Fire alarm linkage method and system applied to intelligent lock

Through the linkage between intelligent locks and fire alarm systems, multi-source fusion, deep reinforcement learning and knowledge graph inference technology are used to build an adaptive decision-making model, solving the shortcomings of intelligent locks in fire safety, improving fire response capabilities and system stability .

CN119026072AActive Publication Date: 2024-11-26YIMAITONG (SHENZHEN) INTELLIGENT TECH CO LTD

Patent Information

Application Number
CN202411013379.X
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-26
Publication Date
2024-11-26
Estimated Expiration
2044-07-26

AI Technical Summary

Technical Problem

The existing smart lock and fire alarm systems lack effective linkage mechanisms, which leads to the inability to fully utilize the functional advantages of smart locks when a fire occurs. The false alarm rate is high and the response speed is slow, making it difficult to deal with complex fire situations, and the smart lock is in the face. Lack of flexible response strategies and adaptability during fires.

Method used

By pairing and exchanging communications of smart locks and fire alarm systems, building equipment association topology diagrams, performing multi-source fusion and spatiotemporal correlation analysis, introducing deep reinforcement learning and situational awareness processing, building adaptive decision-making models and differentiated response strategies, Using edge computing and knowledge graph inference technology, intelligent lock linkage control and system optimization are realized.

Benefits of technology

Improve the accuracy of fire situation assessment and the system's ability to respond to complex fire situations, enhance the accuracy and response flexibility of intelligent lock linkage control, ensure the system's rapid adaptation to environmental changes and continuous optimization, and improve the stability of fire safety. Sex and reliability.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119026072A_ABST
    Figure CN119026072A_ABST
Patent Text Reader

Abstract

The invention relates to the technical field of fire alarm, and discloses a fire alarm linkage method and system applied to an intelligent lock. The method comprises the following steps: performing communication pairing and data exchange on an intelligent lock and a fire alarm system to obtain local sensing data and fire monitoring data; performing multi-source fusion and space-time correlation analysis to obtain a comprehensive fire situation assessment result; constructing an adaptive decision model and a differential response strategy; performing task decomposition and cooperative execution to obtain an intelligent lock linkage control instruction set and obtain an execution effect evaluation index; performing real-time monitoring and edge calculation analysis on the execution effect evaluation index and the environmental parameter change data to obtain a target feedback data stream; and performing knowledge graph reasoning on the target feedback data stream to obtain decision model updating parameters, and performing model updating on the adaptive decision model based on the decision model updating parameters to obtain a target decision model, thereby realizing intelligent linkage of the intelligent lock and the fire alarm system.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of fire alarm technology, and in particular to a fire alarm linkage method and system applied to a smart lock. Background Art

[0002] With the rapid development of smart home technology, smart locks have been widely used as important home security devices. However, traditional smart lock systems still have many shortcomings in fire alarm and emergency response. Existing fire alarm systems are usually independent of smart locks and lack an effective linkage mechanism, which makes it impossible to give full play to the functional advantages of smart locks when a fire occurs.

[0003] In addition, current fire alarm systems generally have problems such as high false alarm rate and slow response speed, making it difficult to cope with complex and changing fire situations. Especially in large buildings or multi-story residential buildings, the lack of accurate fire situation assessment and intelligent decision-making support often leads to low evacuation efficiency and increased risk of casualties. At the same time, existing smart lock systems often lack flexible response strategies and adaptive capabilities when facing emergencies such as fires. This not only limits the role of smart locks in fire safety, but also fails to meet personalized safety needs in different scenarios. Summary of the invention

[0004] The present application provides a fire alarm linkage method and system applied to a smart lock, which is used to realize the intelligent linkage between the smart lock and the fire alarm system.

[0005] In a first aspect, the present application provides a fire alarm linkage method applied to a smart lock, and the fire alarm linkage method applied to a smart lock includes:

[0006] Perform communication pairing and data exchange between the smart lock and the fire alarm system to obtain a device association topology diagram, and acquire local sensor data of the smart lock and fire monitoring data of the fire alarm system;

[0007] Based on the equipment association topology diagram, multi-source fusion and spatiotemporal association analysis are performed on the local sensor data and the fire monitoring data to obtain a comprehensive fire situation assessment result;

[0008] Perform deep reinforcement learning and situational awareness processing on the comprehensive fire situation assessment results to build an adaptive decision-making model and differentiated response strategy;

[0009] Decomposing and co-executing the adaptive decision model and the differentiated response strategy, obtaining a smart lock linkage control instruction set, and obtaining an execution effect evaluation index of the smart lock linkage control instruction set;

[0010] Perform real-time monitoring and edge computing analysis on the execution effect evaluation index and environmental parameter change data to obtain a target feedback data stream;

[0011] Knowledge graph reasoning is performed on the target feedback data stream to obtain decision model update parameters, and the adaptive decision model is updated based on the decision model update parameters to obtain a target decision model.

[0012] In a second aspect, the present application provides a fire alarm linkage system applied to a smart lock, and the fire alarm linkage system applied to a smart lock includes:

[0013] An acquisition module is used to perform communication pairing and data exchange between the smart lock and the fire alarm system, obtain a device association topology diagram, and obtain local sensor data of the smart lock and fire monitoring data of the fire alarm system;

[0014] A fusion module, used for performing multi-source fusion and spatiotemporal correlation analysis on the local sensor data and the fire monitoring data based on the device association topology diagram to obtain a comprehensive fire situation assessment result;

[0015] A construction module is used to perform deep reinforcement learning and situational awareness processing on the comprehensive fire situation assessment results, and to construct an adaptive decision model and a differentiated response strategy;

[0016] An execution module, used to perform task decomposition and collaborative execution on the adaptive decision model and the differentiated response strategy, obtain a smart lock linkage control instruction set, and obtain an execution effect evaluation index of the smart lock linkage control instruction set;

[0017] An analysis module is used to perform real-time monitoring and edge computing analysis on the execution effect evaluation index and environmental parameter change data to obtain a target feedback data stream;

[0018] An updating module is used to perform knowledge graph reasoning on the target feedback data stream to obtain decision model update parameters, and to update the adaptive decision model based on the decision model update parameters to obtain a target decision model.

[0019] The third aspect of the present application provides a fire alarm linkage device applied to a smart lock, comprising: a memory and at least one processor, wherein the memory stores instructions; the at least one processor calls the instructions in the memory so that the fire alarm linkage device applied to the smart lock executes the above-mentioned fire alarm linkage method applied to the smart lock.

[0020] A fourth aspect of the present application provides a computer-readable storage medium, which stores instructions. When the computer-readable storage medium is run on a computer, it enables the computer to execute the above-mentioned fire alarm linkage method applied to smart locks.

