Fire protection and security management method and system

Abstract

The invention discloses a fire-fighting security management method and system, and the method comprises the steps: collecting fire-fighting, security and environment data through a standardized interface, and carrying out the time-space calibration, and then forming a structural data set; denoising by a self-adaptive algorithm, and extracting characteristic parameters such as water pressure fluctuation rate; the mixed detection model is combined with a rule engine to detect abnormity, and weighted score grading alarm is carried out; a dynamic plan is generated and executed during emergency warning; the multi-agent reinforcement learning model is used for daily routing inspection route optimization and the like. The technical effects of pushing differential information, recording whole-process data to form a closed-loop log, continuously improving the inspection efficiency through autonomous optimization, reducing the false alarm rate, shortening the response time, and comprehensively improving the intelligent level of fire-fighting security management, the emergency disposal capability and the long-term adaptability are achieved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of fire safety risk early warning, in particular to a fire safety and security management method and system. BACKGROUND

[0002] With the acceleration of urbanization and the improvement of building complexity, the importance of fire safety and security management is increasingly prominent. Traditional fire safety and security systems mostly adopt independent operation mode, which has problems such as serious data island and insufficient linkage: fire safety equipment focuses on fire monitoring and extinguishing, and security system focuses on intrusion prevention, and the data of the two cannot form effective interconnection, which is difficult to deal with complex security risks. In the prior art, device data collection mostly relies on manual inspection or single sensor, which has defects such as poor real-time performance and limited coverage. For example, fire water pressure detection often uses periodic manual recording method, which cannot capture the pressure drop caused by pipeline leakage in time; environmental parameter monitoring is mostly limited to basic indicators such as temperature and smoke, and there is insufficient integration of related data such as gas concentration and humidity change, which is easy to cause misjudgment or omission of fire hazards. In the aspect of anomaly detection, the traditional system mostly triggers alarm based on fixed threshold, and lacks intelligent analysis capability. When the device operating parameters fluctuate but do not reach the threshold, the potential risks are difficult to be identified; at the same time, the alarm grading mechanism is simple, which causes the management personnel to be unable to quickly judge the priority of the danger, and delays the disposal opportunity. In addition, the emergency plan is mostly a fixed scheme set in advance, which does not consider the dynamic factors on site, has poor adaptability, and has low execution efficiency in complex scenes. SUMMARY

[0003] The purpose of the present application is to provide a fire safety and security management method and system, which solves the problem that in the existing fire safety and security management, device anomaly alarm, inspection task allocation and maintenance record management are independent of each other, which leads to lag in abnormal response, insufficient targeting of inspection tasks, disconnection between maintenance records and alarm traceability, and affects the efficiency and accuracy of safety management.

[0004] In order to achieve the above purpose, the technical scheme adopted by the present application is as follows: a fire safety and security management method, comprising the following steps: Step 1, multi-source device data collection, real-time collection of operating data of fire safety equipment, security equipment and environmental sensors through standardized interface, time and space calibration of the collected data, association of device ID, position coordinates, collection time and parameter value to form a structured data set; Step 2, data preprocessing and feature extraction, denoising of original data by using adaptive algorithm, extraction of device operating characteristic parameters, including water pressure fluctuation rate of fire safety equipment, abnormal behavior feature vector of security equipment and trend slope of environmental parameters; Step 3, intelligent anomaly detection and alarm grading, build a hybrid detection model to detect anomalies in real-time data, calculate anomaly scores and verify them through a rule engine, and grade alarms based on a hierarchical weighted scoring algorithm, dividing them into general, serious, and urgent alarms; Step 4, dynamic emergency plan generation and execution, when an emergency alarm is detected, a hybrid decision algorithm is started to generate a dynamic plan, which includes resource scheduling paths, personnel evacuation routes, and device linkage control logic, control instructions are sent to field devices through edge computing nodes, and disposal guidelines are pushed to responsible persons; Step 5, system self-optimization based on reinforcement learning, build a multi-agent reinforcement learning model, including device inspection agents, parameter threshold agents, and resource scheduling agents, train daily based on historical data to automatically optimize inspection routes, alarm thresholds, and resource allocation; Step 6, human-computer interaction and closed-loop management, push differentiated information to different role users, record alarm handling process data to form closed-loop management logs as training data input for reinforcement learning models.

[0005] Further, the operation data of the fire-fighting equipment includes fire hydrant water pressure, sprinkler system water level, and fire detector detection value; the operation data of the security equipment includes camera online status, video recording status, and access controller switch status; the operation data of the environmental sensor includes temperature and humidity, smoke concentration, and gas concentration.

