A fire management system based on artificial intelligence

By introducing artificial intelligence and fire protection knowledge graphs into the fire protection management system, the problem of slow task allocation and imbalance in the existing system is solved, and the efficiency of fire protection task execution is improved.

CN118710016BActive Publication Date: 2024-11-08SICHUAN JINWEI FIRE PROTECTION TECHNOLOGY CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411178915.1
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-08-27
Publication Date
2024-11-08
Estimated Expiration
2044-08-27

AI Technical Summary

Technical Problem

The task allocation speed and imbalance in the existing fire protection management system are slow, resulting in difficulty in execution of assigned personnel and affecting the efficiency of fire protection safety and maintenance.

Method used

Adopt a fire protection management system based on artificial intelligence, dynamically allocate fire protection tasks, use fire protection knowledge graphs to map content, and feedback to the executive staff to improve task execution efficiency. The system includes a collection unit, a construction unit, a task allocation unit, a task evaluation unit and an interaction unit, dynamically assign tasks and evaluate execution, and generate suggestions information using a multi-objective optimization model and fire protection knowledge graph.

Benefits of technology

The allocation and execution efficiency of fire protection tasks has been improved. By rationally allocating fire protection tasks, the task execution efficiency of personnel is improved, the task execution situation can be more intuitively displayed, and the shortcomings in task execution can be corrected in a timely manner.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118710016B_ABST
    Figure CN118710016B_ABST
Patent Text Reader

Abstract

The present application discloses a fire management system based on artificial intelligence, which relates to the field of fire management technology. The system includes: a collection unit, which obtains first target information based on a data collection device deployed at a target location; a construction unit, which is used to construct a fire knowledge graph; a task allocation unit, which is used to generate a first target task and dynamically allocate fire tasks to be performed by target personnel; a task evaluation unit, which determines the fire task execution status of the target personnel based on an evaluation model and generates sub-evaluation information; based on the sub-evaluation information, generates evaluation information of the target location; an interaction unit, which is used to generate suggestion information based on the evaluation information and the sub-evaluation information using the fire knowledge graph and send it to the target personnel; the present application combines real-time monitoring data and the fire knowledge graph, optimizes the fire safety task allocation strategy according to the information of fire management personnel, and improves the efficiency of task execution.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the technical field of fire management, and in particular to a fire management system based on artificial intelligence. Background Art

[0002] With the acceleration of urbanization, densely packed buildings such as high-rise buildings, large commercial complexes, and industrial parks are increasing, and the complexity and challenges of fire safety management have increased significantly. The current task allocation in the fire management system mostly relies on manual operation and experience judgment, and faces problems such as slow allocation, unbalanced allocation, and difficulty in execution by the assigned personnel. This makes the fire safety maintenance personnel of various organizations inefficient in performing fire safety tasks. These problems directly affect the safety of personnel's lives and the protection of property. Therefore, it is particularly important to develop an intelligent fire safety management system that can achieve automated management and real-time response.

[0003] For example, the Chinese patent application with publication number CN117575285A discloses an artificial intelligence-based fire protection smart information management platform, which involves the field of intelligent fire protection technology. The platform includes: a fire protection information management module for obtaining fire protection information and generating a fire protection information set; a fire protection equipment management module for obtaining fire protection equipment and generating an equipment information set; an information coordination and scheduling module for constructing a coordinated scheduling model based on the fire protection information set and the equipment information set; a fire protection plan layout module for generating fire protection emergency plans and fire protection operation diagrams based on the coordinated scheduling model; and a remote human-computer interaction module for generating human-computer interaction instructions and performing human-computer interaction based on the coordinated scheduling model. The invention uses a fire protection big data application platform to carry out real-time intelligent fire safety monitoring, evaluation and early warning, and realize the collection and sharing of fire protection data; relieve the pressure on the duty room through remote monitoring and real-time information management; and improve the timeliness of fire prevention and rescue by generating fire protection emergency plans and fire protection operation diagrams.

[0004] The above method has the problems raised by the background technology. The system may face problems such as slow task allocation, unbalanced allocation, and difficulty in execution by the assigned personnel. Summary of the invention

[0005] In response to the shortcomings of the existing technology, the present application discloses a fire management system based on artificial intelligence, which dynamically assigns tasks based on the information of the executors and evaluates the task execution. For the deficiencies in the evaluation, the fire knowledge graph is used for content mapping and feedback is provided to the executors, thereby improving the efficiency of personnel allocation and execution.

[0006] The embodiment of the present application discloses a fire management system based on artificial intelligence, which includes:

[0007] A collection unit, used to obtain first target information of the target location based on a data collection device deployed at the target location; the collection unit is also used to obtain second target information for constructing a fire protection knowledge graph;

[0008] A construction unit, configured to construct a fire protection knowledge graph based on the second target information;

[0009] A task allocation unit is used to generate a first target task based on the first target information; based on the first target task and a preset second target task, dynamically allocate the firefighting task to be performed by the target personnel of the target place; the firefighting task includes: the first target task and the second target task;

[0010] A task evaluation unit, used to determine the firefighting task execution status of the target personnel based on a pre-built evaluation model, and generate sub-evaluation information; based on the sub-evaluation information, generate evaluation information of the target location;

[0011] An interactive unit is used to generate suggestion information based on the evaluation information and the sub-evaluation information using the fire protection knowledge graph, and send the suggestion information to the target person.

[0012] As an optional implementation, the second target information includes: fire regulations data, fire equipment management data, and emergency plan data; and constructing the fire knowledge graph based on the second target information includes:

[0013] Set the node set to V, the relationship set to E, and represent the fire protection knowledge graph as a directed graph G=(V,E);

[0014] Among them, the node set , each Represents a knowledge node; a set of relationships , each Representation Node and nodes The relationship between.

[0015] As an optional implementation manner, generating the first target task based on the first target information includes:

[0016] There is at least one preset abnormal target information in the first target information, and based on the fire protection knowledge graph, the first target task corresponding to the abnormal target information is generated.

[0017] As an optional implementation manner, the task allocation unit is further used for:

[0018] Determining the priority of each firefighting task based on the urgency and importance of the firefighting task;

[0019] Acquire the personnel information of each target person in the target location, and the skill information corresponding to each firefighting task; wherein the personnel information includes: position information, personal skill information, location information, and workload;

[0020] Determining the skill matching degree of each of the target personnel for the firefighting task based on the personal skill information and the skill information corresponding to the firefighting task;

[0021] Based on the priority of the firefighting task, the skill matching degree, the workload, the location information and the task completion time, a multi-objective optimization model is constructed, and a first task allocation plan is generated by solving the optimization strategy.

