A method, apparatus and device for data analysis of a combustible gas sensor

Abstract

The application discloses a combustible gas sensor data analysis method, device and equipment and a readable storage medium, and relates to the technical field of data analysis. The combustible gas sensor data analysis method comprises the following steps: establishing a space-time heterogeneous causal graph skeleton in a target building, and analyzing the gas condition in the target building according to the combustible gas sensor data, so that the gas condition in the building can be adjusted according to real-time conditions, and the safety of gas use is ensured.

Description

Technical Field

[0001] This invention relates to the field of data analysis technology, and more specifically to a data analysis method, apparatus, device, and readable storage medium for a combustible gas sensor. Background Technology

[0002] With the increasing complexity of building gas usage and air conditioning / ventilation systems, safety monitoring and response face the challenge of "multi-room, low-concentration, cross-channel propagation." Traditional methods relying on single-point threshold alarms and manual verification are insufficient to address the hidden spread along pipelines and airflows. Dynamic changes in doors, windows, occupancy, and HVAC (heating, ventilation, and air conditioning) operating conditions further increase decision-making uncertainty. When multiple rooms simultaneously exhibit low-concentration anomalies but do not reach the single-point threshold, determining "when, where, and to what extent" to escalate to main valve shut-off becomes a critical challenge.

[0003] Existing technologies primarily focus on two paths: the first is threshold-based distributed monitoring and linkage control. Each combustible gas sensor sets an alarm threshold and time filter. Upon reaching the threshold, the BMS triggers exhaust, ventilation boosting, or valve closure, and linkage across devices is achieved through zone-level interlocking tables. When multiple anomalies do not reach the threshold, a weighted average or rule-based "synthetic alarm" is often used, followed by manual confirmation to determine whether to shut off the main valve. The second is risk diffusion estimation based on topology or regional models. The system maintains a schematic diagram of gas pipelines and HVAC zoning, using a fixed connectivity matrix or simple time delay to infer propagation paths, and executes triggering conditions according to a preset control sequence (exhaust first, then ventilation boosting, then local valve positions). Some studies introduce statistical fusion or graph algorithms to identify multi-point correlations, but these often approximate homogeneous node networks, failing to distinguish between "pipeline connectivity" and "airflow channels." Door and window status and occupancy restrictions are usually treated as additional indicators, making it difficult to translate into executable temporal constraints. Main valve upgrades and shutdowns often rely on high-concentration alarms, manual approval, or scenario lists, making it difficult to form credible upgrade decisions in the early stages of low-concentration, cross-room, and directional diffusion. The control side often uses fixed timing or simple interlocks, lacking unified optimization with propagation upper bound and prohibition conditions; even when MPC is used, it is mostly aimed at energy consumption or air volume balance, without incorporating "preventing back diffusion" and "protecting high occupancy areas" as hard constraints into sequence generation.

[0004] Therefore, there is an urgent need for a data analysis method for combustible gas sensors that can overcome the above-mentioned defects in order to increase the safety of gas use. Summary of the Invention

[0005] The purpose of this invention is to provide a data analysis method, apparatus, device, and readable storage medium for combustible gas sensors. By establishing a spatiotemporal heterogeneous causal graph skeleton within a target building, the gas situation within the target building is analyzed based on the data from the combustible gas sensor. This allows for adjustments to the gas situation within the building based on real-time conditions, ensuring the safety of gas usage.

[0006] To achieve the above objectives, the present invention provides the following technical solution:

[0007] In a first aspect, the present invention provides a data analysis method for a combustible gas sensor, the method comprising:

[0008] Based on the gas pipeline topology, air conditioning and ventilation zones, door and window status, and door and window occupancy areas of the target building, a spatiotemporal heterogeneous causal graph skeleton is constructed.

[0009] The real-time response of the combustible gas sensor is embedded into the spatiotemporal heterogeneous causal graph skeleton as the node state trajectory, and causal evidence is generated on the pipeline connection edge and the airflow channel edge based on the consistency of response sequence and direction, thus obtaining a set of candidate causal fragments.

[0010] Candidate causal segments are pieced together segment by segment according to the consistency of the gas pipeline and the direction of the dominant airflow, and segments inconsistent with cleaning activities or local ventilation are excluded to obtain the upgrade permit conclusion, the affected zone and the causal chain direction.

[0011] Based on the upgrade permit conclusion, the affected zones and the causal chain, and combined with the occupied area, a safety window is generated, and the control scope and timing requirements are determined.

[0012] An event-driven state machine is established based on the control scope and timing requirements, defining the state transition conditions for normal monitoring, preventive actions, escalation cutoff, and post-recovery recovery. At the lower level, constraint model predictive control is used to perform timing optimization on directional exhaust, zoned air enhancement, and segmented valve positions, and outputs the control sequence.

[0013] Based on the control sequence, escalation permit conclusion, and safety window, directional exhaust and zoned ventilation are performed within the safety window. The combustible gas sensor network is used to observe whether the leak-diffusion causal chain is maintained, and the maintenance conclusion and residual impact zone are output. When the maintenance conclusion is satisfied and the escalation permit conclusion remains valid, the main building gas valve is shut off within the safety window according to the control sequence, and the sectional valves are isolated with minimal impact. The aftermath recovery is completed according to the control sequence, so that the event-driven state machine returns to the normal monitoring state.

[0014] In some embodiments, a spatiotemporal heterogeneous causal graph skeleton is constructed based on the gas pipeline topology, air conditioning and ventilation zones, door and window status, and occupied areas of the target building, including:

[0015] Establish pipeline nodes and pipeline connectivity relationships based on the gas pipeline topology; pipeline nodes include pipe segment inflection points, valve locations, and equipment interface locations.

[0016] The air conditioning ventilation zones are modeled as air conditioning ventilation zone nodes, and causal edges of airflow channels are established based on the correspondence between supply air vents and return air vents and the connectivity between corridors and rooms.

[0017] Based on pipeline nodes, pipeline connectivity, air conditioning and ventilation zones, door and window status, and occupied areas, establish causal edge constraints for airflow channels;

[0018] Construct a spatiotemporal heterogeneous causal graph skeleton based on the constraint relationships.

[0019] In some embodiments, the real-time response of the combustible gas sensor is embedded into a spatiotemporal heterogeneous causal graph skeleton as a node state trajectory, and causal evidence is generated on the pipeline connection edge and the gas flow channel edge based on the consistency of response sequence and direction, resulting in a candidate causal fragment set, including:

[0020] The real-time response of each combustible gas sensor node is formed into a state sequence according to the sampling time, and the state sequence is labeled into the spatiotemporal heterogeneous causal graph skeleton according to the response stage.

[0021] For each pipeline connection causal edge, read the node state trajectory of its starting node and ending node, and make a consistency judgment based on the node state trajectory.

[0022] Based on the connectivity and consistency determination results of adjacent causal edges in the spatiotemporal heterogeneous causal graph skeleton, multiple causal evidences are spliced ​​into fragments according to direction and chronological order to form a candidate causal fragment set.

[0023] In some embodiments, candidate causal segments in the candidate causal segment set are pieced together segment by segment according to the consistency of the gas pipeline and the dominant airflow direction, and segments inconsistent with cleaning activities or local ventilation are excluded to obtain upgrade permission conclusions, affected zones, and causal chain directions, including:

[0024] According to the effective direction of the causal edge connecting the pipeline and the causal edge of the airflow channel, the directional consistency test is performed on each causal evidence, and the sequential splicing of adjacent causal evidence is verified within the propagation time bound.

[0025] If the verification passes, the candidate causal segments in the candidate causal segment set are spliced ​​together segment by segment according to the continuous relationship between the end node and the start node to form a leakage-diffusion causal chain.

[0026] Determine whether the leak-diffusion causal chain simultaneously covers multiple rooms where the combustible gas sensor node is located, and whether the causal edge along the pipeline connects from the gas supply direction to the user side or equipment interface location, and whether it crosses the supply air to return air path of the air conditioning ventilation zone on the causal edge of the airflow channel, and generate an upgrade permission conclusion based on the judgment result.

[0027] Read the set of air conditioning and ventilation zone nodes through which the leakage-diffusion causal chain passes, and mark the downwind area as the affected zone according to the direction of the causal edge of the airflow channel.

[0028] Read the directional properties of each pipeline connection causal edge and airflow channel causal edge within the leak-diffusion causal chain to obtain the direction of the causal chain.

[0029] In some embodiments, based on the upgrade license conclusion, the affected partition and the causal chain, combined with the occupied area, a security window is generated, and the control scope and timing requirements are determined, including:

[0030] Under the condition that the upgrade permission conclusion is permitted, the affected zone is read, and the downwind area within the affected zone is marked as the priority area according to the causal chain, forming the spatial part of the control domain;

[0031] Based on the causal chain direction, select the set of pipeline nodes that intersect with the affected zone along the direction of the pipeline connecting causal edge, and determine the set of segment valve positions with the least impact between adjacent segment valves to form the pipeline part of the control domain;

[0032] A safety window is generated by combining the accessible time period and the time of presence in the occupied area;

[0033] Timing requirements are determined based on the directional priority of the causal chain and the spatial and pipeline components of the control domain.