[0006] Further, the adaptive algorithm is an adaptive sliding window filtering algorithm, with an initial window size of 30 data points, and when data mutation is detected, the window size is automatically reduced to 5 data points; the trend slope of the environmental parameter includes temperature change rate and humidity change rate.

[0007] Further, the hybrid detection model is a hybrid model based on improved isolated forest and rule engine, which uses the isolated forest algorithm to train historical normal data to establish a normal behavior baseline, calculates the anomaly score of real-time data, and triggers the rule engine for secondary verification when the anomaly score is ≥60; in the hierarchical weighted scoring algorithm, the first-level indicator is threat level, with a weight of 60%, the second-level indicator is diffusion speed, with a weight of 20%, and the third-level indicator is impact range, with a weight of 20%.

[0008] Further, the hybrid decision algorithm is a hybrid algorithm based on case-based reasoning and rule-based reasoning, the case-based reasoning module retrieves historical event solutions with a similarity ≥85% from the case base, and the rule-based reasoning module adjusts the case solution based on real-time data; the disposal guidelines include AR navigation information.

[0009] Further, the device inspection intelligent agent takes the minimum failure rate as the reward function, the parameter threshold intelligent agent takes the lowest false alarm rate as the reward function, and the resource scheduling intelligent agent takes the shortest response time as the reward function; and the historical data is operation data in the last 30 days.

[0010] Further, a system global statistical report is pushed to an administrator, a task work order with an electronic fence is pushed to an inspector, and real-time alarm video stream is pushed to an on-duty personnel; and the closed-loop management log includes response time, disposal measures and result feedback.

[0011] A system global statistical report is pushed to an administrator, a task work order with an electronic fence is pushed to an inspector, and real-time alarm video stream is pushed to an on-duty personnel; and the closed-loop management log includes response time, disposal measures and result feedback.

[0012] A fire safety management system for implementing the fire safety management method, comprising a data collection module, a data processing module, a detection module, an execution module and an interaction module; the data collection module is connected with the data processing module, and is used for collecting operation data of fire equipment, security equipment and environmental sensors; the data processing module is connected with the detection module, and is used for receiving and processing the operation data; the detection module is connected with the execution module, and is used for receiving and detecting the processed data; the execution module is connected with the interaction module, and is used for receiving the abnormal detection result and performing corresponding processing operation; and the interaction module is connected with the data processing module, and is used for displaying data and receiving operation instructions.

[0013] Compared with the prior art, the present application has the following advantages: Through a standardized interface, multi-source equipment data is collected and structured data groups are formed, adaptive algorithm preprocessing and feature extraction are performed, intelligent abnormal detection and alarm grading are realized by using a hybrid detection model combined with a hierarchical weighted scoring algorithm, a dynamic emergency plan is generated by using a hybrid decision algorithm and is executed by means of edge computing, and a multi-agent reinforcement learning model is used to realize system self-optimization, while human-computer interaction and closed-loop management are performed, thereby solving the technical problems of low management efficiency, untimely response and poor adaptability caused by the dispersion of multi-source data in traditional fire safety management, the lag and low accuracy of abnormal detection, unreasonable alarm grading, the fixation of emergency plans and the lack of self-optimization capability of the system, and achieving the technical effects of realizing integrated linkage processing of multi-source data, improving the accuracy and timeliness of abnormal detection, enhancing the accuracy of alarm response and scene adaptability, continuously improving the inspection efficiency, reducing the false alarm rate and shortening the response time through autonomous optimization, and comprehensively improving the intelligent level, emergency disposal capability and long-term adaptability of fire safety management. BRIEF DESCRIPTION OF DRAWINGS

[0014] The present application will be further described below with reference to the drawings: Fig. 1 This is a schematic diagram of the fire safety management method of the present invention; Fig. 2 This is a structural diagram of the fire protection and security management system of the present invention; Fig. 3 This is a flowchart of the emergency response of the present invention. Detailed Implementation

[0015] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0016] The technical solution of the present invention will be described in detail below with reference to specific embodiments. The following specific embodiments may be combined or substituted with each other according to the actual situation, and the same or similar concepts or processes may not be described again in some embodiments.

[0017] like Figs. 1 to 3 As shown, this invention provides a fire safety management method and system. The specific steps of the management method include the system synchronously accessing the operating data of fire-fighting equipment, security equipment, and environmental sensors through a standardized interface. The fire hydrant water pressure and sprinkler system water level of the fire-fighting equipment, the online status of the security equipment's cameras and the on / off status of the access control controllers, and the temperature, humidity, and smoke concentration of the environmental sensors are all stored in the system database as structured data groups with associated device IDs, location coordinates, and acquisition time, laying the foundation for subsequent processing.

