Distributed fire-fighting linkage control method and system based on edge calculation

Abstract

The invention discloses a distributed fire-fighting linkage control method and system based on edge computing, and the method comprises the steps: generating a preliminary action intention based on local sensor data through a first edge node, broadcasting the preliminary action intention to other edge nodes, carrying out the conflict detection of the intention of the second edge node and the received intention through a second edge node receiving the broadcast, and carrying out the detection of the conflict between the intention of the second edge node and the received intention; and in the negotiation process, the edge node performs dynamic utility evaluation on each action intention by using the shared context data to generate a utility score card, and the score card further introduces a probability risk evaluation mechanism to overcome the defect that the traditional deterministic model ignores the tail risk, and finally, the score card can be used as a new score card. And the second edge node generates and executes a final decision action based on the utility score card, thereby realizing intelligent and cooperative control of the fire. Through autonomous conflict detection and negotiation between edge nodes, the system can flexibly cope with local fire behavior changes, instruction conflicts are avoided, and a global optimal linkage strategy is generated.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of intelligent control, and more specifically, to a distributed fire linkage control method and system based on edge computing. BACKGROUND

[0002] As a highly dynamic and complex disaster, the key parameters of fire, such as spreading path, speed, smoke diffusion direction and concentration, can change significantly in a very short time. The traditional fire linkage control system often relies on static preset rules or centralized decision mechanism, and it is difficult to respond to the changing situation of the fire scene in real time. This lag may lead to failure of evacuation path planning, delay of fire equipment linkage or conflict of instructions, thereby seriously threatening the safety of life and property. The distributed architecture based on edge computing, by deploying data processing and decision-making capabilities on edge nodes close to data sources, is expected to solve the problems of response delay and information island in traditional solutions, thereby providing more flexible and efficient protection for fire protection.

[0003] However, although the industry has begun to explore distributed fire linkage control, existing solutions still face many technical challenges. First, the high dynamicity and uncertainty of the fire environment make any path planning and strategy making based on static information quickly outdated or even wrong; second, the complexity of the fire environment makes the entire building model a huge graph structure, and the calculation of the optimal evacuation path or linkage strategy is extremely frequent and complex, if all computing tasks are concentrated in the cloud, it will face huge concurrent computing pressure and network delay, forming a centralized bottleneck; in addition, there is a lack of effective collaborative cognition mechanism in existing distributed systems, and a single edge node can only perceive the local situation within its coverage, and cannot realize global optimal safety path planning across regions; more critically, the existing linkage strategy evaluation mechanism has a fundamental flaw, it relies on the deterministic prediction vector of the simplified physical model, ignoring the inherent uncertainty of the fire scene. This risk-neutral evaluation framework cannot effectively identify and quantify the tail risk events with small probability but extremely serious consequences, leading the system to choose a strategy that appears to have good average effect but actually has catastrophic hidden dangers, which poses a major safety hazard in the fire linkage control system which prioritizes safety.

[0004] Therefore, an optimized distributed fire linkage control method based on edge computing is expected. SUMMARY

[0005] To solve the above technical problems, the present application is proposed. Embodiments of the present application provide a distributed fire-fighting linkage control method and system based on edge computing, which generates a preliminary action intention based on local sensor data by a first edge node, and broadcasts it to other edge nodes. A second edge node receiving the broadcast will perform conflict detection on its own intention and the received intention, and initiate a negotiation session when a conflict is detected. In the negotiation process, the edge nodes will use shared context data to dynamically evaluate the utility of each action intention, generate a utility scorecard, which will further introduce a probabilistic risk assessment mechanism to overcome the defect of traditional deterministic models that ignore tail risks. Finally, the second edge node will generate and execute the final resolution action based on the utility scorecard, achieving intelligent and collaborative control of the fire. Through autonomous conflict detection and negotiation between edge nodes, the system can flexibly respond to local fire changes, avoid command conflicts, and generate a globally optimal linkage strategy.