[0034] In some embodiments, an event-driven state machine is established based on the control scope and timing requirements, defining state transition conditions for normal monitoring, preventive actions, escalation cutoff, and post-incident recovery; at the lower level, constraint model predictive control is used to perform timing optimization on directional exhaust, zoned air enhancement, and segmented valve positions, outputting a control sequence, including:

[0035] The order of entering and exiting the state is determined by the sequence of actions and the triggering conditions in the timing requirements, and the normal monitoring state is set as the default state.

[0036] When the action triggering conditions in the timing requirements are met, the system will switch from the normal monitoring state to the preventive action state.

[0037] When the upgrade trigger condition in the timing requirements is met, the state changes from the preventive action state to the upgrade cutoff state.

[0038] During the upgraded shut-off state, the segmented valve position adjustment is called according to the timing requirements to prepare for the execution time of the main valve shut-off. The action arrangement after shut-off and the start time and duration of the recovery are written into the subsequent segment of the control sequence, and the control sequence is output.

[0039] Secondly, the present invention also provides a data analysis device for a combustible gas sensor, the device comprising:

[0040] The skeleton construction module is used to construct a spatiotemporal heterogeneous causal graph skeleton based on the gas pipeline topology, air conditioning and ventilation zones, door and window status, and door and window occupancy areas of the target building.

[0041] The fragment acquisition module is used to embed the real-time response of the combustible gas sensor into the spatiotemporal heterogeneous causal graph skeleton as the node state trajectory, and generate causal evidence on the pipeline connection edge and the airflow channel edge according to the consistency of response sequence and direction, so as to obtain a set of candidate causal fragments.

[0042] The conclusion acquisition module is used to piece together the candidate causal segments in the candidate causal segment set segment by segment according to the consistency of the gas pipeline and the direction dominated by the airflow, and to exclude segments that are inconsistent with the cleaning activities or local ventilation, so as to obtain the upgrade permission conclusion, the affected zone and the causal chain direction.

[0043] The module is required to generate a security window and determine the control scope and timing requirements based on the upgrade license conclusion, the affected partition and the causal chain, combined with the occupied area.

[0044] The sequence output module is used to establish an event-driven state machine based on the control scope and timing requirements, and to define the state transition conditions for normal monitoring, preventive actions, escalation cutoff, and post-recovery recovery. At the lower level, it uses constraint model predictive control to perform timing optimization on directional exhaust, zoned air enhancement, and segmented valve positions, and outputs the control sequence.

[0045] The partition output module is used to perform directional exhaust and partitioned ventilation within the safety window according to the control sequence, upgrade permission conclusion, and safety window. It also uses the combustible gas sensor network to observe whether the leak-diffusion causal chain is maintained, and outputs the maintenance conclusion and residual impact partition. When the maintenance conclusion is satisfied and the upgrade permission conclusion continues to be valid, it is used to cut off the main gas valve of the building according to the control sequence within the safety window, and implement minimum impact isolation for the sectional valves. It also completes the follow-up recovery according to the control sequence, so that the event-driven state machine returns to the normal monitoring state.

[0046] Thirdly, the present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the data analysis method for the combustible gas sensor provided in the first aspect.

[0047] Fourthly, the present invention also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the data analysis method for the combustible gas sensor provided in the first aspect.

[0048] Fifthly, the present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the data analysis method for the combustible gas sensor provided in the first aspect.

[0049] The beneficial effects of the present invention are as follows: The data analysis method of the combustible gas sensor in the present invention analyzes the gas situation in the target building based on the data of the combustible gas sensor by establishing a spatiotemporal heterogeneous causal graph skeleton in the target building, and then adjusts the gas situation in the building according to the real-time situation, thereby ensuring the safety of gas use.

[0050] The above description is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention and to implement it in accordance with the contents of the specification, the preferred embodiments of the present invention are described in detail below with reference to the accompanying drawings. Attached Figure Description

[0051] Figure 1 This is a flowchart illustrating a data analysis method for a combustible gas sensor according to an embodiment of the present invention;

[0052] Figure 2 This is a schematic diagram of the structure of a data analysis device for a combustible gas sensor according to an embodiment of the present invention;

[0053] Figure 3 This is a schematic diagram of an electronic device structure provided in an embodiment of this application. Detailed Implementation

[0054] The technical solution of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention. It should be noted that references to "an embodiment," "embodiment," "example embodiment," etc., in this specification refer to the described embodiment including specific features, structures, or characteristics; however, not every embodiment must include these specific features, structures, or characteristics. Furthermore, such expressions do not refer to the same embodiment. Moreover, when describing specific features, structures, or characteristics in conjunction with embodiments, whether or not explicitly described, it is indicated that incorporating such features, structures, or characteristics into other embodiments is within the knowledge scope of those skilled in the art.

[0055] Furthermore, the technical features involved in the different embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0056] In some embodiments, such as Figure 1 The diagram shows a flowchart of a data analysis method for a combustible gas sensor, including:

[0057] S101, based on the gas pipeline topology, air conditioning and ventilation zones, door and window status and door and window occupancy areas of the target building, construct a spatiotemporal heterogeneous causal graph skeleton.

[0058] Mapping the combustible gas sensor network into heterogeneous nodes means establishing combustible gas sensors, gas pipeline locations, air conditioning and ventilation zones, doors and windows, and occupied areas within the building as node types. Heterogeneous nodes include at least combustible gas sensor nodes, pipeline nodes, air conditioning and ventilation zone nodes, door and window nodes, and occupied area nodes. Causal edges include pipeline connection causal edges and airflow channel causal edges. The spatiotemporal heterogeneous causal graph skeleton is a graph model composed of the above heterogeneous nodes, causal edges, and associated spatiotemporal constraints, used to describe the possible transmission paths and directions of leakage in pipelines and air channels.

[0059] Optionally, pipeline nodes and pipeline connectivity can be established based on the gas pipeline topology; pipeline nodes include pipe segment inflection points, valve locations, and equipment interface locations; air conditioning ventilation zones can be modeled as air conditioning ventilation zone nodes, and causal edges of airflow channels can be established based on the correspondence between supply air vents and return air vents and the connectivity between corridors and rooms; constraint relationships of causal edges of airflow channels can be established based on pipeline nodes, pipeline connectivity, air conditioning ventilation zones, door and window status, and occupied areas; and a spatiotemporal heterogeneous causal graph skeleton can be constructed according to the constraint relationships.

[0060] Specifically, firstly, pipeline nodes and pipeline connectivity are established based on the gas pipeline topology. Pipeline nodes include pipe segment inflection points, valve locations, and equipment interface locations. Pipeline connectivity is determined for accessibility based on actual pipeline connections and valve opening / closing states. The direction and availability of pipeline connectivity causal edges are determined by the normal air supply direction and the reverse direction of possible leakage diffusion. Next, air conditioning ventilation zones are modeled as air conditioning ventilation zone nodes. Airflow channel causal edges are established based on the correspondence between supply and return air vents and the connectivity between corridors and rooms. Using the supply air to return air path as the direction, airflow channel causal edges across rooms or from rooms to corridors are activated when doors and windows are open and closed when doors and windows are closed, ensuring that the airflow channel causal edges reflect the actual air exchange path. Finally, the installation location and associated space of each combustible gas sensor are assigned to the corresponding combustible gas sensor node. Its spatiotemporal heterogeneity is determined by its spatial adjacency with the nearest pipeline node, air conditioning ventilation zone node, and door / window node. The anchoring positions in the causal graph framework enable subsequent observations based on combustible gas sensors to make causal judgments along physically accessible paths. After the nodes and causal edges are determined, spatiotemporal constraints are set for each causal edge. These constraints include at least the upper bound and validity conditions of the propagation time along pipeline-connected causal edges, and the upper bound and validity conditions of the propagation time along airflow channels. The upper bound of the propagation time along pipelines is determined based on pipe diameter, common pressure level, and minimum opening of typical leaks, ensuring that the reachability time of leaks along pipelines meets engineering rationality. The upper bound of the propagation time along airflow channels is determined based on the air supply and exhaust conditions and ventilation capacity of the air conditioning ventilation zones, ensuring that the diffusion time of pollution between zones meets the actual airflow path. At the same time, the validity of airflow channel causal edges is constrained primarily in densely populated areas by using the occupied area node, avoiding the formation of false reachable paths when ventilation paths or passage conditions are not available. Finally, the spatiotemporal heterogeneous causal graph framework is constructed according to the above heterogeneous nodes, causal edges, and spatiotemporal constraints.

[0061] S102, the real-time response of the combustible gas sensor is embedded into the spatiotemporal heterogeneous causal graph skeleton as the node state trajectory, and causal evidence is generated on the pipeline connection edge and the airflow channel edge according to the consistency of response sequence and direction, thus obtaining a candidate causal fragment set.