[0022] As an optional implementation manner, the task allocation unit is further used for:

[0023] Set the first objective function , the first objective function is the sum of the urgency and importance of the firefighting task; the second objective function is set , the second objective function is the sum of the skill matching of the target personnel for the firefighting task; the third objective function is set to , the third objective function is the target personnel workload difference; the fourth objective function is set , the fourth objective function is the sum of the distances between the task location and the target personnel location; the fifth objective function is set , the fifth objective function is the expected deviation of task completion time;

[0024] Based on the objective function, the multi-objective optimization model is constructed; the function expression of the multi-objective optimization model is:

[0025] ;

[0026] in, , , , as well as are respectively the weights of the first objective function, the second objective function, the third objective function, the fourth objective function, and the fifth objective function in the multi-objective optimization model at time t.

[0027] As an optional implementation manner, the task allocation unit is further used for:

[0028] Dynamically adjust weights , the adjustment formula is as follows:

[0029] ;

[0030] in, is the initial weight, is the adjustment factor, is at least one of a preset time variation and a preset event triggering amount, .

[0031] As an optional implementation manner, the task evaluation unit is further used for:

[0032] Obtaining the firefighting task completion information of the target personnel; wherein the firefighting task completion information includes: task completion rate, task completion time, task delay rate, task success rate, personnel workload uniformity and task response time;

[0033] Based on the firefighting task completion information of the target personnel, determine the firefighting task execution status of the target personnel, and generate percentage scoring information; wherein each type of information in the firefighting task completion information corresponds to a percentage scoring method;

[0034] Based on the percentage scoring information, the sub-evaluation information is generated; wherein each sub-evaluation information corresponds to one target person;

[0035] Based on the sub-evaluation information corresponding to each of the target persons in the target place, the evaluation information of the target place is determined by using a weighted average method.

[0036] As an optional implementation manner, the task evaluation unit is further used for:

[0037] In response to the target personnel's firefighting task completion information being updated, the sub-evaluation information is updated in real time based on the updated firefighting task completion information;

[0038] The evaluation information of the target place is updated by using a weighted average method based on the updated sub-evaluation information corresponding to each target person of the target place.

[0039] As an optional implementation manner, the determining the evaluation information of the target place by using a weighted average method based on the sub-evaluation information corresponding to each target person of the target place includes:

[0040] Based on the position information, the urgency and importance of the firefighting task, the weight of the sub-evaluation information corresponding to each of the target personnel is adjusted when using the weighted average method.

[0041] As an optional implementation, the evaluation information includes: information on insufficient completion of firefighting tasks;

[0042] The interaction unit is used for:

[0043] Based on at least one of the firefighting task completion insufficient information, mapping the firefighting task completion insufficient information to the firefighting knowledge graph; wherein the firefighting task completion insufficient information includes: task response speed insufficient information, task failure information, and task completion efficiency insufficient information;

[0044] Using the nodes and relationships in the fire protection knowledge graph, searching for a correction path;

[0045] Based on the corrected path search results, suggestion information is generated and sent to the target person.

[0046] Compared with the prior art, the beneficial effects of the present invention are as follows: the present application discloses a fire management system based on artificial intelligence, which utilizes a dynamic task allocation mechanism, which rationally allocates fire tasks based on personnel's position information, personal skill information, location information, workload and other information. A scoring method is disclosed for the task execution status of each person and the task execution status of the entire site, which can more intuitively display the task execution status of the current site. For the shortcomings in task execution, the present application improves the efficiency of personnel's task execution by mapping the fire knowledge graph. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 A schematic diagram of the architecture of an artificial intelligence-based fire management system provided in an embodiment of the present application;

[0048] Figure 2 An example diagram of sensor types deployed at a target location provided in an embodiment of the present application;

[0049] Figure 3 A structural framework diagram of the multi-objective optimization model provided in the embodiments of the present application.

[0050] Reference numerals:

[0051] 10. Collection unit; 11. Smoke sensor; 12. Temperature sensor; 13. Humidity sensor; 14. Carbon dioxide sensor; 20. Construction unit; 30. Task allocation unit; 40. Task evaluation unit; 50. Interaction unit. DETAILED DESCRIPTION

[0052] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0053] See also Figure 1 , Figure 1 A schematic diagram of the architecture of an artificial intelligence-based fire management system provided in an embodiment of the present application, the system includes:

[0054] The acquisition unit 10 is used to acquire first target information of the target location based on the data acquisition device deployed at the target location; the acquisition unit 10 is also used to acquire second target information for constructing a fire protection knowledge graph;

[0055] A construction unit 20, configured to construct the fire protection knowledge graph based on the second target information;

[0056] The task allocation unit 30 is used to generate a first target task based on the first target information; based on the first target task and a preset second target task, dynamically allocate the firefighting task to be performed by the target personnel of the target place; the firefighting task includes: the first target task and the second target task;

[0057] The task evaluation unit 40 is used to determine the fire task execution status of the target personnel based on the evaluation model and generate sub-evaluation information; based on a plurality of the sub-evaluation information, generate evaluation information of the target location;

[0058] The interaction unit 50 is used to generate suggestion information based on the evaluation information and the sub-evaluation information using the fire protection knowledge graph, and send the suggestion information to the target person.

[0059] The artificial intelligence-based fire management system provided in this application is applicable to a variety of target locations, including shopping malls, corporate buildings, administrative office buildings, campuses, and townships. For different target locations, this system can increase or decrease the system structure and functions according to the specific building structure, organizational structure, execution capability, etc. of the target location.

[0060] The data acquisition device includes a sensor for data acquisition. The present application collects real-time data uploaded by sensors deployed at a target location as first target information through an acquisition unit 10, and obtains second target information for constructing a fire safety knowledge graph through a target data source such as a database and a data interface; the construction unit 20 constructs a fire safety knowledge graph through the second target information obtained by the acquisition unit 10; the task allocation unit 30 detects and analyzes the content of the first target information after receiving the first target information obtained by the acquisition unit 10. If there is a preset emergency situation, a first target task corresponding to the emergency situation can be generated. In addition to the first target task The task assignment unit 30 will also preset a second target task for daily fire safety maintenance and management based on the site information of the target site, such as the specific building structure, organizational structure, and execution capabilities of the target site, and dynamically assign the first target task and the second target task to each target person in the target site; the execution of the fire tasks of each target person and the overall execution of the fire tasks of the target site will be evaluated by the task evaluation unit 40; the interaction unit 50 will, based on the evaluation results of the task evaluation unit 40 and using the fire knowledge graph constructed by the construction unit 20, send the fire safety information to the target person, and guide the target person to improve the completion of the fire tasks.

[0061] The target personnel may be the fire department in charge of the target location. In addition, the target location personnel other than the fire department may also be included, and the fire fighting task may also be assigned to them when an emergency occurs.