[0006] According to one aspect of the present application, a distributed fire-fighting linkage control method based on edge computing is provided, which comprises:

[0007] The first edge node processes the obtained local sensor data based on a pre-defined rule base to obtain a first action intention;

[0008] The first edge node broadcasts the first action intention to a second edge node;

[0009] After receiving the first action intention, the second edge node performs conflict detection on the action intention of the node and the first action intention and initiates a negotiation session;

[0010] The second edge node performs dynamic utility evaluation on the first action intention and the action intention of the node in the negotiation session based on shared context data to obtain a utility scorecard;

[0011] The second edge node generates and executes the final resolution action based on the utility scorecard.

[0012] According to another aspect of the present application, a distributed fire-fighting linkage control system based on edge computing is provided, which comprises:

[0013] A data processing module for the first edge node to process the obtained local sensor data based on a pre-defined rule base to obtain a first action intention;

[0014] A broadcast module for the first edge node to broadcast the first action intention to a second edge node;

[0015] A conflict detection module for the second edge node to perform conflict detection on the action intention of the node and the first action intention after receiving the first action intention and initiate a negotiation session;

[0016] a dynamic utility evaluation module, configured to perform dynamic utility evaluation on the first action intention in the negotiation session and the action intention of the second edge node based on the shared context data to obtain an utility scorecard;

[0017] a final resolution action module, configured to generate and execute a final resolution action based on the utility scorecard.

[0018] Compared with the prior art, the application provides a distributed fire linkage control method and system based on edge computing, which generates a preliminary action intention based on local sensor data by a first edge node, and broadcasts the preliminary action intention to other edge nodes, a second edge node receiving the broadcast performs conflict detection on the received intention and the intention of the second edge node, and initiates a negotiation session when a conflict is detected, in the negotiation process, the edge nodes perform dynamic utility evaluation on the action intentions by using shared context data to generate an utility scorecard, the scorecard further introduces a probabilistic risk evaluation mechanism to overcome the defect of traditional deterministic models that ignore tail risks, finally, the second edge node generates and executes a final resolution action based on the utility scorecard, realizing intelligent and collaborative control of the fire. Through autonomous conflict detection and negotiation between edge nodes, the system can flexibly respond to local fire changes, avoid command conflicts, and generate a globally optimal linkage strategy. BRIEF DESCRIPTION OF DRAWINGS

[0019] The above and other objects, features and advantages of the present application will become more apparent from the following detailed description of embodiments of the present application taken in conjunction with the accompanying drawings. The drawings provided in the disclosure are used to provide further understanding of the embodiments of the present application, and constitute a part of the specification, and are used to explain the present application together with the embodiments of the present application, and do not constitute a limitation of the present application. In the drawings, the same reference numerals generally represent the same components or steps.

[0020] Figure 1 a flowchart of the distributed fire linkage control method based on edge computing according to the embodiments of the present application;

[0021] Figure 2 a data flow schematic diagram of the distributed fire linkage control method based on edge computing according to the embodiments of the present application;

[0022] Figure 3 a block diagram of the distributed fire linkage control system based on edge computing according to the embodiments of the present application. DETAILED DESCRIPTION

[0023] Hereinafter, example embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the example embodiments described herein.

[0024] As shown in the present application and claims, unless the context clearly indicates otherwise, the words "a," "an," "the," and / or "this" are not limited in scope to the singular, but rather include the plural. Generally, the terms "comprises" and "comprising" are not used in an exclusive sense, but rather are used in an inclusive sense to indicate that the steps and elements identified are included, but other steps or elements are not necessarily excluded.

[0025] Although the present application makes various references to certain modules in the system according to embodiments of the present application, however, any number of different modules can be used and run on the user terminal and / or server. The modules are merely illustrative, and different aspects of the system and method can use different modules.

[0026] Flowcharts are used in the present application to illustrate the operations performed by the system according to embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed in sequence. Instead, various steps can be processed in reverse order or simultaneously, as desired. Other operations can also be added to or removed from these processes.

[0027] In the following, example embodiments according to the present application will be described in detail with reference to the accompanying drawings. It is obvious that the described embodiments are only a part of the embodiments of the present application, and the present application is not limited to the example embodiments described herein.