[0062] Among them, cross-point response embedding and causal evidence generation refers to using the spatiotemporal heterogeneous causal graph skeleton as a structural constraint to transform the real-time response of combustible gas sensor nodes into node state trajectories, and forming causal evidence on the pipeline connection causal edge and the gas flow channel causal edge for subsequent identification of the leakage-diffusion causal chain; the causal evidence includes causal edge identifier, starting node identifier, ending node identifier, direction attribute, time difference, spatiotemporal constraint satisfaction and consistency judgment result; the candidate causal segment set is a set of segments formed by sequentially splicing causal evidence that satisfies the connectivity relationship of adjacent causal edges and consistency judgment.

[0063] Optionally, the real-time response of each combustible gas sensor node can be formed into a state sequence according to the sampling time, and the state sequence can be labeled into the spatiotemporal heterogeneous causal graph skeleton according to the response stage; for each pipeline connection causal edge, the node state trajectory of its starting node and ending node can be read, and consistency judgment can be made based on the node state trajectory; according to the connection relationship of adjacent causal edges in the spatiotemporal heterogeneous causal graph skeleton and the consistency judgment result, multiple causal evidences can be spliced ​​into fragments according to direction and sequence to form a candidate causal fragment set.

[0064] Specifically, using a spatiotemporal heterogeneous causal graph skeleton as input, the real-time response of each combustible gas sensor node is formed into a state sequence according to the sampling time, and the state sequence is labeled according to the response stage. The response stage includes at least the response start time, the rising stage, the plateau stage, and the recovery stage. The relative concentration change trend and persistence are marked in each response stage, so that the node state trajectory can reflect the state characteristics of "start of an anomaly", "whether it continues" and "whether it falls back" without relying on single-point thresholds. For combustible gas sensor nodes located in the same pipeline node neighborhood or the same air conditioning ventilation zone neighborhood, the node state trajectory retains the temporal correlation field with the adjacent nodes when it is formed, which is used for subsequent determination of the response sequence and direction consistency along the causal edge of the pipeline connection or the causal edge of the airflow channel. The above node state trajectory, as a structured output, provides the response sequence and direction reference of the starting node and the ending node for the generation of causal evidence. For each pipeline-connected causal edge, the node state trajectories of its starting and ending nodes are read. When the response start time of the starting node is earlier than that of the ending node and the time difference between the two does not exceed the upper bound of the propagation time of the causal edge, and the direction attribute is consistent with the effective direction of the pipeline-connected causal edge, causal evidence of a pipeline-connected causal edge is generated, and the time difference and spatiotemporal constraint satisfaction are recorded. For each airflow channel causal edge, the node state trajectories of its starting and ending nodes are read. When the response start time of the starting node is earlier than that of the ending node and the time difference between the two does not exceed the upper bound of the propagation time of the causal edge, and the direction attribute is consistent with the effective direction of the airflow channel causal edge, causal evidence of an airflow channel causal edge is generated, and the time difference and spatiotemporal constraint satisfaction are recorded. After the above causal evidence is generated, according to the connectivity relationship and consistency judgment result of adjacent causal edges in the spatiotemporal heterogeneous causal graph skeleton, multiple causal evidences are spliced ​​into fragments according to direction and sequence to form a candidate causal fragment set and output.

[0065] S103, according to the consistency of the gas pipeline and the direction of airflow, the candidate causal segments in the candidate causal segment set are spliced ​​segment by segment, and segments inconsistent with cleaning activities or local ventilation are excluded, so as to obtain the upgrade permission conclusion, the affected zone and the causal chain direction.

[0066] The process of segmenting the causal fragments according to the consistency of the direction dominated by pipelines and airflow refers to using the effective direction of the causal edges connecting pipelines and the causal edges of airflow channels in the spatiotemporal heterogeneous causal graph skeleton as a reference. The fragments are then sequentially spliced ​​according to the continuous relationship between adjacent causal evidence in the candidate causal fragment set at the end node and the starting node. Each splice satisfies the spatiotemporal constraints of directional attributes and the upper bound of propagation time, thus forming a continuous chain of fragments. The process of upgrading the permission determination refers to making building-level risk decisions on the formed chain of fragments and outputting a permission or non-permission conclusion, which is used as the triggering condition for subsequent steps. The process of excluding fragments inconsistent with cleaning activities or local ventilation refers to removing fragments from the candidate causal fragment set that are inconsistent with the closed state of doors and windows, the air supply and exhaust conditions of air conditioning ventilation zones, or temporary local air sources, as well as fragments that are inconsistent with short-term local anomalies caused by cleaning activities. This ensures that the fragments participating in the splicing maintain physical rationality consistent with the actual air channels and pipeline connections.

[0067] Optionally, the directional consistency of each causal evidence can be checked according to the effective direction of the pipeline connection causal edge and the airflow channel causal edge, and the sequential splicing of adjacent causal evidence can be verified within the propagation time bound. If the verification is successful, the candidate causal segments in the candidate causal segment set are spliced ​​segment by segment according to the continuous relationship between the end node and the start node to form a leakage-diffusion causal chain. It is determined whether the leakage-diffusion causal chain simultaneously covers multiple rooms where the combustible gas sensor node is located, and whether it points from the gas supply direction to the user side or equipment interface location along the pipeline connection causal edge, and crosses the supply air to return air path of the air conditioning ventilation zone on the airflow channel causal edge, and an upgrade permission conclusion is generated based on the judgment result. The set of air conditioning ventilation zone nodes passed through by the leakage-diffusion causal chain is read, and the downwind area is marked as the affected zone according to the direction of the airflow channel causal edge. The directional attributes of each pipeline connection causal edge and airflow channel causal edge in the leakage-diffusion causal chain are read to obtain the causal chain direction.

[0068] Specifically, using the candidate causal fragment set as input, the system first performs a directional consistency check on each causal evidence according to the valid directions of the pipeline connection causal edge and the airflow channel causal edge, and verifies the sequential splicability of adjacent causal evidence within the propagation time bound. Then, among the causal evidence that passes the above checks, segments are spliced ​​sequentially according to the continuity relationship between the endpoint and the starting node, forming a chain structure extending from the potential leak initiation position along the pipeline connection causal edge and diffusing through the airflow channel causal edge at the adjacent air conditioning ventilation zone node. During the splicing process, segments that do not conform to the actual air exchange path are excluded based on the opening or closing status of door and window nodes, the air supply and exhaust relationship of the air conditioning ventilation zone, and the constraints of the occupied area nodes, ensuring that the chain structure extends only along reachable paths. Finally, the above chain structure is confirmed as the leak-diffusion causal chain, and this leak-diffusion causal chain is used as the basis for upgrade permission determination and output item generation. After a leakage-diffusion causal chain is formed, it is determined whether the causal chain simultaneously covers multiple rooms where the combustible gas sensor nodes are located, and whether the causal edge connecting the pipeline points from the gas supply direction to the user side or equipment interface location, and crosses the supply to return air path of the air conditioning ventilation zone on the causal edge of the airflow channel. When the above coverage and direction conditions are met, and the chain is not blocked by door and window nodes or restricted by area nodes, an upgrade permission conclusion is output; otherwise, an upgrade permission conclusion is output. This upgrade permission conclusion serves as the trigger condition for subsequent steps and, together with the affected zone and the direction of the causal chain, serves as the input for the next step. After determining the leakage-diffusion causal chain, the set of air conditioning ventilation zone nodes traversed by the chain is read. Based on the direction of the causal edge of the airflow channel, the downwind area is marked as the priority affected zone, and based on the connection relationship between the pipeline connecting causal edges and the segment valves, the room range corresponding to the pipeline isolation section with the least impact is marked. Finally, the affected zone is output as the scope of action for zone air enhancement, directional exhaust, and segment isolation. For each pipeline connection causal edge and airflow channel causal edge within the leakage-diffusion causal chain, its directional attribute is read. The direction along the pipeline is marked as the direction from the gas supply direction to the user side or from the user side to the return supply direction, and the direction along the airflow is marked as the direction from supply air to return air or from room to corridor. At the chain level, a combined direction of the pipeline connection causal edge direction and the airflow channel causal edge direction is formed, and the causal chain direction is output to determine the priority implementation of directional exhaust on the downwind side and the implementation of segmented valve position adjustment at key pipeline locations.

[0069] S104. Based on the upgrade license conclusion, the affected partition and the causal chain, combined with the occupied area, generate a security window and determine the control scope and timing requirements.

[0070] Among them, the safety window refers to the time interval during which control actions are allowed to be implemented under the constraints of the occupied area and the causal chain. The safety window includes the start time of the action, the duration of the action, and the conditions under which it can be executed; the control domain is the space and pipeline range defined within the affected zone. The control domain includes the downwind area, the corresponding set of pipeline nodes, and the set of segmented valve positions; the timing requirements are the regulations on the execution order and waiting interval of control actions within the safety window. The timing requirements include the action sequence, the action triggering conditions, and the review time point.

[0071] Optionally, if the upgrade permit conclusion is valid, the affected zone can be read, and the downwind area within the affected zone can be marked as a priority area according to the causal chain direction, forming the spatial part of the control domain; the set of pipeline nodes intersecting with the affected zone can be selected according to the direction of the causal chain along the pipeline connecting causal edge, and the set of segment valve positions with the least impact can be determined between adjacent segment valves, forming the pipeline part of the control domain; a safety window can be generated by combining the accessible time period and the on-site time period of the occupied area; and the timing requirements can be determined based on the directional priority of the causal chain direction and the spatial and pipeline parts of the control domain.