[0062] For the above-mentioned acquisition unit 10:

[0063] See also Figure 2 , Figure 2 An example diagram of sensor types deployed at a target location provided in an embodiment of the present application. In a specific implementation, it is necessary to deploy multiple types of sensors at the target location, and the sensor types include: a smoke sensor 11, a temperature sensor 12, a humidity sensor 13, and a carbon dioxide sensor 14. These sensors can monitor changes in environmental parameters in real time and promptly detect potential fire hazards. Specifically, the smoke sensor 11 can detect the smoke concentration in the air, the temperature sensor 12 monitors changes in ambient temperature, the humidity sensor 13 records air humidity, and the carbon dioxide sensor 14 monitors carbon dioxide concentration. Each sensor has specific parameter definitions in the acquisition unit 10, and these parameter definitions include: sensor type, specific value of the collected data, sensor unique identifier, sensor installation location, etc.

[0064] In the specific implementation, the sensor network is connected through the Internet of Things (IoT) platform to achieve real-time data transmission and processing. In order to ensure the efficiency and reliability of data transmission, the MQTT protocol can be used, which is a lightweight publish / subscribe message transmission protocol suitable for low-bandwidth, high-latency or unreliable network environments. The data collected by the sensor will be transmitted to the local edge computing device through the MQTT protocol. The edge computing device is responsible for preliminary processing of the data, such as data cleaning, filtering and preliminary analysis. For example, low-cost, high-performance hardware devices such as Raspberry Pi are deployed in the target location. These devices can process near the data source, reduce data transmission delays, and improve real-time response capabilities.

[0065] Before data is transmitted to the cloud, edge computing devices preprocess the data to ensure data quality and accuracy. The preprocessing steps include data cleaning, outlier detection, and data format conversion. Data cleaning is done using Python's pandas library to remove missing and duplicate values ​​and handle inconsistent data formats; outlier detection uses the z-score method to identify and handle extreme data points through statistical means; data format conversion ensures that different types of sensor data are unified into a standard format for subsequent processing.

[0066] The preprocessed data is sent to the big data platform in the cloud through a secure transmission channel for storage and further analysis.

[0067] In the data processing stage, big data analysis technology and machine learning algorithms can be used to conduct in-depth analysis of the collected data. By analyzing sensor data, abnormal patterns can be identified and potential fire risks can be warned in advance.

[0068] For example, time series analysis (such as the ARIMA model) can be used to predict the changing trends of temperature and smoke concentration, and historical data and current data can be combined to monitor and predict the possibility of fire in real time. Video data is processed through computer vision technology, such as flame detection, smoke detection, and personnel behavior analysis, and deep learning models (such as convolutional neural networks (CNNs)) are used to identify flames and smoke, achieving early fire warning.

[0069] In order to construct the fire knowledge graph, the acquisition unit 10 needs to obtain fire regulations data, fire equipment management data, emergency plan data, etc. These data will be used to construct a comprehensive and dynamically updated fire knowledge graph to support task allocation and the generation of correction suggestions.

[0070] The main functions of the acquisition unit 10 include acquiring and processing fire regulations data, acquiring and processing fire equipment management data, and acquiring and processing emergency plan data.

[0071] First, the sources of fire regulations data are fire regulations issued by the government, industry standards, local fire regulations, etc. The collection method can be through the API interface or regularly downloading official regulations documents (such as PDF, Word, HTML format), and then using text parsing technology (such as natural language processing NLP) to convert unstructured regulations documents into structured data.

[0072] During the data processing stage, key clauses, implementation rules, penalty provisions and other information in the regulations are extracted, and different types of regulatory content (such as fire equipment requirements, emergency plan requirements, etc.) are labeled and classified.

[0073] Secondly, the sources of fire equipment management data include fire equipment manufacturers, installation and maintenance service providers, on-site equipment monitoring systems, etc. The collection method can obtain real-time data (such as equipment operating status, maintenance records, fault alarms, etc.) from the equipment management system through the API interface, or use Internet of Things (IoT) devices (such as sensors) to directly collect real-time operation data of the equipment.

[0074] During the data processing stage, the equipment’s maintenance history, failure frequency, performance indicators and other data are summarized and analyzed, and the equipment data is labeled to form information such as equipment category, model, maintenance cycle, etc.

[0075] Finally, the sources for obtaining emergency plan data are internal enterprise emergency plan documents, fire drill records, emergency response training records, etc.

[0076] The collection method is to obtain the plan text through the company's internal information system or document management system, and use text parsing technology to convert the plan content into structured data.

[0077] During the data processing stage, key steps, responsible persons, emergency materials and other information in the plan are extracted, and different types of emergency plans (such as fire emergency plans, equipment failure emergency plans, etc.) are labeled and classified.

[0078] For the above building block 20:

[0079] As an optional implementation, the second target information includes: fire regulations data, fire equipment management data, and emergency plan data; the construction unit 20, when constructing the fire knowledge graph based on the second target information, includes:

[0080] Set the node set to V, the relationship set to E, and represent the fire protection knowledge graph as a directed graph G=(V,E);

[0081] Among them, the node set , each Represents a knowledge node; a set of relationships , each Representation Node and nodes The relationship between.

[0082] It should be noted that {} here represents a set, which includes multiple nodes, and () indicates that each element in the set is a triple.

[0083] In the specific implementation, when constructing the fire protection knowledge graph, it is first necessary to define the structure of the knowledge graph, including node types and relationship types. Node types include regulatory clauses, equipment, plans, training courses, etc., and relationship types include needs, recommendations, compliance, and related. Next, the collected data is converted into nodes and relationships of the knowledge graph. The regulatory node is a node for each regulatory clause or standard, which contains attributes such as clause content and implementation details. The equipment node is a node for each device and its attributes (such as model, maintenance cycle, and fault history). The plan node is a node for each emergency plan and its steps, responsible persons, materials, etc. Relationships include the "need" relationship between regulations and equipment, the "need" relationship between regulations and plans, the "related" relationship between equipment and plans, and the "comply" relationship between training courses and regulations.

[0084] Finally, use a graph database (such as Neo4j) to store the fire protection knowledge graph. Define the schema of the graph database, including node types and relationship types. The fire protection knowledge graph can be represented as a directed graph G=(V,E), where V is a node set and E is a relationship set. Each node in the node set corresponds to a node type. Node type T includes regulation nodes, equipment nodes, plan nodes, training nodes, etc. The relationship between two nodes in the relationship set corresponds to at least one relationship type R. Relationship types R include need, recommendation, compliance, and relevance.

[0085] For example, in a fire protection knowledge graph, different types of nodes represent different fire protection management elements. Each node type needs to contain specific attributes in order to store and retrieve relevant information.

[0086] : Regulation node, representing fire regulations and standards. Attributes: Regulation name (name), effective date (effective_date), content (content), implementation details (implementation_details), etc.

[0087] : Equipment node, representing various firefighting equipment. Attributes: equipment name (name), model (model), maintenance cycle (maintenance_cycle), fault history (fault_history), current status (current_status), etc.