[0028] In the technical solutions of the present application, a distributed fire linkage control method based on edge computing is proposed. Figure 1 A flowchart of the distributed fire linkage control method based on edge computing according to embodiments of the present application. Figure 2 A system architecture diagram of the distributed fire linkage control method based on edge computing according to embodiments of the present application. As shown in Figure 1 and Figure 2 The distributed fire linkage control method based on edge computing according to embodiments of the present application includes the following steps: S1, a first edge node processes the obtained local sensor data based on a predefined rule base to obtain a first action intention; S2, the first edge node broadcasts the first action intention to a second edge node; S3, after receiving the first action intention, the second edge node performs conflict detection on the action intention of the node and the first action intention and initiates a negotiation session; S4, the second edge node performs dynamic utility evaluation on the first action intention and the action intention of the node in the negotiation session based on shared context data to obtain a utility scorecard; S5, the second edge node generates and executes a final resolution action based on the utility scorecard.

[0029] In particular, the S1, the first edge node, processes the acquired local sensor data based on a predefined rule base to obtain a first action intention. It should be understood that the spread path, speed, and smoke diffusion direction and concentration of a fire can change significantly in a very short time, and a traditional system that relies on static preset rules or centralized decision mechanisms often cannot respond to such rapidly changing fire conditions in a timely and effective manner. By sinking decision intelligence to the first edge node at the fire scene, the local sensor data can be processed in real time, so that when a local abnormality occurs at the initial stage of a fire, a preliminary response strategy, i.e., a first action intention, can be quickly formed, providing a timely and necessary decision basis for subsequent distributed coordination, thereby effectively avoiding response delays caused by information transmission delays and centralized bottlenecks, and maximizing valuable emergency response time.

[0030] Here, the action intention includes an intention ID, a source node ID, a target device ID, a proposed action, an action reason, an impact area, and a primary priority score. The intention ID is a unique identifier of the action intention; the source node ID indicates the identity of the edge node that issued the intention, i.e., the first edge node here; the target device ID specifies the fire-fighting device or system to which the action intention is directed, such as a smoke exhaust fan, a sprinkler head, an evacuation sign, etc.; the proposed action is the specific operation that the intention suggests to perform, such as "start", "stop", "start", "guide", etc.; the action reason explains the reason why the action is proposed, such as "high temperature detected", "smoke exceeds standard", or "evacuation passage blocked"; the impact area defines the geographical scope of the action intention; and the primary priority score reflects the degree of urgency and importance of the action intention at the beginning, providing a preliminary weight basis for subsequent conflict detection and negotiation. In addition, the first edge node refers to a local device or server with certain computing, storage, and communication capabilities, deployed near the sensor data source at the fire scene, responsible for local data processing and preliminary decision-making; and the local sensor data refers to various types of real-time environmental parameters related to the area governed by the first edge node, collected directly by the first edge node.

[0031] In practice, the first edge node continuously acquires local sensor data within its jurisdiction. These sensor data can cover various real-time physical parameters of the fire environment, such as temperature, smoke concentration, flame detection information, and personnel location, etc. After acquiring these data, the first edge node efficiently processes and analyzes the real-time data based on an internally stored predefined rule base. The predefined rule base is a set of logic that contains fire safety expert knowledge, industry standards, and emergency plans, guiding the edge node in interpreting sensor data and generating action intentions. It contains a series of pre-set logical judgments and condition-action mapping rules to associate specific sensor data patterns with corresponding fire linkage actions. For example, when the temperature sensor reading exceeds a certain threshold and the smoke concentration sensor reading also rises, the rule base may trigger an action intention to start the local smoke exhaust system or instruct personnel evacuation. This processing process is not simply data aggregation, but intelligent interpretation of data, aiming to extract decision-making information from massive raw data and infer the best preliminary response strategy for the current local fire situation based on pre-set expert experience or logic. This process emphasizes the autonomy and immediacy of the edge node, enabling it to independently complete preliminary situation awareness and action planning without relying on the central cloud.

[0032] In particular, the S2, the first edge node broadcasts the first action intention to the second edge node. It should be understood that traditional fire linkage control systems often rely on static pre-set rules or centralized decision-making mechanisms, making it difficult to respond to rapidly changing fire conditions in real time. This centralized mode is prone to delay in information transmission, and once the central node fails, the decision-making ability of the entire system will be severely affected, leading to response lag or system paralysis. By broadcasting the first action intention to other edge nodes (i.e., the second edge node), information can be decentralized and delivered in parallel, ensuring rapid spread of fire information and preliminary decisions in the distributed network, thereby laying the foundation for subsequent conflict detection, negotiation sessions, and coordinated actions, greatly improving the real-time response capability and overall robustness of the system in highly dynamic fire scenarios.