[0072] Specifically, firstly, under the condition that the upgrade permit conclusion is permitted, the affected zone is read, and the downwind area within the affected zone is marked as the priority area according to the causal chain direction, forming the spatial part of the control domain; secondly, according to the causal chain direction, the set of pipeline nodes intersecting with the affected zone is selected along the direction of the pipeline connecting the causal edge, and the set of segment valve positions with the least impact is determined between adjacent segment valves, forming the pipeline part of the control domain; thirdly, a safety window is generated by combining the accessible time period and the on-site time period of the occupied area. The safety window is constrained by not disturbing the high-occupancy area, and the start time and duration of the action are determined. The allowed execution conditions are limited by the causal chain direction, so that the control action is given priority on the downwind side; finally, based on the directional priority of the causal chain direction and the spatial and pipeline parts of the control domain, the timing requirements are determined. The timing requirements place directional exhaust first on the downwind side, followed by zonal air enhancement and segment valve position adjustment, and reserve the trigger time point and review time point for the main valve cut-off, forming a unified description of the action sequence and waiting interval.

[0073] S105 establishes an event-driven state machine based on the control scope and timing requirements, defining the state transition conditions for normal monitoring, preventive actions, escalation cutoff, and post-recovery recovery; at the lower level, constraint model predictive control is used to perform timing optimization on directional exhaust, zoned air enhancement, and segmented valve positions, and outputs the control sequence.

[0074] Among them, the event-driven state machine is a control framework used to execute the control scope and timing requirements. The event-driven state machine includes a normal monitoring state, a preventive action state, an escalation cutoff state, and a recovery state. The state transition conditions are based on the action trigger conditions, review time points, and prohibition conditions in the timing requirements, and are used to switch between the above states in one direction or two directions. The constraint model predictive control is a method for jointly orchestrating directional exhaust, zoned air enhancement, and segmented valve positions within the control scope. The joint orchestration uses the timing requirements as sequence constraints and takes preventing back diffusion and protecting high-occupancy areas as control measures. The system includes: constraint control; timing optimization, which generates an executable control arrangement within a continuous control period according to the action sequence and waiting interval specified in the timing requirements; back diffusion prevention, which prohibits pushing pollution from the downwind side back to the upwind side or introducing it into non-affected zones within the control domain; high occupancy protection, which prioritizes avoiding directing airflow to densely populated spaces identified by occupancy area nodes within the control domain; and control sequence, which is a set of control commands for directional exhaust, zoned air enhancement, and segmented valve positions arranged in chronological order. The control sequence includes the command's effective time, duration, target, and target setpoint.

[0075] Optionally, the order of entering and exiting the state can be determined by the action sequence and action triggering conditions in the timing requirements, and the normal monitoring state can be set as the default state; when the action triggering conditions in the timing requirements are met, the state can enter the preventive action state from the normal monitoring state; when the escalation triggering conditions in the timing requirements are met, the state can enter the escalation cutoff state from the preventive action state; in the escalation cutoff state, the segmented valve position adjustment is called according to the timing requirements and the execution time of the main valve cutoff is prepared, the action arrangement after the cutoff and the start time and duration of the recovery are written into the subsequent segment of the control sequence, and the control sequence is output.

[0076] Specifically, firstly, the order of entering and exiting the state is determined according to the action sequence and action triggering conditions in the timing requirements. The normal monitoring state is set as the default state. When the action triggering conditions in the timing requirements are met, the state transitions from normal monitoring to preventive action state. Then, in the preventive action state, downwind priority directional exhaust and zoned air enhancement are executed according to the timing requirements. At the review time point specified in the timing requirements, a judgment is made on whether to continue execution. When the escalation triggering conditions in the timing requirements are met, the state transitions from preventive action state to escalation cutoff state; otherwise, the state returns to normal monitoring state or maintains preventive action state until the next review time point. Subsequently, in the escalation cutoff state, the segmented valve position adjustment is called according to the timing requirements, and the execution time for main valve cutoff is prepared. The action arrangement after cutoff and the start time and duration of post-cutoff recovery are written into the subsequent segment of the control sequence. Finally, in the post-recovery recovery state, directional exhaust and gradual recovery of zoned air enhancement are executed according to the timing requirements, and the temporary adjustment of segmented valve positions is released. When post-recovery recovery is completed, the state returns to normal monitoring state. The entire state transition condition depends only on the action triggering condition, the verification time point, and the prohibition condition in the timing requirements, ensuring that the control scope and timing requirements are the only inputs to complete the establishment and use of the state machine. Within the downwind area and corresponding pipeline node set defined by the control domain, available directional exhaust equipment, zoned air supply equipment, and segmented valve positions are selected as controllable objects. Priority and restrictions on simultaneous execution are set according to the action sequence, waiting intervals, and prohibition conditions in the timing requirements, forming a sequential and mutually exclusive relationship. During continuous control periods, in accordance with the back-diffusion prevention constraint, combinations of control commands that would cause pollution to flow back from the downwind side to the upwind side or enter non-affected zones are prohibited. In accordance with the high-occupancy zone protection constraint, wind direction configurations pointing towards the occupied area are restricted in the space where the nodes in the occupied area are located, or valve position adjustments that might guide pollution into the occupied area are triggered during the occupied period. Under the premise of satisfying the above constraints and timing requirements, the directional exhaust volume, zoned air supply volume, and segmented valve opening are determined sequentially for each control moment, ensuring that control commands within the same control period can be continuously connected without violating waiting intervals, resulting in a time-arranged control arrangement. This control arrangement is then summarized into a preventative action segment, an escalation cutoff preparation segment, and a post-control recovery segment of the control sequence.

[0077] S106, based on the control sequence, escalation permit conclusion, and safety window, perform directional exhaust and zoned ventilation within the safety window, and use the combustible gas sensor network to observe whether the leak-diffusion causal chain is maintained, outputting the maintenance conclusion and residual impact zone. When the maintenance conclusion satisfies the maintenance requirement and the escalation permit conclusion continues to be valid, execute the building gas main valve shut-off within the safety window according to the control sequence, and implement minimum impact isolation for the sectional valves. Complete the aftermath recovery according to the control sequence, so that the event-driven state machine returns to the normal monitoring state.

[0078] Specifically, according to the control sequence, upgrade permission conclusion, and safety window obtained in the above steps, directional exhaust and zoned ventilation are performed within the safety window. The combustible gas sensor network is used to observe whether the leak-diffusion causal chain is maintained. If the maintenance conclusion is satisfied and the upgrade permission conclusion continues to be valid, the main building gas valve can be shut off within the safety window according to the control sequence, and the sectional valves are isolated with minimal impact. The recovery is completed according to the control sequence, so that the event-driven state machine returns to the normal monitoring state.

[0079] The data analysis method of the combustible gas sensor in the above embodiments analyzes the gas situation in the target building based on the data of the combustible gas sensor by establishing a spatiotemporal heterogeneous causal graph skeleton in the target building. In this way, the gas situation in the building can be adjusted according to the real-time situation, thus ensuring the safety of gas use.

[0080] In another embodiment, a specific example will be used to illustrate the data analysis method of the above-described combustible gas sensor:

[0081] First, information on the building's gas pipeline topology, air conditioning ventilation zones, door and window status, and occupied areas was collected, and a spatiotemporal heterogeneous causal graph skeleton was established. The symbols and terminology are defined as follows: This represents the "skeleton of a spatiotemporal heterogeneous causal graph"; Represents a "heterogeneous set of nodes", where For "combustible gas sensor node set", For "pipeline node set", For "Air Conditioning and Ventilation Zone Node Set", For "door and window node set", For "the set of nodes occupying the area"; This represents the "set of causal edges connected to pipelines". Represents "the set of causal edges of airflow channels"; Indicates "pipeline connectivity causal edge" "Upper bound of propagation time" Indicates "pipeline connectivity causal edge" "Validity conditions"; Indicates "airflow channel causal edge" "Upper bound of propagation time" Indicates "airflow channel causal edge" "Validity conditions"; Indicates "causal edge" "Valid directional attribute"; Represents "node" "the set of adjacent nodes"; Indicates "occupied area node" "The on-site time period." Specifically, "mapping the combustible gas sensor network into heterogeneous nodes" means establishing each combustible gas sensor, gas pipeline location, air conditioning ventilation zone, door and window, and occupied area within the building as a heterogeneous node and assigning it to a specific node. , , , , "Establishing causal relationships based on pipeline connections and airflow channels" refers to establishing a causal framework based on pipeline connections and actual airflow paths. and "Forming a spatiotemporally constrained spatiotemporal heterogeneous causal graph framework" refers to... and Set separately and This ensures that each causal edge has executable time and validity constraints.