[0088] : Plan node, representing the emergency plan. Attributes: plan name (name), steps (steps), responsible person (responsible_person), emergency materials (emergency_materials), drill records (drill_records), etc.

[0089] : Training node, representing a fire safety training course. Attributes: course name (name), training content (content), training frequency (frequency), training evaluation (evaluation), etc.

[0090] In addition, different types of relationships represent associations and dependencies between nodes. Relationship types need to contain specific attributes to describe the nature and details of the relationship.

[0091] Relation 1: requires, indicating that one node requires another node. Attributes: reason, urgency level, etc.

[0092] Relation 2: recommendations, which means one node recommends another node. Attributes: recommendation_reason, recommendation_level, etc.

[0093] Relation 3: complies_with, which means one node complies with another node. Attributes: compliance_standard, compliance_requirements, etc.

[0094] Relation 4: related_to, indicating the relationship between nodes. Attributes: relation_type, relation_description, etc.

[0095] Use the node creation function of a graph database (such as Neo4j) to create regulatory nodes, equipment nodes, plan nodes, and training nodes based on the defined node types and attributes.

[0096] Use the relationship creation function of the graph database to establish the "need" relationship between regulations and equipment, the "need" relationship between regulations and plans, the "related" relationship between equipment and plans, and the "compliance" relationship between training courses and regulations based on the defined relationship types and attributes. Import the processed structured data into the graph database to generate the corresponding nodes and relationships.

[0097] Through the above implementation methods, the fire protection knowledge graph constructed by the construction unit 20 can effectively integrate and manage data from different sources, support the generation of task allocation and correction suggestions, and improve the intelligence and efficiency of fire protection management.

[0098] Regarding the task assignment unit 30: As an optional implementation, the task assignment unit 30, when generating the first target task based on the first target information, performs the following steps: in response to the presence of at least one preset abnormal target information in the first target information, generating the first target task corresponding to the abnormal target information. The task assignment unit 30 can generate the first target task based on the first target information of the target place, especially when there is abnormal target information. The purpose of this function is to automatically generate corresponding tasks when an abnormal situation is detected in the target place, so as to facilitate the allocation and execution of subsequent tasks.

[0099] In a specific implementation, when the acquisition unit 10 obtains the first target information from sensors, equipment and other monitoring systems at the target location, the information includes various real-time data, such as temperature, smoke concentration, equipment operation status and video monitoring data. The system first monitors and analyzes these data in real time, and uses a preset anomaly detection algorithm to identify potential abnormal target information.

[0100] Anomaly detection algorithms can be based on a variety of techniques, including but not limited to threshold comparison, statistical analysis, and machine learning models. For example, thresholds for certain key parameters are set, such as when the temperature exceeds a certain value or when the smoke concentration reaches a dangerous level. When the monitoring data exceeds these thresholds, the system will identify it as an abnormal situation. In addition, using machine learning models (such as anomaly detection algorithms, classification algorithms, etc.), complex abnormal situations such as equipment failures and fire precursors can be detected more accurately.

[0101] Once abnormal target information is detected, the system will generate the corresponding first target task according to the preset rules and fire knowledge graph. The fire knowledge graph stores the mapping relationship between various abnormal situations and corresponding processing tasks. For example, if a fire precursor is detected, the system will generate tasks such as "check fire hazards", "start fire extinguishing equipment" and "evacuate personnel"; if equipment failure is detected, the system will generate tasks such as "equipment fault diagnosis" and "emergency maintenance".

[0102] When generating the first target task, different abnormality types correspond to different tasks. For example, fire abnormality and equipment failure abnormality will generate different task types.

[0103] For example, if the temperature sensor 12 detects an abnormally high temperature in a certain area, the system will generate a task of "checking the fire hazard in the area". At the same time, the system will also generate a task of "activating the automatic fire extinguishing equipment in the area" to ensure timely response to potential fire risks.

[0104] In this way, the process of generating the first target task based on abnormal target information can realize real-time, automated and efficient task generation and allocation, ensuring rapid response and taking appropriate measures when abnormal situations occur, thereby minimizing risks and losses.

[0105] As an optional implementation, the task allocation unit 30, when dynamically allocating the firefighting tasks to be performed by the target personnel in the target place based on the first target task and the preset second target task, performs the following steps: determining the priority of each of the firefighting tasks based on the urgency and importance of the firefighting tasks; obtaining the personnel information of each target personnel in the target place, and the skill information corresponding to each of the firefighting tasks; wherein the personnel information includes: position information, personal skill information, location information, and workload; determining the skill matching degree of each of the target personnel for the firefighting tasks based on the personal skill information and the skill information corresponding to the firefighting tasks; constructing a multi-objective optimization model based on the priority of the firefighting tasks, the skill matching degree, the workload, the location information, and the completion time, and generating a first task allocation plan by solving the optimization strategy.

[0106] See also Figure 3 , Figure 3 As an optional implementation, the task allocation unit 30, when constructing a multi-objective optimization model based on the priority of the firefighting task, the skill matching degree, the workload, the location information and the completion time, and generating a first task allocation scheme by solving the optimization strategy, performs the following steps: setting the first objective function , the first objective function is the sum of the task urgency and importance; the second objective function is set , the second objective function is the sum of the matching degree between tasks and personnel skills; the third objective function is set , the third objective function is the difference in personnel workload; the fourth objective function is set , the fourth objective function is the sum of the distances between the task location and the personnel location; the fifth objective function is set , the fifth objective function is the expected deviation of the task completion time; based on the objective function, the multi-objective optimization model is constructed; wherein the function expression of the multi-objective optimization model is:

[0107] ;

[0108] in, , , , as well as are respectively the weights of the first objective function, the second objective function, the third objective function, the fourth objective function and the fifth objective function in the multi-objective optimization model at time t.

[0109] In a specific implementation, the first objective function is set The function expression is:

[0110] ;

[0111] in, The weight indicating the urgency, The weight represents the importance, N is the number of firefighting tasks, and the urgency of each firefighting task is , the importance is , is a binary variable indicating whether task i is assigned.

[0112] In the specific implementation, it is necessary to clarify the specific definitions of the two parameters of urgency and importance. : Indicates the urgency of the task, which is determined by real-time data such as sensor data and alarm information. For example, fire alarms and smoke detector triggers will be assigned a higher degree of urgency. : Indicates the importance of the task to overall fire safety, which is determined by the nature of the task and its potential impact. For example, tasks such as regularly checking fire extinguishers and maintaining fire alarm systems have a high level of importance.

[0113] It is understandable that the above urgency and importance The specific value can be set in combination with the specific target location. For example, in some densely populated areas, the importance of certain regular fire safety inspection tasks can be set to a higher value. However, for some target locations with a relatively scattered distribution of personnel and a large total area, the importance of such regular fire inspection tasks can be appropriately reduced to reduce the burden on fire safety personnel.