[0033] Among them, the second edge node refers to other edge nodes that receive the action intention broadcast by the first edge node. These nodes also have the ability to independently process and make decisions, and will conduct subsequent conflict detection and negotiation based on the received information combined with their own situation.

[0034] In a specific implementation, the first edge node encapsulates a message containing all the details of the action intention and sends it through the network in a peer-to-peer (P2P) manner or using an efficient distributed synchronization protocol such as the Gossip protocol. Upon receiving the message, the second edge node updates its knowledge of the current fire situation and the intentions of neighboring nodes based on the message. This broadcast mechanism ensures rapid dissemination of information, and since there is no single master node burdened with all communication tasks, even if some nodes fail, other nodes can continue to calculate and coordinate based on existing information, thereby enhancing the reliability and fault tolerance of the entire distributed system.

[0035] In particular, the S3, upon receiving the first action intention, performs conflict detection on the action intention of the node and the first action intention and initiates a negotiation session. It should be understood that in a multi-node, decentralized distributed architecture, different edge nodes may independently generate action intentions based on their local perception information. Without coordination, these intentions may result in issuing opposite instructions to the same fire-fighting equipment, for example, one edge node decides to open the A air valve to exhaust smoke, while the adjacent node decides to close the A air valve to prevent the spread of fire. If not addressed in time, this potential conflict will seriously interfere with the effectiveness of the fire-fighting operation, and even exacerbate the fire or hinder personnel evacuation. Therefore, by actively identifying and resolving these strategy conflicts through the conflict detection mechanism, the timing consistency and overall safety of the fire-fighting control behavior can be guaranteed, and high-risk strategies due to model prediction bias can be avoided, thereby maximizing the safety of life and property.

[0036] In a specific implementation, when the second edge node successfully receives the first action intent broadcasted from the first edge node, it does not immediately adopt or execute the intent. Instead, the second edge node first compares its own action intent generated based on local sensor data and a predefined rule base; then, the second edge node performs a detailed conflict detection on the two action intents. In one specific example of the present application, a conflict is determined to occur and a negotiation session is initiated in response to the target device ID being the same but the proposed action being opposite in the action intent of the second edge node and the first action intent. This means that if the second edge node finds that its own intent and the first edge node's intent are both targeting the same target device (i.e., the target device ID is the same), but the proposed actions are opposite or contradictory (e.g., one proposes "on" and the other proposes "off"), the system determines that a conflict has occurred. Once such a conflict is detected, the second edge node initiates a negotiation session. Here, the negotiation session is an interactive process initiated by the second edge node after detecting a conflict. The initiation of the negotiation session marks the entry of two or more relevant edge nodes into a collaborative decision-making mode, aiming to find a solution that can take into account the needs of all parties and achieve global optimization through information sharing, utility evaluation, etc., in order to eliminate conflicts and reach a consensus on the final resolution action. This process ensures that in a distributed decision-making environment, collaboration between nodes is orderly and intelligent, rather than simply independent execution.

[0037] Here, the action intent of the second edge node is an action suggestion independently generated by the second edge node based on its own local sensor data and a predefined rule base, and its structure is the same as that of the first action intent, including intent ID, source node ID, target device ID, proposed action, action reason, impact area, and primary priority score.

[0038] In particular, the S4, the second edge node, performs a dynamic utility assessment on the first action intention in the negotiation session and the action intention of the node based on shared context data to obtain a utility scorecard. The shared context data includes a fire stage and a personnel evacuation state. It should be understood that the existing fire linkage strategy evaluation mechanism excessively relies on a deterministic prediction vector derived from a simplified physical model, thereby completely ignoring the inherent uncertainty inherent in the highly complex and chaotic physical scenario of fire. This risk-neutral evaluation framework cannot effectively identify and measure extreme risk events with small probability but extremely serious consequences (i.e., "tail risk"), resulting in a system that may choose a strategy that appears to have good average effects but actually contains catastrophic possibilities. Therefore, in the technical solution of the present application, by introducing a dynamic utility assessment mechanism and considering the probability of risk, the decision evaluation is upgraded from a deterministic model to a probabilistic model, thereby realizing the quantification and avoidance of risks and ensuring that the system can make more robust and safe final decisions when facing strategy conflicts, thereby maximizing the safety of life and property.