[0021] In the technical solution provided by this application, by pairing and exchanging data between smart locks and fire alarm systems, and building a device association topology map, seamless integration between systems is achieved, and the efficiency and reliability of data transmission are improved. Multi-source fusion and spatiotemporal correlation analysis technology are used to comprehensively process local sensor data and fire monitoring data, significantly improving the accuracy and comprehensiveness of fire situation assessment. Deep reinforcement learning and situational awareness processing technology are introduced to build adaptive decision-making models and differentiated response strategies, enhancing the system's ability to cope with complex fire situations. Through task decomposition and collaborative execution mechanisms, the precise generation and execution of smart lock linkage control instructions are achieved, improving the flexibility and efficiency of system response. Edge computing technology is used to monitor and analyze execution effects and environmental parameters in real time, ensuring that the system can quickly adapt to environmental changes and improving the timeliness of decision-making. Knowledge graph reasoning technology is introduced to conduct in-depth analysis and model update of feedback data, achieving continuous optimization and self-improvement of the system, and improving the stability and reliability of long-term operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0023] Figure 1 A schematic diagram of an embodiment of a fire alarm linkage method applied to a smart lock in an embodiment of the present application;

[0024] Figure 2 This is a schematic diagram of an embodiment of a fire alarm linkage system applied to a smart lock in an embodiment of the present application. DETAILED DESCRIPTION

[0025] The embodiments of the present application provide a fire alarm linkage method and system for smart locks. The terms "first", "second", "third", "fourth", etc. (if any) in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "including" or "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0026] For ease of understanding, the specific process of the embodiment of the present application is described below. Figure 1 , an embodiment of the fire alarm linkage method applied to the smart lock in the embodiment of the present application includes:

[0027] Step S101: perform communication pairing and data exchange between the smart lock and the fire alarm system to obtain a device association topology diagram, and obtain local sensor data of the smart lock and fire monitoring data of the fire alarm system;

[0028] It is understandable that the execution subject of the present application can be a fire alarm linkage system applied to a smart lock, or a terminal or a server, which is not limited here. The present application embodiment is described by taking a server as the execution subject as an example.

[0029] Specifically, by matching the protocols of the communication interfaces of the smart lock and the fire alarm system, the communication between the two is ensured to be seamless, and a communication protocol parameter set is generated in the process to guide the subsequent establishment of a secure channel. The secure channel uses encryption technology to ensure the security of data transmission and form an encrypted data transmission link. For the established encrypted data transmission link, bandwidth testing and delay analysis are performed to generate a link quality assessment report. The report describes the performance of the link during the data transmission process, and according to the content of the report, the data transmission strategy is adaptively adjusted to optimize the data exchange scheme and improve the efficiency and reliability of data transmission. The physical location information of the smart lock and the fire alarm system is spatially mapped to obtain the initial device distribution map, and combined with the optimized data exchange scheme, the communication path between the devices is dynamically planned to construct a multi-level network topology. The multi-level network topology is subjected to redundancy analysis and reliability evaluation to generate network robustness indicators. These indicators are used to measure the stability and reliability of the network under different fault conditions. Based on the network robustness indicators, the communication nodes are weighted to form a device association topology map to ensure that the role and status of each node in the network are reasonably distributed. At the same time, the built-in sensors of the smart lock and the monitoring equipment of the fire alarm system are synchronized in data collection frequency to form a unified data sampling strategy to ensure the consistency and coordination of data collection. According to the unified data sampling strategy and the device association topology diagram, the real-time data stream is routed and load balanced to build an efficient data collection network. The network improves data transmission efficiency, reduces network congestion, and ensures the real-time and accuracy of data by optimizing the data routing path and balancing the load of each node. With the support of an efficient data collection network, the local sensor data of the smart lock and the fire monitoring data of the fire alarm system are continuously obtained.

[0030] Step S102: Based on the equipment association topology diagram, multi-source fusion and spatiotemporal association analysis are performed on the local sensor data and the fire monitoring data to obtain a comprehensive fire situation assessment result;

[0031] Specifically, the device association topology is structurally analyzed to determine the data transmission path and delay estimation between devices. Based on the data transmission path and delay estimation, the local sensor data and fire monitoring data are time synchronized to obtain a calibrated data set to ensure that the timestamps of each data source are consistent. Multi-scale feature extraction is performed on the calibrated data set, and multi-level feature representations of the data are obtained through feature representation methods at different levels. These feature representations can not only reveal local details of the data, but also reflect global trends. Based on the multi-level feature representation, anomaly detection and denoising are performed on the calibrated data set to remove noise and outliers and obtain optimized feature data. Dimensionality reduction is performed on the optimized feature data to reduce the data dimension and obtain a reduced dimension feature vector. In the dimensionality reduction process, the main information of the data is maintained while reducing the computational complexity. Based on the reduced dimension feature vector, nonlinear mapping is performed on the calibrated data set to obtain a fused feature space. In the fused feature space, temporal pattern matching is performed to identify the temporal and spatial correlation information of the data. This information reveals the distribution and change rules of the data in time and space. Based on the temporal and spatial correlation information, contextual reasoning is performed to obtain a scene semantic description. Perform multi-factor risk assessment on the scene semantic description, comprehensively consider multiple factors, and calculate the fire risk score. According to the fire risk score and historical data, perform probabilistic reasoning to obtain a comprehensive fire situation assessment result. This assessment result not only reflects the severity of the current fire, but also predicts the development trend of the future fire, providing a scientific basis for the linkage between smart locks and fire alarm systems.

[0032] Step S103: performing deep reinforcement learning and situational awareness processing on the comprehensive fire situation assessment results to build an adaptive decision model and differentiated response strategy;

[0033] Specifically, the features of the comprehensive fire situation assessment results are extracted to obtain fire feature vectors, which contain various key parameters and indicators of the fire. Based on the fire feature vectors, a basic decision framework is created to preliminarily determine the basic strategies and plans for responding to the fire. The basic decision framework is dynamically adjusted to obtain an initial decision model. By adjusting the parameters, the decision model can adapt to different fire conditions more flexibly. According to the initial decision model, the current environmental information is extracted to obtain a situational feature set, which reflects the specific conditions of the current fire and the surrounding environment. The situational feature set is ranked by importance to identify key situational factors. Key situational factors are the main factors affecting fire response decisions. According to these factors, the initial decision model is fine-tuned to obtain an adaptive decision model. The adaptive decision model can be adjusted according to actual conditions to ensure the effectiveness and flexibility of the response strategy. After obtaining the adaptive decision model, multiple groups of fire scenarios are input into the model to generate an alternative response strategy set. The alternative response strategy set includes various possible response measures and plans. Conflict detection and optimization are performed based on the alternative response strategy set to identify and eliminate conflicts between strategies and obtain an executable strategy subset. The effect of executable strategy subsets is estimated and sorted to generate a strategy priority list. The effect estimation evaluates the effectiveness and reliability of different strategies by simulating their execution results. According to the evaluation results, the strategies are sorted and prioritized. The optimal strategy combination is selected from the strategy priority list to obtain a differentiated response strategy. The differentiated response strategy can provide the most appropriate response measures for different fire conditions and environments to ensure efficient and safe fire handling.

[0034] Step S104: performing task decomposition and collaborative execution on the adaptive decision model and the differentiated response strategy to obtain a smart lock linkage control instruction set, and obtaining an execution effect evaluation index of the smart lock linkage control instruction set;

[0035] Specifically, the adaptive decision model is analyzed strategically to obtain a high-level task target set, which reflects the core tasks and goals of fire response. The differentiated response strategies are matched and screened according to the high-level task target set, and the executable strategy subsets that can meet these task goals are screened. The executable strategy subsets are hierarchically decomposed to obtain multi-level subtask sequences. These subtask sequences decompose complex high-level tasks into several executable specific operation steps, and build a task dependency graph based on the dependency relationship between these steps to determine the priority of task execution. According to the priority of task execution, the required resources are reasonably allocated and optimized, and a preliminary execution plan is generated to ensure that each task can be executed efficiently under resource-constrained conditions. Based on the preliminary execution plan, the smart lock control primitives are generated to form the original instruction sequence. These original instruction sequences contain specific control instructions for guiding the operation of the smart lock in the case of fire. After the original instruction sequence is generated, a semantic consistency check is performed to ensure that these instructions will not cause ambiguity or conflict during execution, and the final smart lock linkage control instruction set is obtained. According to the smart lock linkage control instruction set, the execution effect evaluation indicators are designed, including the accuracy of task completion, response speed, resource utilization and other aspects. The evaluation index system is quantified to form a computable scoring standard, so that each indicator can be quantified into a specific value for real-time evaluation. According to the computable scoring standard, the execution results are evaluated in real time to obtain the execution effect evaluation index of the smart lock linkage control instruction set. Through the evaluation, the performance of the smart lock in the fire linkage is monitored in real time to ensure that each instruction is effectively executed, and corresponding adjustments and optimizations are made according to the evaluation results.