[0082] In terms of specific modeling methods, the first step is to establish a model based on the gas pipeline topology. and Mark the pipe section bends, valve locations, and equipment interface locations as pipeline nodes and add them to the pipeline network. Establish causal relationships for pipeline connectivity based on actual pipeline connections and valve opening / closing states, and add them to the pipeline network. For each set up To reflect the reverse direction of normal airflow and potential leakage diffusion; for each set up To reflect the factors affecting the effectiveness of the edge, such as valve opening and closing, maintenance occupation, etc., the following conditions are set: An upper bound for propagation time along the pipeline is established, determined based on pipe diameter, commonly used pressure rating, and minimum opening of typical leak points, ensuring that the reachability time along the pipeline meets engineering rationality requirements. Subsequently, a system is established based on the relationship between air conditioning ventilation zones and room connectivity. and The space units covered by supply and return air are identified as air conditioning ventilation zone nodes and added to the system. Each air duct is set with the air supply to return direction as the direction. of When doors and windows are open, a cross-space airflow channel is established based on the connection between the room and the corridor, creating a causal boundary. When doors and windows are closed, the corresponding boundary is... Set as not satisfied, thus Only effective on real air exchange paths; for each set up The upper limit of the time that reflects the diffusion in different zones is determined based on the supply and exhaust ventilation conditions and the ventilation capacity.

[0083] Next, combustible gas sensor mapping and anchoring are performed: the installation location of each combustible gas sensor and its associated space are identified as follows. The nodes in the array are determined through spatial adjacency relationships. ,in of It should at least include the nearest pipeline node, air conditioning and ventilation zone node, and door and window node, so that the combustible gas sensor node is in It has a clearly defined anchoring position, which facilitates subsequent movement along... and Make causal judgments. Establish. To record the opening and closing status of door and window nodes, and to directly affect the status of these nodes. of ;Establish To record the nodes occupying the area, and with The validity of edges is limited to high-occupancy periods to prevent the formation of false reachable paths or unreasonably controlled entrances under conditions where ventilation paths are unavailable or on-site operations are inconvenient. Through the setting of the aforementioned nodes and causal edges, the spatiotemporal constraints are configured to ensure... It meets the physical constraints of the building in terms of both spatial connectivity and temporal accessibility.

[0084] Finally, the above , , and its corresponding and Combined into a spatiotemporal heterogeneous causal graph skeleton and output As input to S2, the implementation method ensures that "building structure modeling and combustible gas sensor mapping" unifies the physical paths of gas pipelines and air channels with the deployment of combustible gas sensors into the same graphical model, enabling subsequent "cross-point response embedding and causal evidence generation" to proceed along... and The effective direction and propagation time upper bound are carried out to serve the core scenario of "total valve upgrade determination when low concentration anomalies in multiple rooms do not reach the single-point threshold".

[0085] Based on the spatiotemporal heterogeneous causal graph skeleton As input, where From heterogeneous node set With causal edge set , It consists of its spatiotemporal constraints. The notation conventions are as follows: It is a set of combustible gas sensor nodes. For the set of causal edges connecting pipelines, For the set of causal edges of the airflow channel, Connect cause and effect edges for pipelines The upper bound of the propagation time, Connect cause and effect edges for pipelines Validity conditions Causal edge of airflow channel The upper bound of the propagation time, Causal edge of airflow channel Validity conditions Causal edge The effective direction attribute. For each combustible gas sensor node. Define the real-time response curve For the sensor in time The response value defines the node state trajectory. Based on The phase labeling function has the following values: "Normal," "Rising Phase," "Plateau Phase," and "Recovery Phase." Let... for The moment of entering the "ascent phase" makes This marks the moment to enter the "platform phase," The moment to enter the "recovery phase" is approaching. The duration of the "ascending phase" is determined by... This refers to the duration of the "platform phase." Based on the above definition, the real-time response of each combustible gas sensor node is embedded as a node state trajectory, and this is preserved. , , The timing elements used for causal determination provide the necessary input for judging the "response sequence and direction consistency" along the causal edge connecting the pipeline and the causal edge of the airflow channel.

[0086] In the generation of causal evidence, for each causal edge ,in For the starting node, Define the time difference for the endpoint node. And define a unified propagation time upper bound. With uniform validity conditions As follows: When hour, , ;when hour, , To transform the "consistency of response sequence and direction" into a calculable judgment and metric, a causal evidence consistency score is constructed. It is used to measure within spatiotemporal constraints. Whether it can be considered a propagation segment of leakage-diffusion is defined as follows:

[0087] ;

[0088] in, This is a conditional function that evaluates to a value when the condition within the parentheses is true. Otherwise, the value is ; As edge type weight, when hour ,when hour , and These are the constant weights corresponding to the causal edges of pipeline connectivity and airflow channels, respectively. Starting node The duration of the "rising phase"; End point The duration of the "rising phase"; The time difference between the "response start time"; This is the upper bound of the propagation time of the causal edge; This is a condition for the validity of the causal edge; This represents the valid direction attribute of the causal edge. Through this calculation, It simultaneously reflects "response sequence", "satisfaction of spatiotemporal constraints", "consistency of directional attributes" and "persistent elements", when Time indicates It has causal rationality as a transmission segment in the current time period.

[0089] Based on the above Generate causal evidence for each causal edge. Causal evidence includes causal boundary markers. Starting point node identifier End point identifier Directional attributes Time difference 1. Spatiotemporal constraint satisfaction Consistency determination results In the formation of candidate causal fragments, the spatiotemporal heterogeneous causal graph skeleton is followed. The connectivity of adjacent causal edges in the equation, satisfying the condition... The causal evidence is pieced together sequentially to form a set of fragments. Specifically, let the fragments... From ordered causal edge sequence Composition, satisfaction and and satisfy and If the sequence is valid, then it constitutes a candidate segment; to measure the overall usability of the segments, a segment score is defined. ,in The minimum value operation is performed. Finally, all segments satisfying the above connectivity and timing conditions are summarized to obtain a candidate causal segment set, which is then output. Through the above implementation method, [the following is achieved]: To constrain the structure, the real-time response of the combustible gas sensor node is embedded as the node state trajectory, and calculable causal evidence consistent with the physical path of the building is generated on the causal edges of the pipeline connection and the airflow channel. This enables the subsequent identification of the leakage-diffusion causal chain to be carried out within the path and time window consistent with the gas pipeline and air channel.

[0090] In one embodiment, a set of candidate causal fragments is used as input. The following conventions continue: For the "skeleton of spatiotemporal heterogeneous causal graph", For "the set of causal edges connected to pipelines", For "the set of causal edges of the airflow channel", and These are the causal edges of the pipeline connections. The upper bound of propagation time and the conditions for effectiveness. and These are the causal sides of the airflow channel. The upper bound of propagation time and the conditions for effectiveness. Causal edge The effective direction attribute, For "combustible gas sensor node set", Combustible gas sensor node In time The response For nodes The node state trajectory, For nodes The moment of entering the "ascent phase" For nodes The duration of the "ascent phase" Causal edge Based on the calculated consistency score, the candidate causal segment is denoted as... And its segment rating is Let the set of all candidate causal segments be denoted as . The goal is to... The system constructs a "leakage-diffusion causal chain" that satisfies the physical path and temporal constraints of the building, and outputs "upgrade permit conclusion", "affected zone" and "causal chain direction" based on this.

[0091] First, segment-by-segment splicing is performed to form a chain of candidates. As input, following the requirement of "segmenting sequentially according to the consistency of pipeline and airflow direction", segments are designed to ensure that the end nodes of adjacent segments are consistent with the start nodes and along the same direction. Maintain consistency in direction and in or Fragments within the range that meet the temporal continuity conditions are sequentially spliced ​​together to construct a chain-like candidate sequence. ,in and The end node equals The starting node. To avoid segments inconsistent with cleaning activities or local ventilation from entering the splicing, the set of combustible gas sensor nodes involved in each location to be spliced ​​is verified. Does the duration of the "rising phase" meet the requirements, i.e. ,in A minimum duration threshold is set for the project; if this threshold is not met, the splicing attempt is rejected. Through these rules, a chain-like candidate set that satisfies both physical reachability and durability requirements is obtained.

[0092] Based on the chained candidate set, in order to unify the "formation of a leakage-diffusion causal chain" and the "upgrade permission determination" into a computable criterion, for each chained candidate... Define license score as follows:

[0093] ;

[0094] in, Let this be a conditional decision function; if the condition is true, then... Otherwise take ; For chain candidates The set of all causal edges in the equation; For chain candidates The set of combustible gas sensor nodes covered; For chain candidates The set of air conditioning and ventilation zone nodes traversed; To unify the validity conditions, when hour ,when hour ; To unify the upper bound of the propagation time, when hour ,when hour .

[0095] The above Based on "fragment consistency", the building-level risk coverage criteria are "simultaneously including pipeline propagation and airflow diffusion" and "spanning multiple combustible gas sensor nodes and air conditioning ventilation zones". The criteria for physical rationality and exclusion of short-term local anomalies are "edge validity", "time sequence and propagation upper bound" and "persistence constraints". This ensures that the "upgrade permission determination" is consistent with the building's physical path and combustible gas sensor observations.