[0114] In this application, in addition to setting the urgency and importance of each firefighting task, weights representing the urgency are introduced for these two parameters respectively. and weights indicating importance .

[0115] The purpose of this setting is that as time goes by, the urgency and importance of firefighting tasks may change in the proportion of the total firefighting task priority. For example, if the target location is a shopping mall, the smoking area of ​​the shopping mall often has fire hazards, but the firefighting tasks in the area may be initially set to a lower importance. In this case, the urgency and importance of the firefighting tasks may change in the proportion of the total firefighting task priority. Through dynamic adjustment, a more reasonable firefighting task priority can be determined based on actual scene information.

[0116] As a possible implementation, the dynamic adjustment of weights may be based on the following formula:

[0117] ;

[0118] in, and are the initial weights of the urgency and importance of the firefighting task, and are the weights of the urgency and importance of the firefighting task at time t, and are the adjustment coefficients, is at least one of a preset time variation amount and an event trigger amount;

[0119] For example, the adjustment factor and It can be dynamically calculated based on the analysis results of the feedback data. In addition, it can also be set as a fixed value based on the scene characteristics of the target location. For example, in some important scenes, a higher This makes the scenario more sensitive to the dynamic adjustment of the urgency weight of the task. In addition, machine learning can also be used to optimize the weight parameters.

[0120] Set the second objective function The function expression is:

[0121] ;

[0122] Where M is the number of target personnel, represents the skill information set required for firefighting task i, represents the personal skill set of target person j, is a binary variable indicating whether task i is assigned to target person j.

[0123] In a specific implementation, the personal skill set required by the second objective function can be obtained through personnel information, and the fire safety personnel in the target place can fill in their own skill information through statistical tables or software programs. For example, based on the skill information required to complete each fire task, each skill can be listed in a table or software, and a proficiency evaluation of 1 to 10 points can be set; the target personnel can fill in the above table content by themselves, and in addition, the fire knowledge graph can be constructed in combination with the construction unit 20 to set questions related to each skill for the target personnel to answer, and the skill proficiency score can be automatically calculated based on the answer results.

[0124] In the specific implementation, each firefighting task i has a skill information set , which contains all the skills required for the task.

[0125] Each target person j has a personal skill set , which contains all the skills that the person has.

[0126] Calculate the skill information set of firefighting task i and the personal skill set of target person j The intersection of .

[0127] The size of the intersection It represents the number of skills possessed by target personnel j that match the skills required for firefighting task i.

[0128] Calculate the skill matching degree, the formula is .

[0129] This value indicates the proportion of the skills required for the task that are possessed by the target personnel, ranging from 0 to 1.

[0130] The skill matching degrees of all tasks and target personnel are summed up to obtain the total matching degrees of tasks and target personnel skills.

[0131] Example, firefighting mission A: required skill set ; Firefighting mission B: Required skill set ; Target person 1: Skill set ; Target Personnel 2: Skill Set ;

[0132] Calculate the skill matching degree between each firefighting task and the target personnel: The skill matching degree between firefighting task A and target personnel 1 is: ; The skill matching degree between firefighting mission A and target personnel 2: ; The skill matching degree between firefighting mission B and target personnel 1: ; The skill matching degree between firefighting mission B and target personnel 2: .

[0133] Substitute these matching values ​​into the objective function The sum is taken to obtain the total matching degree between the firefighting task and the target personnel’s skills.

[0134] It is worth noting that if certain firefighting tasks require the target personnel to have certain skills when executing them, then relevant constraints can be added when solving the multi-objective function. The solution optimization of the multi-objective function will be explained below.

[0135] Setting the third objective function The function expression is:

[0136] ;

[0137] in, represents the initial workload of target person j, represents the workload of firefighting task i, is a binary variable indicating whether firefighting task i is assigned to target person j. , otherwise it is 0;

[0138] In a specific implementation, the expression represents the total workload of person j, including the initial workload and the workload of new tasks assigned .

[0139] For example, if the initial workload of target person 1 is , and task 1 and task 2 are assigned (4 hours and 3 hours respectively), the total workload is 9 hours. It can be understood that the simplified representation of the workload here is hours, and in actual execution, the workload can also be measured by work intensity, fatigue value, difficulty, etc., and other function expressions can also be combined in the third objective function in this application.

[0140] expression It means calculating the sum of squares of workload of all personnel.

[0141] For example, if there are two people with a total workload of 9 hours and 7 hours respectively, the sum of the squares of the workload is .

[0142] expression Calculate the average workload of all personnel.

[0143] For example, if two people have a total workload of 9 hours and 7 hours respectively, the average workload is Hour.

[0144] expression Represents the square of the average workload.

[0145] For example, the average workload is 8 hours, and its square is .

[0146] After subtracting the square of the average workload, we get the variance of the workload, which is The smaller the variance, the smaller the workload difference between personnel and the more reasonable the task allocation.

[0147] Set the fourth objective function The function expression is:

[0148] ;

[0149] The location of firefighting task i is , the location of target person j is , is a binary variable indicating whether firefighting task i is assigned to target person j. , otherwise it is 0.

[0150] The fourth objective function Ensure that task allocation minimizes the distance personnel must move while meeting task requirements, thereby improving response speed and efficiency.

[0151] For example, Task A: location coordinates are (2,3), Task B: location coordinates are (4,4), Personnel 1: current location coordinates are (5,1), Personnel 2: current location coordinates are (1,5).

[0152] Calculate the distance between each task and the person, and substitute these distances into the fourth objective function The sum is taken in to obtain the total distance between the task and the personnel position.

[0153] The fourth objective function The measurement method of the total distance between the task location and the personnel location is defined to ensure that task allocation minimizes the movement distance of personnel while meeting task requirements, thereby improving response speed and efficiency.

[0154] Set the fifth objective function The function expression is:

[0155] ;

[0156] in, is the estimated completion time of firefighting task i, is the actual completion time after firefighting task i is assigned to target person j.

[0157] In a specific implementation, the expression represents the expected completion time of task i The actual completion time after being assigned to person j The smaller the difference, the more on-time the task is completed.

[0158] expression Calculate the total deviation between all tasks and people. This objective function optimizes task allocation by minimizing the completion time deviation of all tasks.

[0159] As an optional implementation, the task allocation unit 30, based on the priority of the firefighting task, the skill matching degree, the workload, the location information and the completion time, constructs a multi-objective optimization model, and generates a first task allocation scheme by solving the optimization strategy, performs the following steps:

[0160] Dynamically adjust weights based on the following formula :

[0161] ;

[0162] in, is the initial weight, is the adjustment factor, is at least one of a preset time variation and a preset event triggering amount, .