[0039] In a specific implementation, first, a dynamic weight factor is obtained by accessing a weight mapping table based on shared context data. This means that the second edge node will dynamically adjust the weights of different evaluation indicators according to the global context information such as the current fire stage and personnel evacuation state, to adapt to the changing needs of the fire scene. For example, in the early stage of the fire, more emphasis may be placed on controlling the spread of the fire; while in the peak period of personnel evacuation, more emphasis may be placed on evacuation efficiency and safety;

[0040] Next, the proposed action in the action intention of the node is input into a predictive state inference engine based on a physical model to obtain a predicted state vector, which includes an average visibility prediction, a visibility prediction standard deviation, an average temperature rise prediction, and a temperature rise prediction standard deviation. By introducing a predictive state inference engine based on a physical model and making its output include a prediction state vector with uncertainty metrics, the decision evaluation is upgraded from a deterministic model to a probabilistic model, thereby realizing the quantification and avoidance of risks and providing a solid foundation for subsequent utility integration and safety decisions.

[0041] The engine is a complex computational model that integrates building structure information, fire source location, combustible material characteristics, ventilation conditions, and fire-fighting equipment actions, and other physical parameters and logic. After receiving the proposed action, the engine simulates the physical effects and evolution trends that the action may cause in the current fire scene; unlike traditional deterministic models, the predictive state inference engine can capture the uncertainty factors in the fire process, such as slight fluctuations in smoke diffusion speed, unevenness of temperature gradient, etc. The output prediction state vector describes the future state of key fire parameters and their uncertainty in detail. Specifically, the prediction state vector will include the predicted mean of visibility, the predicted standard deviation of visibility, the predicted mean of temperature rise, and the predicted standard deviation of temperature rise. These means and standard deviations form the basis of a complete probabilistic description of the future visibility and temperature rise, two key state quantities, thereby providing the necessary data support for subsequent construction of state probability density functions, calculation of expected benefits, and quantification of risk costs;

[0042] Further, the prediction state vector is integrated based on probabilistic risk to obtain the utility scorecard. It should be understood that the existing utility function calculation and scorecard generation mechanism has a fundamental technical defect in quantitatively evaluating the linkage strategy. This mechanism relies on a deterministic prediction vector derived from a simplified physical model, i.e., a future state description consisting of a single numerical value. This processing method implicitly assumes absolute accuracy of the prediction, completely ignoring the inherent uncertainty that any prediction model must inevitably have in the highly complex and chaotic physical scenario of a fire. As a result, the evaluation framework of this mechanism is risk-neutral, and cannot effectively identify and measure extreme risk events with small probability but extremely serious consequences (i.e., "tail risk"). It can evaluate the expected effect of a strategy, but cannot distinguish between a strategy with stable and controllable results and another strategy with huge fluctuations and catastrophic possibilities, which is a major safety hazard in a fire-fighting linkage control system that prioritizes safety. To compensate for the above-mentioned defects, in the preferred example of the present application, by introducing a utility integration framework based on probabilistic risk, the decision evaluation is upgraded from a deterministic model to a probabilistic model, thereby realizing the quantification and avoidance of risk and fundamentally improving the intelligence level and decision safety of the distributed fire-fighting linkage control system.

[0043] In this process, first, based on the prediction state vector, a state probability density function is constructed. It should be understood that the uncertainty of the prediction must be expressed in mathematical form to make it a calculable entity. Accordingly, the prediction results output by the prediction model, including the prediction mean of future visibility and temperature rise ( , ) and its prediction standard deviation ( , ). Based on these parameters, both the future visibility and temperature rise, two key state variables, are constructed as continuous random variables subject to normal distribution. From this, the probability density functions and

[0044] The probability density function of visibility is defined as:

[0045]

[0046] where is the visibility prediction mean, and is the visibility prediction standard deviation

[0047] The probability density function of temperature rise is defined as:

[0048]

[0049] where is the temperature rise prediction standard deviation and is the temperature rise prediction mean.