[0018] These raw data will immediately enter the preprocessing stage. The system adopts an adaptive sliding window filtering algorithm, initially using 30 data points as a window for noise reduction. When a sudden change in data is detected, the window will automatically shrink to 5 data points to accurately filter noise. At the same time, the system extracts key feature parameters, such as the water pressure fluctuation rate of fire-fighting equipment, the abnormal behavior feature vectors such as personnel gathering identified by security equipment through intelligent analysis, and the temperature change rate of environmental parameters, making the data more valuable for analysis.

[0019] The preprocessed data stream will enter the intelligent anomaly detection stage. The system first uses an improved isolated forest algorithm to establish a baseline based on historical normal data and calculate anomaly scores for real-time data. When the score is ≥60, the rule engine will perform secondary verification by combining device location and historical fault records. Then, based on the hierarchical weighted scoring algorithm, the anomalies will be classified into general alarms, critical alarms, and emergency alarms according to the degree of threat, spread speed, and scope of impact.

[0020] If an emergency alarm is triggered, the system will immediately start the dynamic emergency plan generation and execution process. The hybrid decision algorithm first retrieves historical solutions with a similarity of ≥85% from the case library and then adjusts them based on real-time data to determine resource scheduling paths, plan personnel evacuation routes, and generate device linkage logic. The edge computing node sends control instructions to on-site devices and pushes disposal guidelines with AR navigation to inspectors to ensure efficient implementation of emergency responses.

[0021] The system also optimizes itself through multi-agent reinforcement learning. The device inspection agent optimizes the inspection route with the goal of minimizing fault rates, the parameter threshold agent adjusts the environmental sensor threshold with the goal of minimizing false alarm rates, and the resource scheduling agent allocates the nearest maintenance personnel with the goal of minimizing response times. These agents are trained daily based on the operation data of the past 30 days to automatically update the inspection plan, threshold parameters, and resource strategy.

[0022] Finally, the system achieves closed-loop management through human-computer interaction, pushing global statistical reports to administrators, sending task orders with electronic fences to inspectors, and pushing real-time alarm video streams to on-duty personnel. The system records the entire alarm handling process, including response time, disposal measures, and result feedback, to form a closed-loop log that feeds back to the reinforcement learning model, continuously improving system performance.

[0023] The fire safety and security management system is composed of a data acquisition module, a data processing module, a detection module, an execution module, and an interaction module connected in sequence. Each module not only functions independently but also forms an organic whole through data flow.

[0024] The data acquisition module is the starting point of the system and is directly connected to the data processing module. It is responsible for real-time acquisition of operation data from fire safety equipment, security equipment, and environmental sensors. These data include water pressure and water level of fire safety equipment, online status and video recording status of security equipment, and temperature and humidity, smoke concentration of environmental sensors. After acquisition, the data is transmitted to the data processing module.

[0025] The data processing module receives raw data from the data acquisition module and is connected to the detection module. Its main task is to preprocess the received operation data, including denoising using adaptive algorithms and extracting device operation characteristic parameters. The structured data after processing is transmitted to the detection module in real time.

[0026] The detection module receives the processed data from the data processing module and is connected to the execution module. Its function is to detect abnormalities in these data, identify abnormalities through a hybrid detection model, and perform alarm grading using a hierarchical weighted scoring algorithm. The abnormal detection results and alarm levels are then sent to the execution module.

[0027] The execution module is connected with the detection module and the interaction module, when receiving the abnormal detection result transmitted by the detection module, especially the emergency alarm, the dynamic emergency plan generation and execution process is started, including generating resource scheduling path, personnel evacuation route and equipment linkage control logic, sending control instructions to the field device, and synchronizing the execution situation to the interaction module.

[0028] The interaction module is connected with the execution module at one end and connected with the data processing module at the other end, which receives the execution situation data transmitted by the execution module and obtains various types of system data processed from the data processing module, and is mainly used for displaying differentiated information to different role users, such as pushing statistical report to the administrator and pushing task work order to the inspector, receiving the operation instruction of the user and feeding back to the data processing module, forming a closed loop of data flow.

[0029] The connection relationship ensures the smooth operation of the whole process from data collection, processing, detection, execution to interaction, so that the system can efficiently realize the intelligent management of fire safety and security.

