A method and system for diagnosing a breathing valve fault based on pressure residual propagation matching

Abstract

The present application relates to the technical field of fault diagnosis and health management, and particularly relates to a breathing valve fault diagnosis method and system based on pressure residual error propagation matching. The method comprises: acquiring topological information and operation data of a gas phase pressure monitoring system, and constructing a basic graph structure with pressure monitoring nodes and breathing valve nodes as graph nodes, and with valve association relationships and gas phase coupling relationships as edges; performing working condition gate update on the basic graph structure according to the operation data to generate a working condition gate space-time graph; determining pressure residual error information of each pressure monitoring node and generating a pressure residual error graph sequence; determining residual error propagation responses with each breathing valve node as a candidate abnormal source; matching the residual error propagation responses with the pressure residual error graph sequence to determine abnormal breathing valve nodes and fault types, and generating a breathing valve fault diagnosis result. The present application can improve the accuracy of breathing valve fault positioning and type identification under gas phase coupling conditions.

Description

Technical Field

[0001] This invention relates to the field of fault diagnosis and health management technology, specifically to a method and system for diagnosing breathing valve faults based on pressure residual propagation matching. Background Technology

[0002] Breather valves are crucial safety accessories for maintaining pressure balance in industrial gas-phase pressure systems. During material input, output, temperature changes, and media evaporation, breather valves limit pressure changes in the pressurized area by venting or drawing in gas. With increased operating time, factors such as valve disc adhesion, sealing surface contamination, spring performance degradation, flame arrestor blockage, and mechanical wear can cause delayed opening, obstructed venting, obstructed intake, incomplete closure, or abnormal reseating. These faults typically do not manifest as obvious machine shutdowns in the early stages, but rather as deviations in pressure response speed, recovery process, or stable pressure, representing typical diagnostic problems in equipment fault detection and health management.

[0003] Existing gas phase pressure systems primarily utilize pressure transmitters, level gauges, thermometers, and valve position signals to collect operational data, and employ methods such as fixed pressure threshold alarms, trend monitoring, periodic inspections, and manual patrols to determine the status of the breather valve. While these methods can detect significant overpressure, negative pressure, and complete failure, they have limited ability to identify performance degradation and intermittent anomalies that have not yet reached alarm limits, and it is difficult to accurately distinguish fault types based on pressure changes.

[0004] Especially in scenarios where multiple pressure monitoring locations share a gas phase balance pipe, exhaust gas recovery pipeline, or venting pipeline, pressure disturbances caused by material input / output, temperature changes, and exhaust gas extraction can be transmitted between different pressure monitoring locations. A pressure anomaly at one location may be caused by a malfunction of its associated breather valve, or it may be affected by changes in operating conditions at other locations or pressure fluctuations in shared pipelines. Existing monitoring methods based on single-point pressure or local parameters are insufficient to reflect dynamic gas phase coupling relationships, easily leading to confusion between normal disturbances and fault responses, inaccurate identification of anomaly sources, and delayed fault type identification.

[0005] Therefore, how to eliminate interference from normal operating conditions under gas-phase coupling operation conditions, accurately locate abnormal breathing valves and identify their fault types are technical problems that urgently need to be solved in breathing valve fault detection and health management. Summary of the Invention

[0006] To address the shortcomings of existing technologies, the present invention aims to provide a method and system for diagnosing breathing valve faults based on pressure residual propagation matching. By constructing a basic graph structure of a gas phase pressure monitoring system, generating a working condition gating spatiotemporal graph by combining operational data, and matching the pressure residual graph sequence with the residual propagation response of each candidate breathing valve, abnormal breathing valve nodes and their fault types are determined, thereby solving the problems of confusion between normal disturbances and fault responses and difficulty in locating the source of anomalies under gas phase coupling conditions.

[0007] To achieve the above objectives, the present invention provides the following technical solution:

[0008] A method for diagnosing breathing valve faults based on pressure residual propagation matching, the method comprising:

[0009] Obtain the topology information and operating data of the gas phase pressure monitoring system, and construct a basic graph structure based on the topology information, with pressure monitoring nodes and breathing valve nodes as graph nodes and valve association relationships and gas phase coupling relationships as edges;

[0010] Based on the operational data, the basic graph structure is updated with operational condition gating to generate an operational condition gating spatiotemporal graph.

[0011] Based on the operating data, the pressure residual information of each pressure monitoring node is determined, and the pressure residual information is written into the corresponding pressure monitoring node in the working condition gating spatiotemporal diagram to generate a pressure residual diagram sequence.

[0012] Each breathing valve node is used as a candidate anomaly source, and the residual propagation response corresponding to each candidate anomaly source is determined based on the working condition gating spatiotemporal diagram.

[0013] Based on the matching results between the residual propagation response and the pressure residual diagram sequence, abnormal breathing valve nodes and their fault types are determined, and breathing valve fault diagnosis results are generated.