[0050] In this way, the original single-point prediction value can be transformed into a mathematical function that can fully describe all possible outcomes and their corresponding probabilities, to output two probability density functions that constitute a complete probabilistic description of the future state;

[0051] Next, the state probability density functions are quantified for the dispersion benefits and the spread costs based on the basic utility function to obtain the expected dispersion benefits and the risk-adjusted spread costs. It should be understood that two key indicators that can represent the overall pros and cons of the strategy need to be extracted from the probability distribution: one reflects its average case benefits, and the other measures its potential loss in the worst case. In the implementation process, first, by integrating the basic benefit function over the entire probability distribution of visibility, the expected dispersion benefit is calculated. This is equivalent to a probability-weighted sum of all possible benefit outcomes. This process is represented by the formula:

[0052]

[0053] where is the basic utility function of the dispersion benefit, which maps a specific visibility value to a normalized benefit score;

[0054] ​​​Secondly, the contagion cost is quantified by adopting the Conditional Value at Risk (CVaR) method in the field of financial engineering. CVaR aims to measure the conditional expectation of the part of cost beyond the Value at Risk (VaR) threshold with a high confidence level (e.g. 95%); the process is formulated as:

[0055]

[0056] where, is the base utility function of the contagion cost, which maps a specific temperature rise value to a normalized cost score;

[0057]

[0058] Further, the expected evacuation benefit and the risk-adjusted contagion cost are probabilistically utility aggregated based on dynamic weight factors to obtain the utility scorecard. It should be understood that the expected benefit and the risk cost calculated in the previous step need to be weighed according to the current macro situation of the fire scene, and finally fused into a single, comparable comprehensive utility score. Specifically, the expected benefit and the risk-adjusted cost are weighted and aggregated by the weight factors and generated dynamically by global context information such as fire stage and evacuation state, so as to calculate the final probabilistic utility score ; the process is formulated as:

[0059]

[0060] where, and are dynamic weight factors, and are the expected evacuation benefit and the risk-adjusted contagion cost, respectively.

[0061] ​​​​​​In this way, a final score that can reflect both the average performance and risk aversion characteristics of each linkage strategy to be evaluated can be generated. The output probability utility score U_prob endogenously contains the penalty for risk. A strategy with an average return is considerable but there is high tail risk, its C_CVaR value will be large, resulting in its final U_prob score being significantly lowered, so as to be at a disadvantage when compared with other more robust strategies, ensuring that the finally selected strategy has higher safety and reliability.

[0062] In summary, the preferred technical solution successfully incorporates the predicted uncertainty into the utility evaluation framework by introducing mathematical tools such as probability density function, expected integral and conditional value at risk (CVaR), transforming the original deterministic, risk-neutral evaluation model into a probabilistic, risk-averse decision model. This model not only evaluates the average performance of the strategy, but also accurately quantifies and amplifies its potential catastrophic risk. In this way, the intelligent level and decision safety of the distributed fire linkage control system are fundamentally improved. By giving the system the ability to see and avoid the worst-case scenario at the algorithm level, it can ensure that the system's choices are not only theoretically optimal, but also robust and safe in practice when faced with strategy conflicts. This can effectively prevent the adoption of high-risk strategies due to model prediction bias, avoid catastrophic consequences that may lead to uncontrolled fires or evacuation failures, and thus maximize the safety of life and property.

[0063] In particular, the S5, the second edge node generates and executes the final resolution action based on the utility scorecard. That is, the complex evaluation results are converted into specific on-site interventions, thereby completely solving the decision-making dilemma caused by strategy conflicts and uncertainty risks in a distributed environment. In the technical solution of the present application, the final action is generated and executed based on the utility scorecard that comprehensively considers risks and benefits, which can ensure that in the highly dynamic and high-risk scenario of fire, the decisions made by the system are not only theoretically optimal, but also robust and safe in practice, effectively preventing the adoption of high-risk strategies due to model prediction bias, avoiding catastrophic consequences that may lead to uncontrolled fires or evacuation failures, and thus maximizing the safety of life and property.

