithm, each human resource deployment plan can be treated as an individual to form a population, and the individuals can be selected through a survival of the fittest process to find the optimal individual solution.

[0048] The meta-swarm intelligent information processing system provided by this embodiment solves the information island problem and the low timeliness of information analysis that exist in existing security systems when processing security tasks. This embodiment selects alternative locations based on the location of the incident, obtains data from the monitoring network of the alternative locations, and modally unifies the data. The data is quickly screened as supporting data for the security task, and the supporting data is used to generate a manpower deployment plan. The manpower deployment plan is quickly optimized and evaluated through a genetic algorithm to select the optimal manpower deployment plan. In addition, when screening alternative locations, a radiation location map with the location of the incident as the root node is first constructed, and the radiation locations are scored using the constructed possibility model. The radiation locations with the highest scores are selected as alternative locations to quickly select the most relevant alternative locations and arrange security activities reasonably and quickly.

[0049] In an optional embodiment, the meta-swarm intelligent information processing system of this embodiment further includes: an evaluation module 308; the evaluation module 308 is configured to evaluate the urgency of the security task before generating a manpower allocation plan based on the data set based on the constraints, obtain the urgency level of the security task, and search and match the security task in the security database to obtain a reference manpower allocation; and the manpower allocation module 306 generates a manpower allocation plan based on the data set based on the constraints, including: the manpower allocation module 306 generates a manpower allocation plan based on the constraints based on the data set, the urgency level, and the reference manpower allocation. Specifically, after obtaining a security task, security personnel will evaluate the urgency of the event (e.g., "super urgent," "urgent," or "normal"), then match the event against previous events in the event database to provide information such as a reference completion time and number of participants for similar previous events. The urgency level and the reference manpower allocation (such as the reference completion time and number of participants for similar previous events) should also be considered when making subsequent manpower allocation plans, thereby rationally allocating manpower to the security task and avoiding insufficient or wasted manpower. According to the setting of the security system architecture of the present invention, the evaluation module 308 can generally be set at a unit node (such as a basic security unit at the incident site).

[0050] In the description of the embodiments of the present invention, it should be understood that the terms "upper", "lower", "front", "back", "left", "right", "vertical", "horizontal", "center", "top", "bottom", "top", "bottom", "inside", "outside", "inner side", "outer side" and the like indicate orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. Among them, "inside" refers to an internal or enclosed area or space. "Periphery" refers to the area surrounding a specific component or specific area.

[0051] In the description of the embodiments of the present invention, the terms "first," "second," "third," and "fourth" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly specifying the number of the technical features indicated. Therefore, a feature specified as "first," "second," "third," or "fourth" may explicitly or implicitly include one or more of the features. In the description of the present invention, unless otherwise specified, "plurality" means two or more.

[0052] In the description of the embodiments of the present invention, it should be noted that, unless otherwise expressly specified or limited, the terms "installed," "connected," "connected," and "assembled" should be understood in a broad sense. For example, they may refer to fixed connections, detachable connections, or integral connections; they may refer to direct connections, indirect connections through an intermediate medium, or internal connections between two components. Those skilled in the art will understand the specific meanings of the above terms in the present invention in specific circumstances.

[0053] In the description of the embodiments of the present invention, specific features, structures, materials or characteristics may be combined in an appropriate manner in any one or more embodiments or examples.

[0054] In describing the embodiments of the present invention, it should be understood that "-" and "~" represent a range between two values, and the range includes the endpoints. For example, "AB" represents a range greater than or equal to A and less than or equal to B. "A~B" represents a range greater than or equal to A and less than or equal to B.

[0055] In describing the embodiments of the present invention, the term "and / or" is used herein to describe a relationship between associated objects, indicating that three possible relationships exist. For example, "A and / or B" can represent three situations: A exists alone, A and B exist simultaneously, and B exists alone. Furthermore, the character " / " is generally used herein to indicate that the associated objects are in an "or" relationship.

[0056] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A method for processing meta-swarm intelligent information, characterized in that: include: Obtaining event information corresponding to the security task, and obtaining the corresponding incident location based on the event information; Determining an alternative location based on the incident location according to a preset decision-making strategy; Acquiring monitoring network data corresponding to the candidate location, hierarchically integrating the event information and the monitoring network data to obtain a key format sentence, wherein the key format sentence is presented in a text modality of a preset format; Processing the key format sentence through the pre-trained model to obtain a pre-trained key format sentence; The pre-trained key format sentences are screened according to the event information and a screening principle to obtain a data set, wherein the screening principle includes at least one of the following: a time correlation principle, a spatial correlation principle, and an event similarity principle; Generating a resource allocation plan based on the data set based on constraints, wherein the constraints include at least one of the following: a maximum manpower constraint for a single security task, a total manpower exhaustion constraint, and a maximum available manpower constraint for a single person; A genetic algorithm is used to optimize the fitness of the human resource deployment plan to obtain a human resource deployment plan with the highest fitness.