[0096] After obtaining each chain of candidates of Then, the chain candidate with the highest permission score is selected as the leakage-diffusion causal chain, i.e. .when At that time, confirm This is a "leakage-diffusion causal chain". Based on Generate an "upgrade license conclusion", ,in For the binary variable representing the upgrade licensing conclusion, The allowable threshold constant; when When permitted, At that time, permission was not granted. Subsequently, the "causal chain" was extracted, i.e., read... Each causal edge within Directional attribute , will along The direction markings are from the gas supply direction to the user side or from the user side to the return supply direction, along... The direction markers are set as supply air pointing to return air or room pointing to corridor. These are combined in the order of the links to obtain an ordered set of causal chain directions, which serves as the output item for the "causal chain direction". Finally, the "influence zone" is determined, i.e., based on... Based on this, the downwind air conditioning ventilation zone indicated by the causal chain direction is marked as the priority influence zone, and will be compared with... The set of intersecting pipeline nodes is used to indicate the segmented isolation range with minimal impact, forming the "impact partition" output item.

[0097] In one embodiment, using "upgrade license conclusion," "affected partition," and "causal chain direction" as inputs, the following standardized symbols and terminology are used: A binary variable representing the "upgrade license conclusion", when a license is granted. When not permitted ; This indicates a chain reaction confirming a "leakage-diffusion causal chain"; express The set of all causal edges included; express The set of covered air conditioning and ventilation zone nodes is used as a set representation of the "affected zone"; Indicates causal edge Valid directional attributes; and These represent the set of causal edges for pipeline connectivity and the set of causal edges for airflow channels, respectively. Represents the set of nodes occupying the region. Indicates the occupied area node The period of presence of a node includes the start and end times of its presence. Indicates the upper bound of the uniform propagation time, when hour ,when hour ; This represents the time difference between the defined "response start time"; This represents a conditional function that evaluates to a value when the condition is true. Otherwise ; This indicates the current reference time. The goal is to generate a "safety window" and determine the "control scope and timing requirements" based on the "impact partition" and "causal chain direction" combined with the presence time of the occupied area.

[0098] First, based on the "causal chain direction," the downwind area is identified within the "influence zone," forming the spatial and pipeline components of the "control domain." Specifically: [The process involves] reading... And in accordance with Internal airflow channel causal edge of The direction is defined as follows: the air conditioning ventilation zone covered at the end of the path from the supply air to the return air is marked as the downwind area, serving as the spatial portion of the "control area"; in the pipeline connection section, read... Internal pipeline connection cause edge The set of pipeline nodes adjacent to these causal edges is defined as the pipeline portion of the "control domain." A set of sectional valve positions with minimal impact is determined between adjacent sectional valves for subsequent sectional isolation and main valve upgrade preparation. The spatial and pipeline portions of the aforementioned "control domain" are bounded by "affected zones," with the directional attribute of the "causal chain direction" as priority. This ensures that directional exhaust and sectional airflow enhancement first act on the downwind region, and sectional valve position adjustments first act on the areas adjacent to the main valve. The set of intersecting pipeline nodes.

[0099] When generating the "safety window," the "upgrade license conclusion," "affected partition," "causal chain direction," and the presence time of the occupied area are unified into a calculable time interval. The set of occupied area nodes that overlap with the "affected partition" space, i.e., the space corresponding to the coverage area. For each ,make Let be the earliest moment when the validity condition of the causal edge changes from unsatisfied to satisfied, i.e. The initial value is The moment; the order to line The duration of directional exhaust action, wind The duration of the zonal air enhancement operation, valve The duration of the segmented valve position adjustment action, slow This is a constant for the safety buffer duration. Define the "safety window". for:

[0100] ;

[0101] in, For occupying area nodes At the current reference time The moment of the end of the presence within; Ensure that control actions are carried out only when the condition of doors and windows and the air supply and exhaust conditions of the air conditioning and ventilation zones allow; The duration of the "safe window" is limited by the duration of the action and the remaining time beyond the upper bound of the propagation time, preventing actions from being performed outside the upper bound of the propagation time. Through calculation, the start and end times of the "safe window" are strictly bound to the "upgrade permission conclusion," "affected partition," "causal chain direction," and the presence time of the occupied area.

[0102] When determining the "timing requirements," the spatial and pipeline portions of the "control scope" are the objects of action, and the "safety window" is used as the reference point. For the execution interval, generate the action sequence, action trigger conditions, and review time points. Specifically: place directional exhaust as the first action in the downwind air conditioning ventilation zone, and... As the starting moment of the action, continuous sorting After directional exhaust is completed, the zoned airflow is activated, triggered by a zero waiting interval or the fulfillment of the prohibition condition. After the above two types of actions are completed, ventilation is carried out. As a review point, segmented valve position adjustments were performed within the pipeline section of the "control area," and the valve position was continuously adjusted. The prohibition conditions for the aforementioned actions are given by the directional attributes determined by the "causal chain direction" and the spatial boundaries of the "control area." It is prohibited to direct airflow towards the upwind side of areas not within the "affected zone," and it is prohibited to trigger valve position adjustments that could guide pollution into the occupied area during the period the occupied area is present. Finally, the spatial and pipeline components of the "control area," along with the action sequence, triggering conditions, and review time points of the "timing requirements," are summarized into "Control Area and Timing Requirements."

[0103] In one embodiment, using "control scope and timing requirements" as input, an upper-level event-driven state machine is built, and constraint model predictive control orchestration is performed at the lower level. Standardized terminology and symbols are used, and the following conventions apply: For "safety window", The moment of the start of the action. The moment when the action ends; the spatial portion of the "control scope" is denoted as This refers to the set of air conditioning and ventilation zones located downwind of the "affected zone"; the pipeline portion of the "control area" is denoted as... Let be the set of pipeline nodes that intersect with the "leakage-diffusion causal chain"; the set of segmented valve positions is denoted as . The collection of directional exhaust equipment is denoted as exhaust. The collection of zoned air supply equipment is denoted as wind. ; respectively row ,wind ,valve The "sequence requirements" include the order of actions, the triggering conditions for the actions, and the review time point. The review time point is recorded as the exhaust time point. The prohibitions include "preventing back diffusion" and "protecting high-occupancy areas." The "preventing back diffusion" constraint prohibits airflow from the leeward side to the upwind side or to areas outside the "affected zone." The "protecting high-occupancy areas" constraint prohibits creating winds pointing towards the occupied area during the presence of nodes in the occupied area, or performing valve position adjustments that could guide pollution into the occupied area during that period. The control sequence is denoted as... A set of control instructions arranged in chronological order. Each instruction includes the time when it takes effect, its duration, the target object, and the target set value.

[0104] In establishing an event-driven state machine, four states are defined, and a single active state is maintained: normal monitoring state. Preventive action state Upgrade the cut-off status and post-disaster recovery status Default entry When the action triggering condition in the "Timing Requirements" coincides with the start time of the "Safety Window" When both conditions are met, by Leap to .Enter Afterwards, in the row Inside, only for Downwind side air conditioning ventilation zone call-up Perform directional exhaust; exhaust air Inside, for the same Call the wind Implement zoned ventilation to maintain exhaust channels and record the verification time points. When the upgrade trigger condition in "Timing Requirements" is met and the prohibition condition is not triggered, the upgrade is initiated by... Leap to Otherwise Time return Or continue to maintain Until the "safe window" closes. In, call Perform minimally impactful segmented valve position adjustments to restrict pipeline connectivity. Internally, prepare the valve position combination and timing points required for the main valve to shut off; then proceed to... ,exist In accordance with the "time-sequence requirements" for post-disaster recovery, the directional exhaust and zoned ventilation were reduced, and the temporary sectional valve position adjustments were cancelled. Return before the end All transitions in the aforementioned state machine are based solely on "control scope and timing requirements," and all actions are limited to... Execute internally.

[0105] In constrained model predictive control orchestration, the lower-level control is based on... , ,Row ,wind and As a controllable object, the sequence of actions and waiting intervals according to "timing requirements" are used as sequence constraints, and "preventing back diffusion" and "protecting high-occupancy areas" are used as control constraints to generate objects that meet the requirements. Continuously executed control arrangements. During the directional exhaust phase, priority is given to... The downwind end area is used as the exhaust point, and the zonal air enhancement of adjacent areas is delayed until the directional exhaust is completed to avoid forming a backflow path; during the segmented valve position adjustment phase, only the... The valve position corresponding to the pipeline node intersecting the "leakage-diffusion causal chain" will activate, prohibiting crossing to pipeline segments outside the "affected zone". At any time, if the prohibition condition of "protecting the high occupancy area" is met, the instruction that guides airflow to the occupancy area node will be delayed or replaced until the prohibition condition is lifted or the "safety window" ends.

[0106] control sequence It consists of three consecutive instruction segments: a preventative action segment, an escalation cutoff preparation segment, and a post-incident recovery segment. The preventative action segment begins at... And sequentially includes pairs Directional exhaust command and wind The zoned ventilation increase command; the upgrade cutoff preparation section begins at And includes the following: The segmented valve position adjustment command and the trigger time mark reserved for the main valve cutoff; the post-recovery section covering the exhaust valve. It also includes instructions for directional exhaust and zoned airflow enhancement and desiccation. Each instruction is defined by the spatial and pipeline components of the "control scope," and the action triggering conditions and verification time points provided by the "timing requirements" serve as the time anchors for execution and confirmation.