[0064] In a specific implementation, after the second edge node completes dynamic utility evaluation on each action intention in the negotiation session, and generates a corresponding utility scorecard for each intention, the system will compare and analyze these scorecards. Since the utility scorecard inherently contains a penalty for risk, a strategy with an average considerable return but high tail risk will have its scorecard score significantly lowered. Therefore, the second edge node will select the action intention with the highest or optimal utility scorecard, and determine it as the final resolution action. This selection process ensures that the system can identify the best strategy that not only brings the maximum expected return, but also effectively avoids significant risks, among multiple possible action plans. Once the final resolution action is determined, the second edge node will immediately convert it into specific instructions and send them to the corresponding fire linkage equipment or system, and execute the action. For example, if the final resolution action is to "turn on the smoke exhaust fan in a certain area", the second edge node will issue a start command to the control module of the fan; if it involves personnel evacuation, it may send a path update instruction to the intelligent evacuation indication system to guide personnel to avoid dangerous areas. The entire execution process is completed locally on the edge node, minimizing the time from decision to execution, achieving rapid and precise fire emergency response, and embodying the advantages of distributed edge computing.

[0065] wherein the final resolution action refers to the only certain action selected by the second edge node as optimal and about to be implemented among multiple action intentions that have undergone utility evaluation. This action is generated after conflict detection, negotiation session, and probabilistic risk assessment, aiming to achieve global optimality and highest safety under the current fire situation.

[0066] In summary, the distributed fire linkage control method based on edge computing according to the embodiments of the present application is illustrated, which generates preliminary action intentions based on local sensor data by the first edge node, and broadcasts them to other edge nodes. The second edge node receiving the broadcast will perform conflict detection on its own intentions and the received intentions, and initiate a negotiation session when a conflict is detected. During the negotiation process, the edge nodes will use shared context data to dynamically evaluate the utility of each action intention, generating a utility scorecard. This scorecard will further introduce a probabilistic risk assessment mechanism to overcome the shortcomings of traditional deterministic models that ignore tail risks. Finally, the second edge node will generate and execute the final resolution action based on this utility scorecard, achieving intelligent and collaborative control of the fire. Through autonomous conflict detection and negotiation among edge nodes, the system can flexibly respond to local fire changes, avoid instruction conflicts, and generate a globally optimal linkage strategy.

[0067] Further, a distributed fire linkage control system based on edge computing is also provided.

[0068] Figure 3A block diagram of a distributed fire linkage control system based on edge computing according to an embodiment of the present application. As shown in Figure 3 The distributed fire linkage control system based on edge computing 300 according to an embodiment of the present application comprises: a data processing module 310, configured to process local sensor data obtained by a first edge node based on a predefined rule base to obtain a first action intention; a broadcast module 320, configured to broadcast the first action intention by the first edge node to a second edge node; a conflict detection module 330, configured to, after receiving the first action intention, perform conflict detection on an action intention of the node and the first action intention by the second edge node and initiate a negotiation session; a dynamic utility evaluation module 340, configured to perform dynamic utility evaluation on the first action intention in the negotiation session and the action intention of the node based on shared context data by the second edge node to obtain a utility scorecard; and a final resolution action module 350, configured to generate and execute a final resolution action based on the utility scorecard by the second edge node.

[0069] As described above, the distributed fire linkage control system based on edge computing 300 according to an embodiment of the present application can be implemented in various wireless terminals, such as a server with a distributed fire linkage control algorithm based on edge computing, and the like. In one possible implementation, the distributed fire linkage control system based on edge computing 300 according to an embodiment of the present application can be integrated into a wireless terminal as a software module and / or a hardware module. For example, the distributed fire linkage control system based on edge computing 300 can be a software module in the operating system of the wireless terminal, or can be an application program developed for the wireless terminal; of course, the distributed fire linkage control system based on edge computing 300 can also be one of the many hardware modules of the wireless terminal.

[0070] Alternatively, in another example, the distributed fire linkage control system based on edge computing 300 and the wireless terminal can also be separate devices, and the distributed fire linkage control system based on edge computing 300 can be connected to the wireless terminal through a wired and / or wireless network, and transmit interactive information in an agreed data format.

[0071] The above has described embodiments of the present disclosure, and the above description is exemplary, not exhaustive, and is not limited to the disclosed embodiments. Many modifications and changes are obvious to those skilled in the art without departing from the scope and spirit of the described embodiments. The choice of terms used herein is intended to best explain the principles, practical applications, or improvements to the technology in the market of the embodiments, or to enable other ordinary skilled persons in the art to understand the embodiments disclosed herein.