[0036] Step S105: Real-time monitoring and edge computing analysis of the execution effect evaluation index and environmental parameter change data are performed to obtain a target feedback data stream;

[0037] Specifically, the data of the execution effect evaluation indicators are standardized to obtain a normalized evaluation indicator set, eliminating the dimensional differences between different indicators and making the indicators comparable. A multidimensional evaluation space is constructed based on the normalized evaluation indicator set to form an indicator vector matrix to ensure comprehensive analysis of various evaluation indicators in the multidimensional space. The environmental parameter change data is segmented in time series to divide the continuous environmental data into multiple time window data blocks, and the environmental change characteristics are extracted based on the time window data blocks to obtain a dynamic environmental feature set. The dynamic environmental feature set reflects the changing trends and characteristics of environmental parameters in different time periods. The dynamic environmental feature set and the indicator vector matrix are analyzed for correlation to obtain an association weight table. The association weight table is used to quantify the degree of correlation between each environmental feature and the evaluation indicator. According to these correlation weights, the data is prioritized to form a weighted data stream. The weighted data stream is analyzed by sliding windows to obtain real-time data slices. Through the sliding window method, the changing trend of the data is continuously monitored to capture short-term fluctuations in the data. According to the real-time data slices, the anomaly detection algorithm is executed to identify abnormal situations in the data and obtain abnormal event markers. Cluster analysis is performed on abnormal event markers to obtain event type classification results. Through cluster analysis, similar abnormal events are classified and different types of abnormal phenomena are identified. According to the event type classification results and weighted data streams, data packets are converted to form target feedback data streams. The target feedback data stream integrates the results of real-time monitoring and analysis and provides real-time feedback on the system operation status.

[0038] Step S106: perform knowledge graph reasoning on the target feedback data stream to obtain decision model update parameters, and update the adaptive decision model based on the decision model update parameters to obtain the target decision model.

[0039] Specifically, semantic parsing is performed on the target feedback data stream to obtain a structured data set. Through semantic parsing, unstructured data is converted into structured data with clear meanings. Key entities and relationships are extracted from the structured data set to form a knowledge triple set. The triple set reflects the basic elements in the data and their relationships. Ontology mapping is performed on the knowledge triple set to obtain a domain knowledge graph. Through ontology mapping, the knowledge triple set is matched with the predefined domain knowledge to ensure the accuracy and consistency of the knowledge graph. Based on the domain knowledge graph, an inference rule library is constructed to form a graph inference engine. The inference rule library contains various inference rules for inference operations in the knowledge graph. Path query analysis is performed on the graph inference engine to obtain a causal relationship chain. Through path query, the causal relationship between entities in the knowledge graph is revealed to form a causal relationship chain. According to the causal relationship chain, a decision impact assessment is performed to obtain a decision sensitivity matrix. The decision sensitivity matrix is ​​used to quantify the degree of influence of each decision factor on the final decision result. The decision sensitivity matrix is ​​sorted by feature importance, the features that have the greatest impact on the decision are identified, and the decision model update parameters are obtained. The updated parameters will be used to adjust the adaptive decision model to improve the accuracy and adaptability of the model. According to the updated parameters of the decision model, the parameters of the adaptive decision model are adjusted to obtain the updated model structure. The updated model structure is verified and tested. Through a series of tests, the performance of the model is evaluated and a model performance evaluation report is generated. The model performance evaluation report contains the performance of the model under various test conditions. According to the model performance evaluation report, model integration is performed to integrate the updated model structure with the existing model to form the final target decision model.

[0040] In the embodiment of the present application, by pairing and exchanging data between the smart lock and the fire alarm system, and building a device association topology map, seamless integration between systems is achieved, and the efficiency and reliability of data transmission are improved. Multi-source fusion and spatiotemporal correlation analysis technology are used to comprehensively process local sensor data and fire monitoring data, significantly improving the accuracy and comprehensiveness of fire situation assessment. Deep reinforcement learning and situational awareness processing technology are introduced to build adaptive decision-making models and differentiated response strategies, enhancing the system's ability to cope with complex fire situations. Through task decomposition and collaborative execution mechanisms, the precise generation and execution of smart lock linkage control instructions are achieved, improving the flexibility and efficiency of system response. Edge computing technology is used to monitor and analyze execution effects and environmental parameters in real time, ensuring that the system can quickly adapt to environmental changes and improving the timeliness of decision-making. The introduction of knowledge graph reasoning technology, in-depth analysis of feedback data and model updates, achieves continuous optimization and self-improvement of the system, and improves the stability and reliability of long-term operation.

[0041] In a specific embodiment, the process of executing step S101 may specifically include the following steps:

[0042] (1) Perform protocol matching on the communication interface of the smart lock and the fire alarm system to obtain a communication protocol parameter set, and establish a secure channel between the smart lock and the fire alarm system according to the communication protocol parameter set to obtain an encrypted data transmission link;

[0043] (2) Perform bandwidth testing and delay analysis on the encrypted data transmission link to obtain a link quality evaluation report, and adaptively adjust the data transmission strategy based on the link quality evaluation report to obtain an optimized data exchange solution;

[0044] (3) The physical location information of the smart lock and the fire alarm system is spatially mapped to obtain an initial device distribution map. The communication paths between devices are dynamically planned based on the initial device distribution map and the optimized data exchange scheme to obtain a multi-level network topology structure.

[0045] (4) Perform redundancy analysis and reliability evaluation on the multi-level network topology structure to obtain network robustness indicators, and assign weights to communication nodes based on the network robustness indicators to obtain a device association topology diagram;

[0046] (5) Synchronize the data collection frequency of the built-in sensor of the smart lock and the monitoring equipment of the fire alarm system to obtain a unified data sampling strategy;

[0047] (6) Based on the unified data sampling strategy and device association topology, the real-time data stream is routed and load balanced to obtain an efficient data acquisition network, and the local sensor data of the smart lock and the fire monitoring data of the fire alarm system are continuously obtained.