[0107] In one embodiment, to control sequence As the basis for execution, and in the generated security window Internal actions and reviews shall be carried out. Standardized terminology and symbols shall be used, as follows: A confirmed leak-diffusion causal chain; for The set of all causal edges included; and These are the sets of causal edges for pipeline connectivity and the sets of causal edges for airflow channels, respectively. Causal edge The validity condition, when hour ,when hour ; To unify the upper bound of the propagation time, when hour ,when hour ; It is a set of combustible gas sensor nodes. For nodes The node state trajectory, For nodes In time The response; Defined as an activity indicator function, when When it is the "ascent phase" or "plateau phase" ,otherwise ; This is a conditional decision function; it evaluates the result when the condition is true. Otherwise take ; To upgrade the license conclusion, when the license is granted. When not permitted ; To control the spatial portion of the area, the leeward side air conditioning and ventilation zones are grouped together; For the pipeline section within the control area, the set of pipeline nodes intersecting with the leakage-diffusion causal chain; For segmented valve position assembly; exhaust For the review time point; [List] ,wind ,valve These represent the duration of actions for directional exhaust, zoned airflow enhancement, and segmented valve position adjustment, respectively. From The candidate set of main valve cut-off trigger times is reserved in the middle; This is the current reference time. The execution flow is as follows: In Neiyi Completed one after another Directional exhaust and zoned airflow within; Conduct a maintenance review; Based on the review and upgrade approval conclusions, the system decides whether to trigger the main valve shut-off and minimum impact isolation, and then completes the follow-up recovery.

[0108] To transform "preservation review" into a calculable conclusion, an observation interval is set. ,in This is the observation duration constant. Internally, along the leakage-diffusion causal chain For each causal edge Calculate persistent consistency items And aggregated into a link persistence score. The definition is as follows:

[0109] ;

[0110] in, and Causal sides The starting node and the ending node at time Activity instructions; This is a condition for edge validity; and "Response start time"; This is the upper bound of the propagation time for that edge; Values or This indicates whether the causal chain remains active throughout the observation interval and satisfies the spatiotemporal constraints. Based on this, a preservation-of-conclusion variable is given. ,when This indicates that the causal chain persists even after preventative actions, demonstrating a continued tendency to spread within pipelines and spaces. This indicates that the causal chain has been broken or has fallen back.

[0111] In the determination and execution of "main valve upgrade and shut-off", based on and This is a joint condition, and is limited to a set of candidate trigger times within a safe window. Select trigger time If the above set is empty, it will not be triggered.

[0112] Define cutoff trigger decision variables ,when At that time, Issue the main valve shut-off command, and according to and Perform segmented isolation with minimal impact, limiting connectivity to [the area between the two points]. Within the range of intersecting pipeline nodes; when If the cutoff is not triggered, maintain the preventive action until... End or await review in a subsequent cycle. Execute together after disconnection. The post-disaster recovery phase, for The directional exhaust and zoned ventilation within the system were reduced, and the temporary sectional valve position adjustments were cancelled to ensure a return to normal monitoring status.

[0113] To provide a basis for the scope of disposal in subsequent management stages, At the end of the process or after the cutoff and isolation are completed, output the set of "residual impact partitions". To define the set of affected partitions, the residual affected partitions are defined as follows:

[0114] ;

[0115] in, The downwind air conditioning ventilation zone, which is maintained in the "rising phase" or "plateau phase" by the combustible gas sensor node at the end of the safety window, is included to guide continued exhaust or further manual verification.

[0116] Through the above implementation method, preventive actions are strictly executed in accordance with the control sequence within the safety window. At the review time point, the node status trajectory of the combustible gas sensor network is used to determine the sustainability of the leakage-diffusion causal chain. When both the permission and sustainability are met, the building gas main valve is cut off and the minimum impact isolation is triggered at the reserved time. Subsequently, the follow-up recovery is completed, forming a closed-loop handling of the main valve upgrade judgment scenario when the low concentration anomaly in multiple rooms does not reach the single-point threshold.

[0117] Based on the same inventive concept, this application also provides a data analysis device for a combustible gas sensor to implement the data analysis method for the combustible gas sensor described above. The solution provided by this device is similar to the solution described in the above method; therefore, the specific limitations in one or more embodiments of the data analysis device for a combustible gas sensor provided below can be found in the limitations of the data analysis method for the combustible gas sensor described above, and will not be repeated here.

[0118] In one embodiment, such as Figure 2 As shown, a data analysis device for a combustible gas sensor is provided, the device comprising:

[0119] The skeleton construction module 20 is used to construct a spatiotemporal heterogeneous causal graph skeleton based on the gas pipeline topology, air conditioning and ventilation zones, door and window status and door and window occupied areas of the target building.

[0120] The fragment acquisition module 21 is used to embed the real-time response of the combustible gas sensor into the spatiotemporal heterogeneous causal graph skeleton as the node state trajectory, and generate causal evidence on the pipeline connection edge and the airflow channel edge according to the consistency of response sequence and direction, so as to obtain a candidate causal fragment set.

[0121] The conclusion acquisition module 22 is used to splice the candidate causal segments in the candidate causal segment set segment by segment according to the consistency of the gas pipeline and the direction dominated by the airflow, and to exclude segments that are inconsistent with the cleaning activities or local ventilation, so as to obtain the upgrade permission conclusion, the affected zone and the causal chain direction.

[0122] Module 23 is required to generate a safety window based on the upgrade license conclusion, the affected partition and the causal chain, combined with the occupied area, and to determine the control scope and timing requirements.

[0123] The sequence output module 24 is used to establish an event-driven state machine based on the control scope and timing requirements, define the state transition conditions for normal monitoring, preventive actions, escalation cutoff and post-recovery recovery; and perform timing optimization of directional exhaust, zoned air enhancement and segmented valve positions using constraint model predictive control at the lower level, and output the control sequence.

[0124] The partition output module 25 is used to perform directional exhaust and partitioned ventilation within the safety window according to the control sequence, upgrade permission conclusion and safety window, and to use the combustible gas sensor network to observe whether the leak-diffusion causal chain is maintained, output the maintenance conclusion and residual impact partition, and when the maintenance conclusion is satisfied and the upgrade permission conclusion continues to be valid, the main gas valve of the building is shut off within the safety window according to the control sequence, and the section valves are isolated with minimal impact. The post-recovery is completed according to the control sequence, so that the event-driven state machine returns to the normal monitoring state.

[0125] This application also provides an electronic device, in some embodiments, referring to... Figure 3 As shown, the electronic device 700 includes an input unit 710, a memory 720, a processor 730, and an output unit 740. The memory 720 stores program instructions that can be executed on the processor 730. The processor 730 can execute the data analysis method and / or technical solution based on the combustible gas sensor in the foregoing embodiments by calling the program instructions. The electronic device 700 can be a mobile terminal device such as a mobile phone or a computer.

[0126] Furthermore, embodiments of this application also provide a computer-readable storage medium for storing a computer program that performs a data analysis method for a combustible gas sensor. For example, computer program instructions, when executed by a computer, can invoke or provide the methods and / or technical solutions according to this application through the operation of the computer. The program instructions for invoking the methods of this application may be stored in a fixed or removable storage medium, and / or transmitted via data streams in broadcast or other signal carrying media, and / or stored in a storage medium that operates according to the program instructions.

[0127] Obviously, those skilled in the art should understand that the modules or steps of this application described above can be implemented using general-purpose computing devices. They can be centralized on a single computing device or distributed across a network of multiple computing devices. Optionally, they can be implemented using computer-executable program code, thereby storing them in a storage device for execution by a computing device, or fabricating them separately as individual integrated circuit modules, or fabricating multiple modules or steps as a single integrated circuit module. Thus, this application is not limited to any particular combination of hardware and software.

[0128] The technical features of the above embodiments can be arbitrarily integrated. For the sake of brevity, not all possible integrations of the technical features in the above embodiments are described. However, as long as the integration of these technical features does not contradict each other, they should be considered to be within the scope of this specification.

[0129] The above embodiments merely illustrate several implementation methods of the present invention, and their descriptions are relatively specific and detailed, but they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these all fall within the protection scope of the present invention. Therefore, the protection scope of this invention patent should be determined by the appended claims.