[0030] In addition to the preferred embodiments described above, the present application has other embodiments, and all other embodiments obtained by those skilled in the art based on the embodiments in the present application without creative labor belong to the scope of the present application.

Claims

1. A fire safety and security management method, characterized by, The method comprises the following steps: Step 1, multi-source equipment data acquisition, real-time acquisition of operation data of fire-fighting equipment, security equipment and environmental sensors through a standardized interface, time and space calibration of the acquired data, correlation of equipment ID, location coordinates, acquisition time and parameter values to form a structured data set; Step 2, data preprocessing and feature extraction, denoising of the original data using an adaptive algorithm, extraction of equipment operation characteristic parameters, including water pressure fluctuation rate of fire-fighting equipment, abnormal behavior characteristic vector of security equipment and trend slope of environmental parameters; Step 3, intelligent anomaly detection and alarm grading, construction of a hybrid detection model based on real-time data anomaly detection, calculation of anomaly scores and secondary verification by a rule engine, alarm grading based on a hierarchical weighted scoring algorithm, divided into general alarm, serious alarm and emergency alarm; Step 4, dynamic emergency plan generation and execution, when an emergency alarm is detected, a hybrid decision algorithm is started to generate a dynamic plan, which includes resource scheduling path, personnel evacuation route and equipment linkage control logic, control instructions are sent to field equipment through an edge computing node, and disposal guidelines are pushed to the person in charge; Step 5, system self-optimization based on reinforcement learning, construction of a multi-agent reinforcement learning model, including equipment inspection agent, parameter threshold agent and resource scheduling agent, training based on historical data every day, automatic optimization of inspection route, alarm threshold and resource allocation; Step 6, human-computer interaction and closed-loop management, differentiated information is pushed to different role users, alarm handling process data is recorded to form a closed-loop management log, which is used as training data input for the reinforcement learning model.

2. The fire safety management method according to claim 1, characterized by, The operation data of the fire-fighting equipment includes fire hydrant water pressure, sprinkler system water level and fire detector detection value; the operation data of the security equipment includes camera online status, video recording status and access controller switch status; the operation data of the environmental sensors includes temperature and humidity, smoke concentration and gas concentration.

3. The fire safety management method according to claim 1, characterized by, The adaptive algorithm is an adaptive sliding window filtering algorithm, the initial window size is set to 30 data points, and when data mutation is detected, the window is automatically reduced to 5 data points; the trend slope of the environmental parameters includes temperature change rate and humidity change rate.

4. The fire safety management method according to claim 1, characterized by, The hybrid detection model is a hybrid model based on improved isolated forest and rule engine, the isolated forest algorithm is used to train the historical normal data to establish a normal behavior baseline, and the real-time data is calculated for anomaly score, and when the anomaly score is greater than or equal to 60, the rule engine is triggered for secondary verification; in the hierarchical weighted scoring algorithm, the first-level index is threat level, the weight is 60%, the second-level index is diffusion speed, the weight is 20%, and the third-level index is influence range, the weight is 20%.

5. The fire safety management method according to claim 1, wherein The hybrid decision algorithm is a hybrid algorithm based on case-based reasoning and rule-based reasoning, the case-based reasoning module retrieves a historical event solution with a similarity greater than or equal to 85% from a case base, and the rule-based reasoning module adjusts the case solution according to real-time data; the disposal guidelines include AR navigation information.

6. The fire safety management method according to claim 1, wherein The device inspection intelligent agent takes the minimum failure rate as a reward function, the parameter threshold intelligent agent takes the lowest false alarm rate as a reward function, and the resource scheduling intelligent agent takes the shortest response time as a reward function; and the historical data is operation data in the last 30 days.

7. The fire safety management method of claim 1, wherein, The system pushes a global statistical report to an administrator, pushes a task work order with an electronic fence to an inspector, and pushes a real-time alarm video stream to a duty officer; and the closed-loop management log includes a response time, a treatment measure, and a result feedback. 8.A fire safety and security management system for implementing the fire safety and security management method according to any one of claims 1-7, comprising a data collection module, a data processing module, a detection module, an execution module, and an interaction module; the data collection module is connected with the data processing module, and is configured to collect operation data of fire equipment, security equipment, and environmental sensors; the data processing module is connected with the detection module, and is configured to receive and process the operation data; the detection module is connected with the execution module, and is configured to receive and detect the processed data; the execution module is connected with the interaction module, and is configured to receive an abnormality detection result and perform a corresponding processing operation; and the interaction module is connected with the data processing module, and is configured to display data and receive an operation instruction.

Images