[0163] In specific implementation, different weights can be set for each objective function in the multi-objective optimization model to more reasonably allocate firefighting tasks. The weight setting can be a fixed value based on the target site information. For example, the fire management department of the target site is more concerned about the quality of task completion, so the second objective function related to skill matching is more concerned about the quality of task completion. The weight can be set higher. In addition, the weight setting can also be set in the initial weight Dynamically updated based on.

[0164] In a specific implementation, the adjustment coefficient Determines how quickly and how much the weight changes over time. Adjustment factors can be set based on the urgency, importance, and feedback data of the task. For example, in an emergency, the urgency-related weight adjustment factor can be set larger to quickly increase the weight.

[0165] Time Variation Represents the time interval at which the system weights are adjusted. This interval can be a fixed time period (such as every hour) or based on a trigger of a specific event (such as the detection of a fire alarm).

[0166] In the specific implementation, the multi-objective optimization model we constructed can be solved by weighted sum method, Pareto optimization, genetic algorithm, and particle swarm optimization.

[0167] For example, taking genetic algorithm as an example, the following steps can be performed to solve the multi-objective optimization model: Step 1: Randomly generate a certain number of initial solutions, each solution represents a task allocation scheme. Each individual in the population is a task-personnel allocation matrix . Step 2: Calculate the fitness value of each individual. The fitness value is determined by the multi-objective optimization model Z, and the goal is to minimize Z. Step 3: Select some individuals for reproduction according to the fitness value. Individuals with high fitness values ​​have a higher probability of being selected. Step 4: Select two individuals for crossover operation to generate new individuals. The crossover operation can exchange some columns or rows in the task allocation matrix. Step 5: Perform mutation operations on some individuals to randomly change certain values ​​in the task allocation matrix to increase the diversity of the population. Step 6: The new individuals generated after selection, crossover and mutation form a new population, and calculate its fitness value. Step 7: Repeat the above steps until the termination condition is met (such as reaching the maximum number of iterations or the fitness value converges). Step 8: Select the individual with the highest fitness value from the final population as the optimal solution, and the corresponding task allocation plan is the optimal task allocation plan.

[0168] Through the above steps, the optimal task allocation solution is obtained is a matrix where each element is a binary value indicating whether task i is assigned to person j.

[0169] According to the matrix Assign tasks to appropriate personnel. elements, assigning task i to person j.

[0170] Use the communication and notification module of the system to send the task assignment results to relevant personnel. The notification content includes task description, task urgency, expected completion time, specific execution steps and precautions, etc. The above content can be obtained in combination with the fire protection knowledge graph constructed by the construction unit 20. Record the results and status of task assignment in the task management system for subsequent tracking and management. Monitor the execution of tasks in real time, collect execution data, and dynamically adjust weights and optimize models according to the completion of tasks.

[0171] For example, Task A: urgency 9, importance 8, expected completion time 2 hours, location coordinates (2,3), required skill set {firefighting, first aid}; Task B: urgency 5, importance 7, expected completion time 3 hours, location coordinates (4,4), required skill set {inspection, report writing}; Personnel 1: current location (5,1), possesses skill set {firefighting, first aid}; Personnel 2: current location (1,5), possesses skill set {inspection, report writing, equipment maintenance}. By solving the problem with the genetic algorithm, the following task allocation scheme is obtained: Task A is assigned to Personnel 1, and Task B is assigned to Personnel 2.

[0172] Finally, the allocation results are notified to the relevant personnel and recorded in the task management system. During the task execution, real-time monitoring and data collection are carried out to further optimize the model and adjust the weights.

[0173] In addition, if a firefighting mission has special requirements, such as the personnel performing the mission must have certain skills, or the personnel performing the mission must be in certain locations, the above multi-objective optimization model can be improved by adding constraints.

[0174] Exemplary, Task Specific skills required , and only people with this skill The following constraints can be introduced:

[0175] ;

[0176] in, is the indicator function, when personnel Possess skills 1 if the task is Must be assigned to at least one person with the skill of personnel.

[0177] For example, if the task Must be carried out by the personnel closest to the mission location Execution, corresponding constraints can also be introduced.

[0178] The above constraints are added to the multi-objective optimization model, and the constrained genetic algorithm is used to solve the optimization problem:

[0179] For example, randomly generate initial solutions that meet the constraints, ensure that the allocation of each task meets specific skill requirements and location constraints, set the crossover operation to ensure that the new individuals still meet all constraints, and set the mutated individuals to meet the constraints.

[0180] In a specific implementation, when an emergency occurs in the target scene, the task allocation unit 30 can generate multiple tasks for the emergency at the same time according to the preset emergency plan or fire knowledge graph. The specific task allocation method can be referred to above and will not be repeated here. In addition, the task allocation method is not limited to the above method. For example, when the task allocation unit 30 is not required to automatically allocate tasks in an emergency, all tasks can be directly allocated to each target person based on the preset emergency plan through manual setting, so as to flexibly respond to emergencies.

[0181] In this way, the AI-based fire management system can efficiently solve multi-objective optimization problems, achieve reasonable task allocation, improve the timeliness and efficiency of task execution, and ensure the efficient operation of fire safety management.

[0182] Regarding the task evaluation unit 40: As an optional implementation, the task evaluation unit 40 is also used to: obtain the firefighting task completion information of the target personnel; wherein the firefighting task completion information includes: task completion rate, task completion time, task delay rate, task success rate, personnel workload uniformity and task response time; based on the firefighting task completion information of the target personnel, determine the firefighting task execution status of the target personnel, and generate percentage scoring information; wherein each type of information in the firefighting task completion information corresponds to a percentage scoring method; based on the percentage scoring information, generate sub-evaluation information; wherein each sub-evaluation information corresponds to a target person; based on the sub-evaluation information corresponding to each target person in the target place, use the weighted average method to determine the evaluation information of the target place.

[0183] In the specific implementation, the task evaluation unit 40 first needs to obtain the fire task completion information of the target personnel. The fire task completion information includes the following aspects:

[0184] Task completion rate: the ratio of the number of tasks completed by the target person in a certain period of time to the total number of tasks assigned to him; Task completion time: the actual time required to complete a task; Task delay rate: the ratio of delayed task completion; Task success rate: the ratio of successfully completed tasks; Personnel workload uniformity: a measure of whether the workload of the target person is balanced among various tasks; Task response time: the time required from task assignment to the time when the target person starts to perform the task.

[0185] Based on the firefighting task completion information of the target personnel, the task evaluation unit 40 generates percentage scoring information. Each type of information corresponds to a percentage scoring method.

[0186] In the specific implementation, for the task completion rate, at the end of each evaluation cycle, the number of tasks completed by a target person (the number of completed tasks) is calculated by dividing the total number of tasks assigned to him (the total number of tasks); for the task completion time, the task completion time is obtained by subtracting the task start time from the task end time; for the task delay rate, the delay time of each task is calculated (if any), and then the ratio of the number of delayed tasks of a target person to the total number of tasks is counted; for the task success rate, the number of tasks successfully completed by a target person (the number of successful tasks) can be calculated by dividing the total number of tasks assigned to him (the total number of tasks); for the personnel workload, the workload of each task completed by the target person is calculated, and then the variance or standard deviation of these workloads is calculated to evaluate the uniformity. The smaller the standard deviation, the more uniform the workload.