Claims

1. An edge-computing-based distributed fire linkage control method, characterized in that, The method comprises: a first edge node processes acquired local sensor data based on a predefined rule base to obtain a first action intent; the first edge node broadcasts the first action intent to a second edge node; after receiving the first action intent, the second edge node performs conflict detection on an action intent of the node and the first action intent and initiates a negotiation session; the second edge node performs dynamic utility evaluation on the first action intent and the action intent of the node in the negotiation session based on shared context data to obtain a utility scorecard; the second edge node generates and executes a final resolution action based on the utility scorecard. 2.The edge computing based distributed fire linkage control method according to claim 1, characterized in that, The action intent comprises an intent ID, a source node ID, a target device ID, a proposed action, an action reason, an impact area, and a primary priority score. 3.The edge computing based distributed fire linkage control method according to claim 2, characterized in that, After receiving the first action intent, the second edge node performs conflict detection on an action intent of the node and the first action intent and initiates a negotiation session, comprising: in response to the target device ID being the same but the proposed action being opposite in the action intent of the node and the first action intent, determining that a conflict occurs and initiating the negotiation session. 4.The edge computing based distributed fire linkage control method according to claim 1, characterized in that, The shared context data comprises a fire stage and a personnel evacuation state. 5.The edge computing based distributed fire linkage control method according to claim 4, characterized in that, The second edge node performs dynamic utility evaluation on the first action intent and the action intent of the node in the negotiation session based on shared context data to obtain a utility scorecard, comprising: accessing a weight mapping table based on the shared context data to obtain a dynamic weight factor; inputting the proposed action in the action intent of the node into a predictive state inference engine based on a physical model to obtain a predicted state vector, the predicted state vector comprising a visibility prediction mean, a visibility prediction standard deviation, a temperature rise prediction mean, and a temperature rise prediction standard deviation; performing probability risk-based utility integration on the predicted state vector to obtain the utility scorecard. 6.The edge computing based distributed fire linkage control method according to claim 5, characterized in that, Performing probability risk-based utility integration on the predicted state vector to obtain the utility scorecard comprises: constructing a state probability density function based on the predicted state vector; quantifying evacuation benefits and spread costs of the state probability density function based on a basic utility function to obtain an expected evacuation benefit and a risk-adjusted spread cost; performing probabilistic utility aggregation on the expected evacuation benefit and the risk-adjusted spread cost based on the dynamic weight factor to obtain the utility scorecard. 7.The edge computing based distributed fire linkage control method according to claim 6, characterized in that, The state probability density function is expressed as: wherein is the visibility forecast mean, is the visibility forecast standard deviation, is the temperature rise forecast standard deviation and is the temperature rise forecast mean. 8.The edge computing based distributed fire linkage control method according to claim 6, characterized in that, Performing probabilistic utility aggregation on the expected evacuation benefit and the risk-adjusted spread cost based on the dynamic weight factor to obtain the utility scorecard comprises: performing probabilistic utility aggregation on the expected evacuation benefit and the risk-adjusted spread cost according to the following formula: wherein, and is a dynamic weight factor, and are the expected evacuation benefit and the risk-adjusted spread cost, respectively.

9. An edge-computing-based distributed fire linkage control system, characterized by, The method comprises: a data processing module for a first edge node to process acquired local sensor data based on a predefined rule base to obtain a first action intent; a broadcast module for the first edge node to broadcast the first action intent to a second edge node; a conflict detection module for the second edge node to perform conflict detection on an action intent of the node and the first action intent after receiving the first action intent and initiate a negotiation session; a utility scorecard generation module for the second edge node to perform dynamic utility evaluation on the first action intent and the action intent of the node in the negotiation session based on shared context data to obtain a utility scorecard; and a final resolution action generation module for the second edge node to generate and execute a final resolution action based on the utility scorecard. a dynamic utility evaluation module, configured to perform dynamic utility evaluation on the first action intention in the negotiation session and the action intention of the second edge node based on the shared context data to obtain an utility scorecard; a final resolution action module, configured to generate and execute a final resolution action based on the utility scorecard.

Images