[0014] Based on the aforementioned topology information, a basic graph structure is constructed, with pressure monitoring nodes and breathing valve nodes as graph nodes and valve association relationships and gas phase coupling relationships as edges, including:

[0015] Based on the topology information, determine each pressure monitoring node, the valve association between each pressure monitoring node and each breathing valve, and the gas phase coupling relationship between each pressure monitoring node through the gas phase pipeline.

[0016] Based on the valve association relationship, a corresponding breathing valve node is set for each breathing valve;

[0017] A valve-associated edge is formed between pressure monitoring nodes with valve association and breather valve nodes, and a gas-phase coupling edge is formed between pressure monitoring nodes with gas-phase coupling.

[0018] The basic graph structure is constructed based on the pressure monitoring node, the breathing valve node, the valve associated edge, and the gas phase coupling edge.

[0019] Based on the operational data, the basic graph structure is updated using operational condition gating to generate an operational condition gating spatiotemporal graph, including:

[0020] Multiple monitoring cycles are determined based on the collection time of the aforementioned operational data;

[0021] For each monitoring cycle, the current operating status of each edge in the basic graph structure is determined based on the operating data within that monitoring cycle;

[0022] Based on the current operating condition, the effective connectivity state, pressure disturbance propagation direction, and pressure disturbance propagation weight of each edge in the basic graph structure are set to obtain the operating condition gating graph for the corresponding monitoring period.

[0023] Organize the gating diagrams for each operating condition according to the time sequence of the multiple monitoring cycles, and generate a spatiotemporal diagram of the operating condition gating.

[0024] Based on the current operating condition, the effective connectivity state, pressure disturbance propagation direction, and pressure disturbance propagation weight of each edge in the basic graph structure are set, including:

[0025] Based on the current working condition and the connection status of each edge, determine whether the corresponding edge is a valid connection edge in the current monitoring period;

[0026] For the effective connected edge, the pressure disturbance propagation direction of the effective connected edge is determined according to the pressure relationship between the nodes on both sides of the edge, the edge type, and the breathing valve direction information.

[0027] Based on the propagation direction of the pressure disturbance and the corresponding edge's operational data, determine the pressure disturbance propagation weight of the effective connected edge;

[0028] Suppress propagation states are set for non-effective connected edges, and the effective connected states, pressure disturbance propagation directions, and pressure disturbance propagation weights corresponding to each edge are written into the basic graph structure to obtain the working condition gating graph for the corresponding monitoring period.

[0029] Based on the operational data, the pressure residual information of each pressure monitoring node is determined, and the pressure residual information is written into the corresponding pressure monitoring node in the operating condition gating spatiotemporal diagram to generate a pressure residual diagram sequence, including:

[0030] Multiple residual calculation times are determined based on the data collection time of the operation;

[0031] For each residual calculation time, the pressure residual information of each pressure monitoring node is determined based on the operating data corresponding to that residual calculation time;

[0032] The pressure residual information of each pressure monitoring node is assigned to the pressure monitoring node at the corresponding time position in the working condition gating spatiotemporal diagram to obtain the pressure residual diagram at the corresponding residual calculation time.

[0033] The pressure residual diagrams are organized according to the time sequence of the multiple residual calculation times to generate the pressure residual diagram sequence.

[0034] For each residual calculation time, the pressure residual information of each pressure monitoring node is determined based on the operating data corresponding to that residual calculation time, including:

[0035] For any pressure monitoring node, the measured pressure change of the pressure monitoring node is determined based on the measured pressure data within the time window corresponding to the current residual calculation time.

[0036] Based on the operating data within the time window corresponding to the current residual calculation time, the predicted pressure change of the pressure monitoring node when the associated breathing valve is in a normal response state is determined;

[0037] Based on the deviation between the measured pressure change and the predicted pressure change, the pressure residual information of the pressure monitoring node at the current residual calculation time is determined.

[0038] The step of assigning the pressure residual information of each pressure monitoring node to the pressure monitoring node at the corresponding time position in the working condition gating spatiotemporal diagram to obtain the pressure residual diagram at the corresponding residual calculation time includes:

[0039] Extract the working condition gating diagram corresponding to the current residual calculation time from the working condition gating spatiotemporal diagram;

[0040] Write the pressure residual information of each pressure monitoring node at the current residual calculation time as the node residual value into the corresponding pressure monitoring node.

[0041] The effective connectivity state, pressure disturbance propagation direction, and pressure disturbance propagation weight of each edge in the working condition gating diagram are retained, and the working condition gating diagram after writing the node residual values ​​is determined as the pressure residual diagram corresponding to the current residual calculation time.

[0042] The step of using each breathing valve node as a candidate anomaly source and determining the residual propagation response corresponding to each candidate anomaly source based on the working condition gating spatiotemporal diagram includes:

[0043] For any candidate anomaly source, the breathing valve node corresponding to the candidate anomaly source is taken as the propagation starting point in the working condition gating spatiotemporal diagram;

[0044] Based on the aforementioned working condition gating spatiotemporal diagram, the propagation paths from the propagation starting point to each pressure monitoring node and the propagation intensity of the paths are determined at each time location.