[0187] Based on the percentage scoring information, the task evaluation unit 40 generates sub-evaluation information. Each sub-evaluation information corresponds to a target person. The sub-evaluation information includes the score of the target person on each indicator and the comprehensive score. Based on the sub-evaluation information corresponding to each target person of the target place, the evaluation information of the target place is determined using the weighted average method.

[0188] In specific implementation, the system can record and monitor information such as task completion rate, task completion time, task delay rate, task success rate, personnel workload uniformity, and task response time. For example, when the target personnel receive a firefighting task through the mobile phone software and confirm the completion, the system can record the task completion time.

[0189] The task success rate can be determined by computer vision and data analysis. For example, the acquisition unit 10 acquires the first target information in the target place through the sensor, which includes abnormal temperature information. After receiving the information, the task assignment unit 30 generates the corresponding first target task and assigns it to a target person for execution. After confirming that the task is completed, the target person can determine whether the temperature has returned to the preset normal value through the sensor data again. If the temperature has successfully returned to the preset normal value range, the task is confirmed to be successfully completed.

[0190] In addition, confirming whether a task is successfully completed by automatic detection may not be applicable to all types of tasks. For confirmation of whether such tasks are successfully completed, the task assignment unit 30 can generate a corresponding first target task, or preset a second target task, and assign it to a target person with a higher skill match. After the target person confirms the completion of the task, the system automatically obtains the task completion information to determine whether the task is successfully completed.

[0191] As an optional implementation, the task evaluation unit 40 is further configured to:

[0192] In response to the update of the firefighting task completion information of the target personnel, the sub-evaluation information is updated in real time based on the updated firefighting task completion information;

[0193] In response to the sub-evaluation information being updated, the evaluation information of the target place is updated by using a weighted average method based on the updated sub-evaluation information corresponding to each of the target persons of the target place.

[0194] In specific implementation, in addition to periodic updates, the sub-evaluation information of each target person can also be updated in real time according to actual conditions, and the evaluation information of the target place can also be updated in real time based on the sub-evaluation information of each target person updated in real time.

[0195] In this way, the execution status of the firefighting mission in this target scene can be reflected more quickly and intuitively in the entire fire management system, so as to help the responsible personnel of the target site to make more accurate subsequent decisions.

[0196] As an optional implementation, based on the sub-evaluation information corresponding to each of the target personnel of the target place, using the weighted average method to determine the evaluation information of the target place includes: based on the position information, the urgency and importance of the fire fighting task, when using the weighted average method, adjusting the weight of the sub-evaluation information corresponding to each of the target personnel. The weighted average calculation formula is:

[0197] ;

[0198] in, is the evaluation information of the target place, is the sub-evaluation information of the j-th target person, is the corresponding weight, and M is the number of target personnel.

[0199] In a specific implementation, the proportion of sub-evaluation information corresponding to different target personnel in the evaluation information for the target place is different. For example, when the task assignment unit 30 performs task assignment, the skill matching degree is included in it as described above. For some target personnel with fewer skills or whose skills cannot meet the firefighting tasks with high urgency or high importance, the firefighting tasks assigned to them by the task assignment unit 30 may be cleaning of some places, filling in report sheets, etc.

[0200] Then the sub-evaluation information of such target personnel will have a smaller weight in the evaluation information. However, the sub-evaluation information corresponding to the target personnel who perform fire equipment inspection and maintenance tasks will have a higher weight in the evaluation information.

[0201] In addition, weights can also be preset based on the target personnel's position information. Higher weights can be set for the executive directors and supervisors of the target sites because the firefighting tasks assigned to them are some important information confirmation, management or emergency decision-making; the authority of firefighting equipment maintenance personnel can be set to medium because the firefighting equipment inspection and maintenance tasks they perform are also critical for the handling and prevention of emergencies; lower weights can be set for target personnel who mainly perform information recording, firefighting knowledge popularization, and environmental cleaning.

[0202] As an optional implementation, a target area may be set above the target location. The target area may include multiple target locations. By collecting evaluation information of each target location in the target area and summarizing and counting them, fire management information of the entire target area may be obtained.

[0203] For example, the target area includes companies, scientific research institutes, and government agencies. The company includes evaluation information of three target locations, the scientific research institute includes evaluation information of one target location, and the government agency includes evaluation information of five target locations. By statistically analyzing the evaluation information corresponding to the target scenes included in each unit organization in the above area, the fire management information corresponding to the target area can be obtained.

[0204] For the interaction unit 50: In a specific implementation, the interaction unit 50, based on the evaluation information and the sub-evaluation information, generates suggestion information using the fire protection knowledge graph, and sends it to the target personnel, and performs the following steps: based on at least one of the fire protection task completion information, map the fire protection task completion information to the fire protection knowledge graph; wherein the fire protection task completion information includes: task response speed deficiency information, task failure information, and task completion efficiency deficiency information; use the nodes and relationships in the fire protection knowledge graph to search for a correction path; based on the correction path search results, generate the suggestion information, and send it to the target personnel.

[0205] In a specific implementation, the evaluation information and sub-evaluation information output by the task evaluation unit 40 include: information on insufficient task response speed, information on uncompleted tasks, and information on insufficient task completion efficiency;

[0206] The task response speed insufficient information is obtained by obtaining the task allocation time and the actual task start execution time to calculate the real response time, and then comparing the real response time with the preset expected response time.

[0207] Mission failure information can be obtained by searching for all failed missions in the mission information.

[0208] The information on insufficient task completion efficiency can be obtained by calculating the actual time used to complete a task through the task start execution time, task response time, and task completion time, and then compared with the preset standard completion time for the task.

[0209] It should be noted that the task assignment time is different from the task start time because the target personnel need to go to the task location first.

[0210] In a specific implementation, the interactive unit 50 directly obtains task completion deficiency information from the evaluation information, including specific tasks and deficiencies. For example, the system automatically reads information from the task completion log in the evaluation information, which records in detail the task ID and specific deficiencies of each task. Directly read the task ID and deficiency information in the log, such as the problem of insufficient response speed for the task with task ID 1. Map the specific task deficiency information extracted from the log to the relevant nodes in the fire protection knowledge graph.

[0211] In a specific implementation, the construction unit 20 uses the Neo4j graph database to store the fire protection knowledge graph. The knowledge graph contains various nodes and relationships, such as training courses, equipment maintenance and other nodes.