[0048] Specifically, perform protocol matching on the communication interfaces of the smart lock and the fire alarm system to obtain a communication protocol parameter set. Identify and compare the communication protocols used by the smart lock and the fire alarm system, and form a unified communication protocol parameter set by matching parameters such as protocol version, data format, and encryption method. For example, assuming that the smart lock uses the Zigbee protocol and the fire alarm system uses the Wi-Fi protocol, communication between the two can be achieved through a protocol conversion gateway. Establish a secure channel between the smart lock and the fire alarm system based on the communication protocol parameter set to ensure the confidentiality and integrity of data transmission. The encrypted data transmission link is implemented through the SSL / TLS protocol, which exchanges keys through a handshake process and uses a symmetric encryption algorithm, such as AES (Advanced Encryption Standard), for encryption in subsequent data transmission. The formula is as follows:

[0049] C=E k (P);

[0050] Among them, C represents the encrypted ciphertext, E represents the encryption function, k represents the symmetric key, and P represents the original data. After establishing the encrypted data transmission link, the link is tested for bandwidth and latency to evaluate its performance. By sending and receiving a series of test data packets, the transmission time and throughput of the data packets are measured. Assuming that the size of the test data packet sent is S and the transmission time is T, the bandwidth B can be expressed as:

[0051]

[0052] The results of bandwidth testing and delay analysis form a link quality assessment report, based on which the data transmission strategy can be adaptively adjusted. For example, if the link bandwidth is insufficient or the delay is too high, the packet size or transmission frequency can be adjusted, and the optimized data exchange scheme ensures the stability and efficiency of data transmission. The physical location information of the smart lock and the fire alarm system is spatially mapped to obtain the initial device distribution map. By combining the physical location and communication range of each device, the connection relationship between the devices is preliminarily drawn. Assuming that the smart lock and the fire alarm system are located in multiple rooms respectively, the location of each device can be represented by a two-dimensional coordinate system, such as the smart lock is located at the coordinates (x1, y1) and the fire alarm system is located at the coordinates (x2, y2). Based on the initial device distribution map and the optimized data exchange scheme, the communication path between devices is dynamically planned to form a multi-level network topology. The optimal path between devices is found through the Dijkstra algorithm or the A* algorithm to ensure low latency and high reliability of data transmission. The network topology can be divided into different layers, such as the core layer, the aggregation layer, and the access layer. The devices in each layer are classified and connected according to their functions and locations. Redundancy analysis and reliability evaluation are performed on the multi-level network topology to obtain network robustness indicators. Redundancy analysis ensures that there are multiple backup paths in the network. When a path fails, the system can quickly switch to the backup path to maintain normal communication. Assume that the reliability of each node in the network is R i , the total reliability R can be expressed as:

[0053]

[0054] Among them, n is the number of network nodes, and the overall reliability of the network can be obtained by calculation. According to the network robustness index, the communication nodes are weighted to form a device association topology. The weight allocation takes into account factors such as node reliability, transmission delay, and bandwidth to ensure that key nodes have a higher priority, thereby optimizing the performance and stability of the entire network. The built-in sensors of the smart lock and the monitoring equipment of the fire alarm system are synchronized to obtain a unified data sampling strategy. By unifying the sampling frequency, it is ensured that the data of each device can be collected and transmitted synchronously to avoid analysis errors caused by data asynchrony. Assuming that the sampling frequency of the smart lock is f1 and the sampling frequency of the fire alarm system is f2, the unified sampling frequency f can be achieved by adjusting the sampling period of the two to meet a common frequency f. According to the unified data sampling strategy and device association topology, the real-time data stream is routed and load balanced to form an efficient data acquisition network. Route optimization selects the optimal path to transmit data and reduces transmission delay; load balancing allocates network resources to avoid overloading of a certain node. Load balancing can be achieved by evenly distributing data streams to each node through a hash algorithm or a polling algorithm. In the optimized data acquisition network, the local sensor data of the smart lock and the fire monitoring data of the fire alarm system are continuously obtained to form an efficient and reliable data collection and transmission system. Assuming that the sensor data of the smart lock is D1 and the monitoring data of the fire alarm system is D2, the data of the two are fused through data fusion algorithms, such as weighted average method or Kalman filtering, to obtain a comprehensive data set D

[0055] D=w1·D1+w2·D2;

[0056] Among them, w1 and w2 are the data weights of the smart lock and fire alarm system respectively.

[0057] In a specific embodiment, the process of executing step S102 may specifically include the following steps:

[0058] (1) Perform structural analysis on the device association topology to obtain the data transmission path and delay estimation between devices;

[0059] (2) Based on the data transmission path and delay estimation, the local sensor data and fire monitoring data are time synchronized to obtain a calibrated data set;

[0060] (3) performing multi-scale feature extraction on the calibrated data set to obtain a multi-level feature representation, and performing anomaly detection and denoising on the calibrated data set based on the multi-level feature representation to obtain optimized feature data;

[0061] (4) Performing dimensionality reduction processing on the optimized feature data to obtain a reduced-dimensionality feature vector, and performing nonlinear mapping on the calibrated data set based on the reduced-dimensionality feature vector to obtain a fused feature space;

[0062] (5) Perform temporal pattern matching on the fused feature space to obtain spatiotemporal correlation information, and perform contextual reasoning based on the spatiotemporal correlation information to obtain a semantic description of the scene;

[0063] (6) Perform multi-factor risk assessment on the scene semantic description to obtain a fire risk score, and perform probabilistic reasoning based on the fire risk score and historical data to obtain a comprehensive fire situation assessment result.

[0064] Specifically, a structural analysis is performed on the device association topology diagram to determine the data transmission path and delay estimate between devices. Suppose there is a network containing multiple smart locks and fire alarm system nodes. The physical location and connection status of each node constitute a topology diagram. Through the shortest path algorithm in graph theory, such as the Dijkstra algorithm, the shortest transmission path between each pair of devices is found, and its transmission delay is calculated. The transmission delay can be expressed as the sum of the node transmission delays on each path. According to the data transmission path and delay estimate, the local sensor data and fire monitoring data are time synchronized to obtain a calibrated data set to ensure that data from different devices can be compared and analyzed under the same time reference. Assume that the data timestamp of the smart lock is t lock , the data timestamp of the fire alarm system is t alarm , through time synchronization algorithms, such as Network Time Protocol, the data timestamps of different devices are calibrated to a unified time base t sync, so that the calibrated data set has a consistent timestamp. Perform multi-scale feature extraction on the calibrated data set to obtain a multi-level feature representation. The methods for multi-scale feature extraction may include wavelet transform, Fourier transform, etc., through which the original data is analyzed at different scales to extract features at different levels. For example, in wavelet transform, the signal is decomposed into a combination of different frequency components, which respectively represent the details and overall trends of the data. By analyzing these features, a multi-level feature representation of the data is obtained. According to the multi-level feature representation, anomaly detection and denoising are performed on the calibrated data set to obtain optimized feature data. The purpose of anomaly detection is to identify and remove outliers in the data, which may be caused by sensor failure or data transmission errors. Denoising is to eliminate noise in the data and improve clarity and accuracy. Use statistical-based methods, such as mean and standard deviation, to identify outliers, or use machine learning algorithms such as isolation forest algorithm for anomaly detection. Perform dimensionality reduction on the optimized feature data to obtain a reduced-dimensional feature vector. The purpose of dimensionality reduction is to reduce the dimension of the feature space, thereby reducing the computational complexity while maintaining the main information of the data. Commonly used dimensionality reduction algorithms include principal component analysis and t-SNE. Assume that the optimized feature data is X, and after PCA processing, the reduced dimension feature vector is Y:

[0065] Y=W T X;

[0066] Among them, W is the eigenvector matrix. By selecting the first k principal components, the dimension of the feature can be effectively reduced. According to the reduced dimensionality eigenvector, the calibrated data set is nonlinearly mapped to obtain the fused feature space. Nonlinear mapping can be achieved through kernel methods or neural networks to map data in high-dimensional space to low-dimensional space while maintaining the structural relationship of the data. For example, using kernel PCA in the kernel method, the eigenvector is mapped into a nonlinear space to obtain the fused feature space. In the fused feature space, time series pattern matching is performed to identify the spatiotemporal correlation information in the data. Time series pattern matching methods can include dynamic time warping and hidden Markov models, through which patterns and regularities in the data are identified. Assuming that the time series data is X(t), through DTW, similar patterns in the data can be found and their distance D can be calculated:

[0067]

[0068] According to the spatiotemporal correlation information, contextual reasoning is performed to obtain the scene semantic description. The contextual reasoning method can be implemented through a rule engine or a Bayesian network to convert the correlation information in the data into a scene description. For example, through a rule engine, according to predefined rules, a specific sensor data pattern is identified as a fire alarm event, thereby generating a corresponding scene semantic description. A multi-factor risk assessment is performed on the scene semantic description to obtain a fire risk score. The multi-factor risk assessment method can include fuzzy logic and multi-criteria decision analysis, and the fire risk score is calculated by comprehensively considering the influence of different factors. Probabilistic reasoning is performed based on the fire risk score and historical data to obtain a comprehensive fire situation assessment result. The probabilistic reasoning method can be implemented through Bayesian reasoning or a Markov decision process to compare the current risk score with the historical data to predict the future development trend of the fire. Assuming that the probability distribution of the historical data is P and the current fire risk score is R, the comprehensive fire situation assessment result can be expressed as a conditional probability:

[0069] P(R|historical data)=P(R∩historical data) / P(historical data);

[0070] In a specific embodiment, the process of executing step S103 may specifically include the following steps:

[0071] (1) Extract features from the comprehensive fire situation assessment results to obtain fire feature vectors, and create a basic decision framework based on the fire feature vectors;

[0072] (2) Dynamically adjust the parameters of the basic decision framework to obtain an initial decision model, and extract the current environmental information based on the initial decision model to obtain a situational feature set;

[0073] (3) Rank the situational feature set by importance to obtain key situational factors, and fine-tune the initial decision model based on the key situational factors to obtain an adaptive decision model;

[0074] (4) Input multiple groups of fire scenarios into the adaptive decision model to obtain a set of alternative response strategies, and perform conflict detection and optimization based on the set of alternative response strategies to obtain an executable strategy subset;

[0075] (5) Estimate and sort the effects of the executable strategy subsets to obtain a strategy priority list, and select the optimal strategy combination from the strategy priority list to obtain a differentiated response strategy.

[0076] Specifically, the comprehensive fire situation assessment results are feature extracted to obtain the fire situation feature vector. Through feature extraction, the complex fire situation assessment results are converted into a set of key parameters, which can accurately reflect various important information of the fire situation. Assuming that the fire situation assessment results include multiple indicators such as temperature, smoke concentration, and flame detection, these indicators are extracted as feature vectors through algorithms such as principal component analysis. The feature vector can be expressed as:

[0077] F=[f1,f2,…,f n ];

[0078] Among them, F represents the fire characteristic vector, f i Represents the specific value of each feature. Create a basic decision framework based on the fire feature vector. The basic decision framework contains basic strategies and measures for dealing with different fire conditions. By mapping the feature vector to specific decision rules, the strategy for dealing with the fire is preliminarily determined. For example, if the temperature feature value f1 exceeds a certain threshold and the smoke concentration feature value f2 exceeds the standard at the same time, the basic decision framework may recommend triggering a fire alarm immediately and opening the smart lock. Dynamically adjust the parameters of the basic decision framework to adapt to different fire scenarios and obtain an initial decision model. Dynamic parameter adjustment can be achieved through machine learning algorithms, such as through reinforcement learning algorithms, to continuously optimize decision parameters based on historical fire data and current environmental feedback. Assuming that the basic decision parameter is θ, after adjustment by reinforcement learning, the optimized initial decision parameter θ′ is obtained:

[0079]

[0080] Among them, α is the learning rate and J(θ) is the decision benefit function. According to the initial decision model, the current environmental information is extracted to form a situational feature set. The situational feature set includes all key information in the current environment, such as temperature, humidity, personnel location, etc. Through the sensor network and data acquisition system, these environmental information are obtained in real time and integrated into the situational feature set. The situational feature set is sorted by importance to identify key situational factors. Through feature selection algorithms, such as information gain or mutual information, each feature is sorted by calculating its contribution to the decision result. After the key situational factors are sorted, determine which factors are most important in the current fire environment, so as to fine-tune the initial decision model. Assume that the key situational factor is C=[c1,c2,…,c m ], and adjust the initial decision model according to these factors to obtain the adaptive decision model:

[0081]

[0082] Among them, β is the adjustment coefficient, w iis the weight of each situational factor. Multiple groups of fire scenarios are input into the adaptive decision model to generate a set of alternative response strategies. These fire scenarios can be generated through simulation experiments or historical data, including fires of different types and severity. The adaptive decision model generates a series of possible response strategies based on these scenarios, such as evacuation route selection, alarm mode adjustment, etc. Conflict detection and optimization are performed on the set of alternative response strategies to ensure that there is no conflict between the strategies. For example, if one strategy requires closing all door locks and another strategy requires opening door locks, the system needs to perform conflict detection and select the optimal solution. The optimized results form an executable strategy subset, and strategy selection and combination are performed through optimization algorithms such as genetic algorithms. The executable strategy subset is estimated and sorted to generate a strategy priority list. The effect estimation evaluates its impact on fire control and personnel safety by simulating the strategy execution process. The estimation results can be obtained through simulation models or historical execution data. Assume that the strategy set is S = {s1, s2,…, s k}, estimate the effect E(s) of each strategy i ), and sort them according to the effect value to form a priority list:

[0083] Priority(s i )=E(s i )

[0084] Select the optimal strategy combination from the strategy priority list to form a differentiated response strategy. The differentiated response strategy takes into account different fire scenarios and environmental changes and provides customized response measures.

[0085] In a specific embodiment, the process of executing step S104 may specifically include the following steps:

[0086] (1) Performing strategy analysis on the adaptive decision model to obtain a high-level task goal set, and matching and screening the differentiated response strategies based on the high-level task goal set to obtain an executable strategy subset;

[0087] (2) Perform hierarchical decomposition on the executable strategy subset to obtain a multi-level subtask sequence, and construct a task dependency graph based on the multi-level subtask sequence to obtain the task execution priority ranking;

[0088] (3) Perform resource allocation and scheduling optimization on the task execution priority to obtain a preliminary execution plan, and generate smart lock control primitives based on the preliminary execution plan to obtain the original instruction sequence;

[0089] (4) Performing a semantic consistency check on the original instruction sequence to obtain a smart lock linkage control instruction set, and designing execution effect evaluation indicators based on the smart lock linkage control instruction set to obtain an evaluation indicator system;

[0090] (5) The evaluation index system is quantified to obtain a computable scoring standard, and the execution results are evaluated in real time according to the computable scoring standard to obtain the execution effect evaluation index of the smart lock linkage control instruction set.