Claims

1. A data analysis method for a combustible gas sensor, characterized in that, The method includes: Based on the gas pipeline topology, air conditioning and ventilation zones, door and window status, and door and window occupancy areas of the target building, a spatiotemporal heterogeneous causal graph skeleton is constructed. The real-time response of the combustible gas sensor is embedded into the spatiotemporal heterogeneous causal graph skeleton as a node state trajectory, and causal evidence is generated on the pipeline connection edge and the gas flow channel edge based on the consistency of response sequence and direction to obtain a candidate causal fragment set. The candidate causal segments in the candidate causal segment set are pieced together segment by segment according to the consistency of the gas pipeline and the dominant airflow direction, and segments inconsistent with cleaning activities or local ventilation are excluded to obtain the upgrade permission conclusion, the affected zone and the causal chain direction; including: according to the effective direction of the pipeline connection causal edge and the airflow channel causal edge, the directional consistency test is performed on each causal evidence, and the sequential splicability of adjacent causal evidence is verified within the propagation time upper bound; if the verification is passed, the candidate causal segments in the candidate causal segment set are pieced together segment by segment according to the continuity relationship between the end node and the start node to form a leakage-diffusion causal chain; Determine whether the leak-diffusion causal chain simultaneously covers multiple rooms where the combustible gas sensor node is located, and whether the causal edge connecting the pipeline points from the gas supply direction to the user side or equipment interface location, and crosses the supply air to return air path of the air conditioning ventilation zone on the causal edge of the airflow channel, and generate an upgrade permission conclusion based on the determination result; read the set of air conditioning ventilation zone nodes through which the leak-diffusion causal chain passes, and mark the downwind area as the affected zone according to the direction of the causal edge of the airflow channel; read the directional attributes of each pipeline connecting causal edge and airflow channel causal edge in the leak-diffusion causal chain to obtain the direction of the causal chain; Based on the upgrade permit conclusion, the affected zones, and the causal chain direction, combined with the occupied area, a safety window is generated, and the control domain and timing requirements are determined. This includes: under the condition that the upgrade permit conclusion is permissible, reading the affected zones, and marking the downwind area within the affected zones as a priority area according to the causal chain direction, forming the spatial portion of the control domain; selecting a set of pipeline nodes intersecting the affected zones according to the direction of the causal chain along the pipeline connecting causal edges, and determining the set of segment valve positions with the least impact between adjacent segment valves, forming the pipeline portion of the control domain; generating a safety window by combining the accessible time period and the on-site time period of the occupied area; and determining the timing requirements based on the directional priority of the causal chain direction and the spatial and pipeline portions of the control domain. Based on the control scope and timing requirements, an event-driven state machine is established, defining the state transition conditions for normal monitoring, preventive actions, escalation cutoff, and post-recovery recovery. At the lower level, constraint model predictive control is used to perform timing optimization on directional exhaust, zoned air enhancement, and segmented valve positions, and outputs a control sequence. Based on the control sequence, upgrade permission conclusion, and safety window, directional exhaust and zoned ventilation are performed within the safety window. The combustible gas sensor network is used to observe whether the leak-diffusion causal chain is maintained, and the maintenance conclusion and residual impact zone are output. When the maintenance conclusion is satisfied and the upgrade permission conclusion continues to be valid, the main building gas valve is shut off within the safety window according to the control sequence, and the sectional valves are isolated with minimal impact. The aftermath recovery is completed according to the control sequence, so that the event-driven state machine returns to the normal monitoring state.

2. The data analysis method for a combustible gas sensor as described in claim 1, characterized in that, Based on the gas pipeline topology, air conditioning and ventilation zones, door and window status, and occupied areas of the target building, a spatiotemporal heterogeneous causal graph skeleton is constructed, including: Establish pipeline nodes and pipeline connectivity based on the gas pipeline topology; the pipeline nodes include pipe segment inflection points, valve locations, and equipment interface locations; The air conditioning ventilation zones are modeled as air conditioning ventilation zone nodes, and causal edges of airflow channels are established based on the correspondence between supply air vents and return air vents and the connectivity between corridors and rooms. Based on the pipeline nodes, pipeline connectivity, air conditioning and ventilation zones, door and window status, and occupied areas, establish the causal constraint relationship of the airflow channel; Construct a spatiotemporal heterogeneous causal graph skeleton based on the aforementioned constraints.

3. The data analysis method for a combustible gas sensor as described in claim 2, characterized in that, The real-time response of the combustible gas sensor is embedded into the spatiotemporal heterogeneous causal graph skeleton as a node state trajectory, and causal evidence is generated on the pipeline connection edge and the gas flow channel edge based on the consistency of response sequence and direction, resulting in a candidate causal fragment set, including: The real-time response of each combustible gas sensor node is formed into a state sequence according to the sampling time, and the state sequence is labeled into the spatiotemporal heterogeneous causal graph skeleton according to the response stage. For each pipeline connection causal edge, read the node state trajectory of its starting node and ending node, and make a consistency judgment based on the node state trajectory. Based on the connectivity and consistency determination results of adjacent causal edges in the spatiotemporal heterogeneous causal graph skeleton, multiple causal evidences are spliced ​​into fragments according to direction and chronological order to form a candidate causal fragment set.

4. The data analysis method for a combustible gas sensor as described in claim 3, characterized in that, An event-driven state machine is established based on the aforementioned control scope and timing requirements, defining state transition conditions for normal monitoring, preventative actions, escalation cutoff, and post-incident recovery. At the lower level, constrained model predictive control is used to perform timing optimization on directional exhaust, zoned airflow enhancement, and segmented valve positions, outputting a control sequence including: The order of entering and exiting the state is determined by the sequence of actions and the triggering conditions in the timing requirements, and the normal monitoring state is set as the default state. When the action triggering conditions in the timing requirements are met, the system will switch from the normal monitoring state to the preventive action state. When the upgrade trigger condition in the timing requirements is met, the state changes from the preventive action state to the upgrade cutoff state. During the upgraded cutoff state, the segmented valve position adjustment is called according to the timing requirements to prepare for the execution time of the main valve cutoff. The action arrangement after the cutoff and the start time and duration of the recovery are written into the subsequent segment of the control sequence, and the control sequence is output.

5. A data analysis device for a combustible gas sensor, characterized in that, The device includes: The skeleton construction module is used to construct a spatiotemporal heterogeneous causal graph skeleton based on the gas pipeline topology, air conditioning and ventilation zones, door and window status, and door and window occupancy areas of the target building. The fragment acquisition module is used to embed the real-time response of the combustible gas sensor into the spatiotemporal heterogeneous causal graph skeleton as a node state trajectory, and generate causal evidence on the pipeline connection edge and the gas flow channel edge according to the consistency of response sequence and direction, so as to obtain a candidate causal fragment set. The conclusion acquisition module is used to piece together candidate causal segments from the candidate causal segment set segment by segment according to the consistency of the direction dominated by the gas pipeline and the airflow, and to exclude segments inconsistent with cleaning activities or local ventilation, thereby obtaining the upgrade permission conclusion, the affected zone, and the causal chain direction; including: checking the directional consistency of each causal evidence according to the effective direction of the causal edge connecting the pipeline and the causal edge of the airflow channel, and verifying the sequential splicability of adjacent causal evidence within the propagation time upper bound; if the verification is successful, then splicing the candidate causal segments from the candidate causal segment set segment by segment according to the continuity relationship between the end node and the start node to form a leak-diffusion. Causal chain; determine whether the leakage-diffusion causal chain simultaneously covers multiple rooms where the combustible gas sensor node is located, and whether the causal edge connecting the pipeline points from the gas supply direction to the user side or equipment interface location, and crosses the supply air to return air path of the air conditioning ventilation zone on the causal edge of the airflow channel, and generate an upgrade permission conclusion based on the judgment result; read the set of air conditioning ventilation zone nodes passed through by the leakage-diffusion causal chain, and mark the downwind area as the affected zone according to the direction of the causal edge of the airflow channel; read the directional attributes of each pipeline connecting causal edge and airflow channel causal edge in the leakage-diffusion causal chain to obtain the direction of the causal chain; The module is required to generate a safety window and determine the control domain and timing requirements based on the upgrade permit conclusion, the affected zone, the causal chain direction, and the occupied area. This includes: under the condition that the upgrade permit conclusion is valid, reading the affected zone and marking the downwind area within the affected zone as a priority area according to the causal chain direction, forming the spatial portion of the control domain; selecting a set of pipeline nodes intersecting the affected zone along the direction of the causal chain connecting the pipeline causal edges, and determining the set of segment valve positions with the least impact between adjacent segment valves, forming the pipeline portion of the control domain; generating a safety window by combining the accessible time period and the on-site time period of the occupied area; and determining the timing requirements based on the directional priority of the causal chain direction and the spatial and pipeline portions of the control domain. The sequence output module is used to establish an event-driven state machine based on the control scope and timing requirements, and define the state transition conditions for normal monitoring, preventive action, escalation cutoff and post-recovery recovery; at the lower level, it uses constraint model predictive control to perform timing optimization on directional exhaust, zoned air enhancement and segmented valve positions, and outputs the control sequence. The partition output module is used to perform directional exhaust and partitioned ventilation within the safety window according to the control sequence, upgrade permission conclusion, and safety window, and to observe whether the leak-diffusion causal chain is maintained using a combustible gas sensor network. It outputs the maintenance conclusion and residual impact partition. When the maintenance conclusion satisfies the maintenance requirement and the upgrade permission conclusion continues to be valid, it performs the building gas main valve shut-off within the safety window according to the control sequence, and implements minimum impact isolation for the sectional valves. It completes the follow-up recovery according to the control sequence, so that the event-driven state machine returns to the normal monitoring state.

6. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the data analysis method for the combustible gas sensor as described in any one of claims 1 to 4.

7. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the data analysis method for the combustible gas sensor according to any one of claims 1 to 4.

8. A computer program product, comprising a computer program, characterized in that, When executed by a processor, the computer program implements the data analysis method for the combustible gas sensor as described in any one of claims 1 to 4.

Images