[0212] Use Cypher query language to find relevant nodes. Based on specific deficiency information (such as insufficient response speed, unfinished tasks, etc.), the system searches for corresponding improvement nodes in the knowledge graph, such as "emergency response speed improvement training" and "fire alarm maintenance."

[0213] According to the mapping result, the starting node is determined to be a specific node with insufficient information.

[0214] Using Neo4j's graph query capabilities, find paths from the starting node to the improvement nodes. These paths may include multiple relationships from the insufficient information node to the relevant training or equipment maintenance nodes.

[0215] The information of relevant nodes is extracted from the correction path, such as the detailed information of "emergency response speed improvement training" and "fire alarm maintenance", and sent to the target personnel.

[0216] In specific implementation, detailed information on "emergency response speed improvement training" and "fire alarm maintenance" in the fire protection knowledge graph can be sent to target personnel through software push, text messages, emails, etc.

[0217] It is understandable that the above execution process can be uniformly called and implemented in Python language, which will not be repeated here.

[0218] In this way, the system realizes the mapping of insufficient firefighting task completion information to the firefighting knowledge graph, and generates and sends targeted advice information. This process not only improves the efficiency and effectiveness of task execution, but also helps the target personnel continuously improve their professional skills and emergency response capabilities, thereby improving the overall level of firefighting management.

[0219] It should be understood that in the various embodiments of the present application, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0220] It should be understood that determining B based on A does not mean determining B only based on A. B can also be determined based on A and / or other information.

[0221] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed in the present invention can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0222] In the description of this specification, the description with reference to the terms "one embodiment", "example", "specific example", etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0223] The preferred embodiments of the present invention disclosed above are only used to help explain the present application. The preferred embodiments do not describe all the details in detail, nor do they limit the present application to specific implementation methods. Obviously, many modifications and changes can be made according to the content of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the present application, so that those skilled in the art can understand and use the present application well. The present application is limited only by the claims and their full scope and equivalents.

Claims

1. A fire management system based on artificial intelligence, characterized in that: include: A collection unit, configured to obtain first target information of the target location based on a data collection device deployed at the target location; The acquisition unit is also used to obtain second target information for building a fire protection knowledge graph; A construction unit, configured to construct a fire protection knowledge graph based on the second target information; A task allocation unit, configured to generate a first target task based on the first target information; Based on the first target task and the preset second target task, dynamically assigning the firefighting tasks to be performed by the target personnel in the target location; The firefighting tasks include: a first target task and a second target task; A task evaluation unit, used to determine the firefighting task execution status of the target personnel based on a pre-built evaluation model, and generate sub-evaluation information; based on the sub-evaluation information, generate evaluation information of the target location; An interaction unit, configured to generate suggestion information based on the evaluation information and the sub-evaluation information using the fire protection knowledge graph, and send the suggestion information to the target person; Wherein, the task allocation unit is also used for: Determining the priority of each firefighting task based on the urgency and importance of the firefighting task; Acquire the personnel information of each target person in the target location, and the skill information corresponding to each firefighting task; wherein the personnel information includes: position information, personal skill information, location information, and workload; Determining the skill matching degree of each of the target personnel for the firefighting task based on the personal skill information and the skill information corresponding to the firefighting task; Based on the priority of the firefighting task, the skill matching degree, the workload, the location information, and the task completion time, a multi-objective optimization model is constructed, and a first task allocation plan is generated by solving the optimization strategy; The task allocation unit is also used for: Set the first objective function , the first objective function is the sum of the urgency and importance of the firefighting task; the second objective function is set , the second objective function is the sum of the skill matching of the target personnel for the firefighting task; the third objective function is set to , the third objective function is the target personnel workload difference; the fourth objective function is set , the fourth objective function is the sum of the distances between the task location and the target personnel location; the fifth objective function is set , the fifth objective function is the expected deviation of task completion time; Based on the objective function, the multi-objective optimization model is constructed; the function expression of the multi-objective optimization model is: ; in, , , , as well as are respectively the weights of the first objective function, the second objective function, the third objective function, the fourth objective function, and the fifth objective function in the multi-objective optimization model at time t; The task allocation unit is also used for: Dynamically adjust weights , the adjustment formula is as follows: ; in, is the initial weight, is the adjustment factor, is at least one of a preset time variation and a preset event triggering amount, .

2. The fire management system based on artificial intelligence according to claim 1 is characterized in that: The second target information includes: fire regulations data, fire equipment management data, and emergency plan data; and constructing the fire knowledge graph based on the second target information includes: Set the node set to V, the relationship set to E, and represent the fire protection knowledge graph as a directed graph G=(V,E); Among them, the node set , each Represents a knowledge node; a set of relationships , each Representation Node and nodes The relationship between.

3. The fire management system based on artificial intelligence according to claim 2 is characterized in that: Generating a first target task based on the first target information includes: There is at least one preset abnormal target information in the first target information, and based on the fire protection knowledge graph, the first target task corresponding to the abnormal target information is generated.

4. The fire management system based on artificial intelligence according to claim 3 is characterized in that: The task evaluation unit is also used for: Obtaining the firefighting task completion information of the target personnel; wherein the firefighting task completion information includes: task completion rate, task completion time, task delay rate, task success rate, personnel workload uniformity and task response time; Based on the firefighting task completion information of the target personnel, determine the firefighting task execution status of the target personnel, and generate percentage scoring information; wherein each type of information in the firefighting task completion information corresponds to a percentage scoring method; Based on the percentage scoring information, the sub-evaluation information is generated; wherein each sub-evaluation information corresponds to one target person; Based on the sub-evaluation information corresponding to each of the target persons in the target place, the evaluation information of the target place is determined by using a weighted average method.

5. The fire management system based on artificial intelligence according to claim 4 is characterized in that: The task evaluation unit is also used for: In response to the target personnel's firefighting task completion information being updated, the sub-evaluation information is updated in real time based on the updated firefighting task completion information; The evaluation information of the target place is updated by using a weighted average method based on the updated sub-evaluation information corresponding to each target person of the target place.

6. The fire management system based on artificial intelligence according to claim 5 is characterized in that: The step of determining the evaluation information of the target place by using a weighted average method based on the sub-evaluation information corresponding to each target person of the target place includes: Based on the position information, the urgency and importance of the firefighting task, the weight of the sub-evaluation information corresponding to each of the target personnel is adjusted when using the weighted average method.

7. The artificial intelligence-based fire management system according to claim 6, characterized in that: The evaluation information includes: information on insufficient completion of firefighting tasks; The interaction unit is used for: Based on at least one of the firefighting task completion insufficient information, mapping the firefighting task completion insufficient information to the firefighting knowledge graph; wherein the firefighting task completion insufficient information includes: task response speed insufficient information, task failure information, and task completion efficiency insufficient information; Using the nodes and relationships in the fire protection knowledge graph, searching for a correction path; Based on the corrected path search results, suggestion information is generated and sent to the target person.