[0091] Specifically, the adaptive decision model is analyzed to obtain a high-level task goal set. By analyzing the decision rules and response strategies in the model, the main task goals that the system needs to achieve in different fire scenarios are identified. These task goals may include personnel evacuation, fire alarm, and activation of fire extinguishing systems. Assume that in a certain fire scenario, the main task goal set is T = {t1, t2, …, t n}, where t i Represents a specific task goal, for example, t1 represents "starting the fire fighting system" and t2 represents "opening the emergency exit". According to the high-level task goal set, the differentiated response strategies are matched and screened to obtain an executable strategy subset. By comparing the task goal with the predefined response strategy, the strategy that meets the current fire scenario is screened out. Assume that the differentiated response strategy set is S = {s1, s2, ..., s m}, through the strategy matching algorithm, filter out the executable strategy subset that meets the task target set T Perform hierarchical decomposition on the executable strategy subset to obtain a multi-level subtask sequence. Hierarchical decomposition decomposes each strategy into specific operation steps or subtasks. For example, starting the fire protection system can be decomposed into "detecting fire signals", "activating pump stations", "opening sprinkler valves", etc. Through decomposition, a more detailed task sequence U = {u1,u2,…,u k}, each subtask u i Corresponding to the specific execution steps. According to the multi-level subtask sequence, a task dependency graph is constructed to obtain the task execution priority ranking. The task dependency graph represents the dependency relationship between tasks through nodes and edges to ensure that tasks are executed in the correct order. Assuming that the nodes in the task dependency graph represent subtasks and the edges represent dependency relationships, a topological sorting algorithm can be used to prioritize the tasks and obtain the sorted task sequence V = {v1, v2, …, v k}, where v i represents the prioritized subtasks. We perform resource allocation and scheduling optimization on the priority sorting of task execution to obtain a preliminary execution plan. Resource allocation and scheduling optimization considers the resource constraints of the system, such as network bandwidth, processing power, energy consumption, etc., and reasonably allocates resources to ensure efficient execution of tasks. Assume that each task v i Resources required i , the total resources of the system are R, and the optimization goal is to maximize resource utilization. The formula is as follows:

[0092]

[0093] Generate smart lock control primitives according to the preliminary execution plan and obtain the original instruction sequence. Smart lock control primitives are basic control commands. By combining commands, complex operations can be achieved. Assume that the tasks determined by the preliminary execution plan are V = {v1, v2, …, v k}, the corresponding control primitive is C = {c1,c2,…,c k}, the original instruction sequence can be expressed as:

[0094] C={c1,c2,…,c k};

[0095] Perform semantic consistency check on the original instruction sequence to ensure that each instruction does not cause conflict or misunderstanding during execution, and obtain the intelligent lock linkage control instruction set. The semantic consistency check verifies the logical relationship and execution order of the instructions to ensure their accuracy and consistency. Design execution effect evaluation indicators based on the intelligent lock linkage control instruction set to form an evaluation indicator system. The evaluation indicator system is used to measure the effect of instruction set execution, including response time, task completion rate, resource utilization, etc. Assume that the evaluation indicator is E = {e1, e2, …, e m}, where e i Represents a specific evaluation indicator, for example, e1 represents the response time, and e2 represents the task completion rate. The evaluation indicator system is quantified to obtain a computable scoring standard. The scoring standard quantifies each evaluation indicator so that it can be compared and analyzed. Assume that each evaluation indicator e i The weight is w i , then the overall score S can be expressed as:

[0096]

[0097] According to the computable scoring criteria, the execution results are evaluated in real time to obtain the execution effect evaluation index of the smart lock linkage control instruction set. By monitoring the actual execution effect of the system, the evaluation results are fed back to the scoring system, and the scores are updated in real time to obtain dynamic evaluation indicators.

[0098] In a specific embodiment, the process of executing step S105 may specifically include the following steps:

[0099] (1) Performing data standardization on the execution effect evaluation index to obtain a normalized evaluation index set, and constructing a multidimensional evaluation space based on the normalized evaluation index set to obtain an index vector matrix;

[0100] (2) Performing time series segmentation on the environmental parameter change data to obtain multiple time window data blocks, and extracting environmental change features based on the time window data blocks to obtain a dynamic environmental feature set;

[0101] (3) Perform correlation analysis on the dynamic environment feature set and the indicator vector matrix to obtain a correlation weight table, and prioritize the data according to the correlation weight table to obtain a weighted data stream;

[0102] (4) Perform sliding window analysis on the weighted data stream to obtain real-time data slices, and execute anomaly detection algorithms based on the real-time data slices to obtain abnormal event markers;

[0103] (5) Cluster analysis is performed on the abnormal event markers to obtain event type classification results, and data packets are converted based on the event type classification results and weighted data streams to obtain target feedback data streams.

[0104] Specifically,

[0105] In a specific embodiment, the process of executing step S106 may specifically include the following steps:

[0106] (1) Perform semantic analysis on the target feedback data stream to obtain a structured data set, and extract key entities and relationships based on the structured data set to obtain a set of knowledge triples;

[0107] (2) Perform ontology mapping on the knowledge triple set to obtain a domain knowledge graph, and build an inference rule base based on the domain knowledge graph to obtain a graph inference engine;

[0108] (3) Perform path query analysis on the graph reasoning engine to obtain the causal chain, and perform decision impact assessment based on the causal chain to obtain the decision sensitivity matrix;

[0109] (4) Sort the feature importance of the decision sensitivity matrix to obtain the update parameters of the decision model;

[0110] (5) adjusting the parameters of the adaptive decision model according to the decision model update parameters to obtain an updated model structure;

[0111] (6) Perform verification tests on the updated model structure to obtain a model performance evaluation report, and perform model integration based on the model performance evaluation report to obtain the target decision model.

[0112] Specifically, the execution effect evaluation indicators are standardized, and data of different dimensions are converted into dimensionless data, so that they can be compared and analyzed on the same scale to obtain a normalized evaluation indicator set. A multidimensional evaluation space is constructed based on the normalized evaluation indicator set. In the multidimensional evaluation space, each normalized indicator represents a dimension, and these dimensions together constitute the evaluation space. Assuming there are m evaluation indicators, each indicator corresponds to a dimension, then a point P in the multidimensional evaluation space can be represented as an m-dimensional vector:

[0113] P=(z1,z2,…,zm );

[0114] In this way, multiple evaluation indicators are integrated into a unified framework to construct an indicator vector matrix M, where each row represents a sample and each column represents an evaluation indicator. The environmental parameter change data is segmented in time series, and the continuous environmental data is divided into multiple time window data blocks. Assuming that the environmental parameter data is E(t), it is divided into multiple time windows W through time series segmentation. k , each time window contains environmental data for a period of time:

[0115] W k = {E(t1),E(t2),…,E(t n )};

[0116] In each time window, the environmental change features are extracted to form a dynamic environmental feature set F. Assuming that a feature is temperature T, humidity H, and wind speed V, the dynamic environmental feature set can be expressed as:

[0117] F k ={T k ,H k ,V k};

[0118] The correlation analysis of the dynamic environmental feature set and the indicator vector matrix is ​​performed to determine the degree of influence of each environmental feature on the evaluation index. By calculating the correlation coefficient r between the feature and the indicator, the correlation weight table is obtained. The calculation formula of the correlation coefficient is as follows:

[0119]

[0120] Among them, r ij Indicates environmental characteristics F i And the evaluation index M j The correlation coefficient of and are the means of the features and indicators, respectively. Based on the correlation coefficient, an association weight table R is constructed, and the data is prioritized according to the weight to obtain a weighted data stream. Sliding window analysis is performed on the weighted data stream to obtain real-time data slices. Sliding window analysis calculates relevant statistics in each window by moving a fixed-size window on the data stream. Assuming the sliding window size is w, the data slice at time t is:

[0121] S(t)={D(t-w+1),D(t-w+2),…,D(t)};