2. The method for processing meta-swarm intelligent information according to claim 1, characterized in that: Before the step of generating a manpower deployment plan according to the data set based on the constraint conditions, the method further includes: Evaluate the urgency of the security task to obtain the urgency level of the security task; Search and match the security tasks in the security database to obtain reference manpower allocation; Generating the manpower deployment plan according to the data set based on the constraint conditions includes: generating the manpower deployment plan according to the data set, the emergency level and the reference manpower allocation based on the constraint conditions.

3. The method for processing meta-swarm intelligent information according to claim 1, characterized in that: The step of determining an alternative location based on the incident location and a preset decision-making strategy includes: Counting radiation points with the incident site as the center, wherein the radiation points refer to places with monitoring systems within a preset distance from the incident site; Constructing a location map based on the incident location and the radiation locations, wherein the location map includes point data and edge data, wherein the point data includes the incident location and the radiation locations, and the edge data includes edges directly or indirectly connecting the incident location and the radiation locations; Establishing a possibility assessment model to assess the possibility scores of all the radiation locations based on the occurrence time of the security task, the location map, and the target movement speed; The radiation locations with the highest preset probability scores are determined as the candidate locations.

4. The method for processing meta-swarm intelligent information according to claim 1, wherein: The key format sentence includes at least one of the following key information: time, place, target, target characteristics and target behavior.

5. The method for processing meta-swarm intelligent information according to claim 4, characterized in that: The hierarchical integration of the event information and the monitoring network data includes: Performing secondary integration on the event information and each of the monitoring network data to obtain text modality data, wherein the secondary integration refers to converting the modalities of the event information and the monitoring network data into text modality; The text modal data is subjected to a first-level integration, wherein the first-level integration refers to extracting the key information from the event information and the monitoring network data, and utilizing multiple components of the key information to form the key format sentence in a Chinese word order expression.

6. The method for processing meta-swarm intelligent information according to claim 1, characterized in that: The fitness of the manpower deployment plan is optimized by using a genetic algorithm to obtain a manpower deployment plan with the highest fitness, including: Convert each of the manpower deployment plans into a matrix; Using a genetic algorithm to make the fitness value of each matrix converge; The human resource deployment plan corresponding to the matrix with the highest fitness value is taken as the human resource deployment plan with the highest fitness value.

7. A meta-swarm intelligent information processing system, characterized in that: include: An event acquisition module is used to obtain event information corresponding to a security task and obtain the corresponding location of the incident based on the event information; An intelligent decision-making module, configured to determine an alternative location based on the location of the incident according to a preset decision-making strategy; an integration processing module, configured to obtain monitoring network data corresponding to the candidate location, hierarchically integrate the event information and the monitoring network data, and obtain a key format sentence, wherein the key format sentence is presented in a text mode of a preset format; A pre-training module, configured to process the key format sentence using a pre-training model to obtain a pre-trained key format sentence; a screening module, configured to screen the pre-trained key format sentences according to the event information and a screening principle to obtain a data set, wherein the screening principle includes at least one of the following: a time correlation principle, a spatial correlation principle, and an event similarity principle; A manpower deployment module, configured to generate a manpower deployment plan based on the data set based on constraints, wherein the constraints include at least one of the following: a maximum manpower constraint for a single security task, a total manpower exhaustion constraint, and a maximum available manpower constraint for a single security personnel; The meta-population optimization module is used to optimize the fitness of the human resource deployment plan by using a genetic algorithm to obtain the human resource deployment plan with the highest fitness.

8. The meta-swarm intelligent information processing system according to claim 7, characterized in that: Also includes: Assessment module; The evaluation module is used to evaluate the urgency of the security task before generating the manpower deployment plan based on the constraint conditions according to the data set to obtain the urgency level of the security task, and search and match the security task in the security database to obtain a reference manpower allocation; The manpower deployment module generates a manpower deployment plan according to the data set based on the constraint conditions, including: the manpower deployment module generates the manpower deployment plan according to the data set, the emergency level and the reference manpower allocation based on the constraint conditions.

9. The meta-swarm intelligent information processing system according to claim 7, characterized in that: The step of determining an alternative location based on the incident location and a preset decision-making strategy includes: Counting radiation points with the incident site as the center, wherein the radiation points refer to places with monitoring systems within a preset distance from the incident site; Constructing a location map based on the incident location and the radiation locations, wherein the location map includes point data and edge data, wherein the point data includes the incident location and the radiation locations, and the edge data includes edges directly or indirectly connecting the incident location and the radiation locations; Establishing a possibility assessment model to assess the possibility scores of all the radiation locations based on the occurrence time of the security task, the location map, and the target movement speed; The radiation locations with the highest preset probability scores are determined as the candidate locations.

10. The meta-swarm intelligent information processing system according to claim 7, characterized in that: The key format sentence includes at least one of the following key information: time, place, target, target characteristics and target behavior.

Images