[0122] Based on the real-time data slices, anomaly detection algorithms are executed to identify abnormal events in the data. Anomaly detection algorithms can use statistical methods, such as z-score detection, or machine learning methods, such as isolation forest. Assume that the detected anomaly value is A = {a1, a2, ..., a k}, these outliers will be marked as abnormal events. Perform cluster analysis on the abnormal event labels to obtain the event type classification results. Cluster analysis can use the k-means algorithm or DBSCAN algorithm to identify different types of abnormal events by clustering similar abnormal events together. Assume that the clustering result is C = {c1, c2, …, c m}, where each c i Represents a type of abnormal event. Data packet conversion is performed based on the event type classification result and the weighted data stream to obtain the target feedback data stream. Data packet conversion combines the classification result with the weighted data stream to form a new data packet format so that it can adapt to the subsequent processing flow. The target feedback data stream T can be expressed as:

[0123] T = {C, S};

[0124] The fire alarm linkage method applied to the smart lock in the embodiment of the present application is described above. The fire alarm linkage system applied to the smart lock in the embodiment of the present application is described below. Figure 2 In the embodiment of the present application, an embodiment of the fire alarm linkage system applied to the smart lock includes:

[0125] The acquisition module 201 is used to perform communication pairing and data exchange between the smart lock and the fire alarm system, obtain a device association topology diagram, and obtain local sensor data of the smart lock and fire monitoring data of the fire alarm system;

[0126] Fusion module 202, used to perform multi-source fusion and spatiotemporal correlation analysis on local sensor data and fire monitoring data based on the equipment association topology diagram to obtain a comprehensive fire situation assessment result;

[0127] Building module 203, for performing deep reinforcement learning and situational awareness processing on the comprehensive fire situation assessment results, and building an adaptive decision model and a differentiated response strategy;

[0128] An execution module 204 is used to perform task decomposition and collaborative execution of the adaptive decision model and the differentiated response strategy, obtain a smart lock linkage control instruction set, and obtain an execution effect evaluation index of the smart lock linkage control instruction set;

[0129] Analysis module 205, used to perform real-time monitoring and edge computing analysis on the execution effect evaluation index and environmental parameter change data to obtain the target feedback data stream;

[0130] The updating module 206 is used to perform knowledge graph reasoning on the target feedback data stream to obtain decision model update parameters, and to update the adaptive decision model based on the decision model update parameters to obtain the target decision model.

[0131] Through the collaboration of the above components, by pairing and exchanging data between smart locks and fire alarm systems, and building a device association topology, seamless integration between systems is achieved, and the efficiency and reliability of data transmission are improved. Multi-source fusion and spatiotemporal correlation analysis technology are used to comprehensively process local sensor data and fire monitoring data, significantly improving the accuracy and comprehensiveness of fire situation assessment. Deep reinforcement learning and situational awareness processing technology are introduced to build adaptive decision-making models and differentiated response strategies, enhancing the system's ability to cope with complex fire situations. Through task decomposition and collaborative execution mechanisms, the precise generation and execution of smart lock linkage control instructions are achieved, improving the flexibility and efficiency of system response. Edge computing technology is used to monitor and analyze execution effects and environmental parameters in real time, ensuring that the system can quickly adapt to environmental changes and improving the timeliness of decision-making. Knowledge graph reasoning technology is introduced to conduct in-depth analysis and model update of feedback data, achieving continuous optimization and self-improvement of the system, and improving the stability and reliability of long-term operation.

[0132] The present application also provides a fire alarm linkage device applied to a smart lock, and the fire alarm linkage device applied to the smart lock includes a memory and a processor, and the memory stores computer-readable instructions. When the computer-readable instructions are executed by the processor, the processor executes the steps of the fire alarm linkage method applied to the smart lock in the above-mentioned embodiments.

[0133] The present application also provides a computer-readable storage medium, which may be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium. Instructions are stored in the computer-readable storage medium. When the instructions are executed on a computer, the computer executes the steps of the fire alarm linkage method applied to a smart lock.

[0134] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described systems, systems and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0135] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk and other media that can store program code.

[0136] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A fire alarm linkage method applied to a smart lock, characterized in that: The method comprises: Perform communication pairing and data exchange between the smart lock and the fire alarm system to obtain a device association topology diagram, and acquire local sensor data of the smart lock and fire monitoring data of the fire alarm system; Based on the equipment association topology diagram, multi-source fusion and spatiotemporal association analysis are performed on the local sensor data and the fire monitoring data to obtain a comprehensive fire situation assessment result; Perform deep reinforcement learning and situational awareness processing on the comprehensive fire situation assessment results to build an adaptive decision-making model and differentiated response strategy; Decomposing and co-executing the adaptive decision model and the differentiated response strategy to obtain a smart lock linkage control instruction set, and obtaining an execution effect evaluation index of the smart lock linkage control instruction set; Perform real-time monitoring and edge computing analysis on the execution effect evaluation index and environmental parameter change data to obtain a target feedback data stream; Knowledge graph reasoning is performed on the target feedback data stream to obtain decision model update parameters, and the adaptive decision model is updated based on the decision model update parameters to obtain a target decision model.

2. The fire alarm linkage method for smart locks according to claim 1 is characterized in that: The communication pairing and data exchange between the smart lock and the fire alarm system to obtain a device association topology diagram, and to obtain the local sensor data of the smart lock and the fire monitoring data of the fire alarm system include: Perform protocol matching on the communication interface of the smart lock and the fire alarm system to obtain a communication protocol parameter set, and establish a secure channel between the smart lock and the fire alarm system according to the communication protocol parameter set to obtain an encrypted data transmission link; Performing bandwidth testing and delay analysis on the encrypted data transmission link to obtain a link quality evaluation report, and adaptively adjusting the data transmission strategy according to the link quality evaluation report to obtain an optimized data exchange solution; Performing spatial mapping on the physical location information of the smart lock and the fire alarm system to obtain an initial device distribution map, and dynamically planning the communication paths between devices according to the initial device distribution map and the optimized data exchange scheme to obtain a multi-level network topology structure; Performing redundancy analysis and reliability evaluation on the multi-level network topology structure to obtain a network robustness index, and assigning weights to communication nodes according to the network robustness index to obtain a device association topology diagram; Synchronize the data acquisition frequency of the built-in sensor of the smart lock and the monitoring equipment of the fire alarm system to obtain a unified data sampling strategy; According to the unified data sampling strategy and the device association topology diagram, the real-time data stream is routed and load balanced to obtain an efficient data acquisition network, and the local sensor data of the smart lock and the fire monitoring data of the fire alarm system are continuously obtained.

3. The fire alarm linkage method applied to smart locks according to claim 2 is characterized in that: The multi-source fusion and spatiotemporal correlation analysis of the local sensor data and the fire monitoring data based on the device association topology diagram is performed to obtain a comprehensive fire situation assessment result, including: Performing structural analysis on the device association topology diagram to obtain data transmission paths and delay estimates between devices; According to the data transmission path and delay estimation, time synchronization processing is performed on the local sensor data and the fire monitoring data to obtain a calibrated data set; Performing multi-scale feature extraction on the calibrated data set to obtain a multi-level feature representation, and performing anomaly detection and denoising on the calibrated data set based on the multi-level feature representation to obtain optimized feature data; Performing dimensionality reduction processing on the optimized feature data to obtain a reduced-dimensionality feature vector, and performing nonlinear mapping on the calibrated data set according to the reduced-dimensionality feature vector to obtain a fused feature space; Performing temporal pattern matching on the fused feature space to obtain spatiotemporal correlation information, and performing contextual reasoning based on the spatiotemporal correlation information to obtain a scene semantic description; A multi-factor risk assessment is performed on the scene semantic description to obtain a fire risk score, and probabilistic reasoning is performed based on the fire risk score and historical data to obtain a comprehensive fire situation assessment result.

4. The fire alarm linkage method applied to smart locks according to claim 1, characterized in that: The method of performing deep reinforcement learning and situational awareness processing on the comprehensive fire situation assessment results to construct an adaptive decision model and a differentiated response strategy includes: Extracting features from the comprehensive fire situation assessment results to obtain a fire situation feature vector, and creating a basic decision framework based on the fire situation feature vector; Dynamically adjusting parameters of the basic decision framework to obtain an initial decision model, and extracting current environmental information according to the initial decision model to obtain a situational feature set; Sorting the situational feature set by importance to obtain key situational factors, and fine-tuning the initial decision model according to the key situational factors to obtain an adaptive decision model; Inputting multiple groups of fire scenarios into the adaptive decision model to obtain a set of alternative response strategies, and performing conflict detection and optimization according to the set of alternative response strategies to obtain an executable strategy subset; The executable strategy subsets are estimated and ranked to obtain a strategy priority list, and an optimal strategy combination is selected from the strategy priority list to obtain a differentiated response strategy.

5. The fire alarm linkage method applied to smart locks according to claim 1, characterized in that: The task decomposition and collaborative execution of the adaptive decision model and the differentiated response strategy to obtain a smart lock linkage control instruction set, and obtaining an execution effect evaluation index of the smart lock linkage control instruction set, include: Performing strategy analysis on the adaptive decision model to obtain a high-level task goal set, and matching and screening the differentiated response strategy according to the high-level task goal set to obtain an executable strategy subset; Decomposing the executable strategy subset hierarchically to obtain a multi-level subtask sequence, and constructing a task dependency graph based on the multi-level subtask sequence to obtain a task execution priority ranking; Perform resource allocation and scheduling optimization on the task execution priority sorting to obtain a preliminary execution plan, and generate smart lock control primitives according to the preliminary execution plan to obtain an original instruction sequence; Performing a semantic consistency check on the original instruction sequence to obtain a smart lock linkage control instruction set, and designing an execution effect evaluation index based on the smart lock linkage control instruction set to obtain an evaluation index system; The evaluation index system is quantified to obtain a computable scoring standard, and the execution result is evaluated in real time according to the computable scoring standard to obtain the execution effect evaluation index of the smart lock linkage control instruction set.

6. The fire alarm linkage method for smart locks according to claim 1, characterized in that: The real-time monitoring and edge computing analysis of the execution effect evaluation index and environmental parameter change data to obtain the target feedback data stream includes: Performing data standardization processing on the execution effect evaluation index to obtain a normalized evaluation index set, and constructing a multidimensional evaluation space according to the normalized evaluation index set to obtain an index vector matrix; Performing time series segmentation on the environmental parameter change data to obtain a plurality of time window data blocks, and extracting environmental change features according to the time window data blocks to obtain a dynamic environmental feature set; Performing correlation analysis on the dynamic environment feature set and the indicator vector matrix to obtain an associated weight table, and prioritizing the data according to the associated weight table to obtain a weighted data stream; Performing sliding window analysis on the weighted data stream to obtain real-time data slices, and executing an anomaly detection algorithm based on the real-time data slices to obtain an abnormal event marker; Cluster analysis is performed on the abnormal event markers to obtain event type classification results, and data packet conversion is performed based on the event type classification results and the weighted data stream to obtain a target feedback data stream.

7. The fire alarm linkage method for smart locks according to claim 1, characterized in that: The performing knowledge graph reasoning on the target feedback data stream to obtain decision model update parameters, and performing model update on the adaptive decision model based on the decision model update parameters to obtain a target decision model, includes: Performing semantic parsing on the target feedback data stream to obtain a structured data set, and extracting key entities and relationships based on the structured data set to obtain a knowledge triple set; Performing ontology mapping on the knowledge triple set to obtain a domain knowledge graph, and constructing an inference rule library based on the domain knowledge graph to obtain a graph inference engine; Performing path query analysis on the graph reasoning engine to obtain a causal relationship chain, and performing decision impact assessment based on the causal relationship chain to obtain a decision sensitivity matrix; Sorting the decision sensitivity matrix by feature importance to obtain decision model update parameters; Adjusting the parameters of the adaptive decision model according to the decision model update parameters to obtain an updated model structure; The updated model structure is verified and tested to obtain a model performance evaluation report, and model integration is performed based on the model performance evaluation report to obtain a target decision model.

8. A fire alarm linkage system applied to a smart lock, characterized in that: The system is used to execute the fire alarm linkage method applied to a smart lock according to any one of claims 1 to 7, and comprises: An acquisition module is used to perform communication pairing and data exchange between the smart lock and the fire alarm system, obtain a device association topology diagram, and obtain local sensor data of the smart lock and fire monitoring data of the fire alarm system; A fusion module, used for performing multi-source fusion and spatiotemporal correlation analysis on the local sensor data and the fire monitoring data based on the device association topology diagram to obtain a comprehensive fire situation assessment result; A construction module is used to perform deep reinforcement learning and situational awareness processing on the comprehensive fire situation assessment results, and to construct an adaptive decision model and a differentiated response strategy; An execution module, used to perform task decomposition and collaborative execution on the adaptive decision model and the differentiated response strategy, obtain a smart lock linkage control instruction set, and obtain an execution effect evaluation index of the smart lock linkage control instruction set; An analysis module is used to perform real-time monitoring and edge computing analysis on the execution effect evaluation index and environmental parameter change data to obtain a target feedback data stream; An updating module is used to perform knowledge graph reasoning on the target feedback data stream to obtain decision model update parameters, and to update the adaptive decision model based on the decision model update parameters to obtain a target decision model.

9. A fire alarm linkage device applied to a smart lock, characterized in that: The fire alarm linkage device applied to the smart lock comprises: a memory and at least one processor, wherein the memory stores instructions; The at least one processor calls the instructions in the memory so that the fire alarm linkage device applied to the smart lock executes the fire alarm linkage method applied to the smart lock as described in any one of claims 1-7.

10. A computer-readable storage medium having instructions stored thereon, characterized in that: When the instruction is executed by the processor, a fire alarm linkage method applied to a smart lock is implemented as described in any one of claims 1 to 7.

Citation Information

Patent Citations

  • Intelligent emergency monitoring system applied to laboratory

    CN111667388A

  • Intelligent elderly emergency rescue system and method for various sudden diseases

    CN117409536A

  • Intelligent fire safety monitoring system

    CN117649130A

  • Digital twin fire control room graphic display system and method

    CN117910673A

  • Intelligent door control security and protection Internet of Things system for outdoor power ring main unit

    CN118230473A

Cited By

  • Intelligent fire and wind prevention linkage method and system based on port quay crane meteorological perception

    CN119784191A

  • Multi-modal industrial operation and maintenance system and method fusing hierarchical large model architecture and adaptive reinforcement learning, and application of multi-modal industrial operation and maintenance system and method

    CN120355134A

  • Electromechanical equipment management method and system based on Internet of Things

    CN120509870A

  • Mechanical and electrical equipment management method and system based on internet of things

    CN120509870B

  • Building fire intelligent detection and positioning method based on multi-source data fusion

    CN121434641A