[0045] Based on the propagable paths and propagation intensity corresponding to each time location, the residual propagation contribution of the candidate anomaly source to each pressure monitoring node at each time location is determined, and then organized in chronological order as the residual propagation response corresponding to the candidate anomaly source.

[0046] Based on the matching results between the residual propagation response and the pressure residual map sequence, abnormal breathing valve nodes and their fault types are determined, and breathing valve fault diagnosis results are generated, including:

[0047] Align the residual propagation response corresponding to any candidate anomaly source with the pressure residual information of the same pressure monitoring node at the same time position in the pressure residual map sequence;

[0048] Based on the alignment results, the consistency between the residual propagation response and the pressure residual information in terms of residual polarity and residual amplitude is compared to obtain the response consistency value corresponding to the candidate anomaly source.

[0049] The breathing valve node corresponding to the candidate anomaly source whose response consistency value meets the preset anomaly judgment condition is determined as the abnormal breathing valve node.

[0050] Based on the correspondence between the residual propagation response corresponding to the abnormal breathing valve node and the pressure residual map sequence in terms of time persistence, residual polarity, and node distribution, a preset fault feature rule is matched to determine the fault type of the abnormal breathing valve node and generate the breathing valve fault diagnosis result.

[0051] A breathing valve fault diagnosis system based on pressure residual propagation matching is provided to implement the aforementioned breathing valve fault diagnosis method based on pressure residual propagation matching. The system includes:

[0052] The graph structure construction module is used to acquire the topology information and operating data of the gas phase pressure monitoring system, and construct a basic graph structure based on the topology information, with pressure monitoring nodes and breathing valve nodes as graph nodes and valve association relationships and gas phase coupling relationships as edges;

[0053] The gating graph generation module is used to update the basic graph structure with working condition gating based on the operating data and generate a working condition gating spatiotemporal graph.

[0054] The residual map generation module is used to determine the pressure residual information of each pressure monitoring node based on the operating data, and write the pressure residual information into the corresponding pressure monitoring node in the working condition gated spatiotemporal map to generate a pressure residual map sequence.

[0055] The propagation response module is used to take each breathing valve node as a candidate anomaly source and determine the residual propagation response corresponding to each candidate anomaly source based on the working condition gating spatiotemporal diagram.

[0056] The diagnostic result generation module is used to determine the abnormal breathing valve node and its fault type based on the matching result of the residual propagation response and the pressure residual diagram sequence, and generate the breathing valve fault diagnosis result.

[0057] Compared with the prior art, the beneficial effects of the present invention are:

[0058] By abstracting pressure monitoring nodes, breathing valve nodes, and gas-phase coupling pipelines into a basic graph structure, and combining operational data to perform condition gating updates on the effective connectivity status of each side, the direction of pressure disturbance propagation, and the weight of pressure disturbance propagation, the monitoring process can reflect the current real gas-phase propagation conditions, avoiding false alarms and mislocation of anomaly sources caused by fixed topology or single measuring point pressure threshold judgment.

[0059] By extracting pressure residual information based on the predicted pressure change under the normal response state of the breather valve, and matching the pressure residual map sequence with the residual propagation response of each candidate breather valve, abnormal propagation characteristics can be identified from normal operating disturbances such as material feeding and discharging, temperature changes and exhaust gas recovery, thereby improving the accuracy of abnormal breather valve positioning and identification of operating states such as delayed opening, incomplete closing and obstructed exhaust. Attached Figure Description

[0060] Other features, objects, and advantages of the invention will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings:

[0061] Figure 1 This is a schematic diagram of an application scenario for breather valve fault diagnosis provided in an embodiment of the present invention;

[0062] Figure 2 A schematic flowchart of a breather valve fault diagnosis method based on pressure residual propagation matching provided in an embodiment of the present invention;

[0063] Figure 3 This is a schematic diagram of the basic structure of the gas phase pressure monitoring system provided in an embodiment of the present invention;

[0064] Figure 4 This is a schematic diagram of a breather valve fault diagnosis system based on pressure residual propagation matching, provided as an embodiment of the present invention. Detailed Implementation

[0065] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0066] Example 1

[0067] Please see Figure 1This invention provides a breather valve fault diagnosis method based on pressure residual propagation matching, applicable to tank farm operation scenarios where multiple tanks are connected by gas phase pipelines, exhaust gas recovery pipelines, or venting pipelines, forming a gas phase coupling relationship. In this embodiment, each tank serves as a specific implementation entity of a pressure monitoring node and corresponds to one pressure monitoring node. The measured pressure, predicted pressure, and pressure residual information of the tank serve as the node information of the corresponding pressure monitoring node. The pressure monitoring node corresponding to the tank will be referred to as the tank node. The tank farm contains multiple tanks, each equipped with a breather valve to maintain gas phase pressure balance within the tank. Operating data of the tank farm is acquired through pressure sensors, level sensors, temperature sensors, valve position acquisition units, and exhaust gas recovery operation acquisition units. Figure 1 In the application scenario shown, the pressure change of a single storage tank is not only affected by its own feeding, discharging, temperature changes and material volatilization, but may also be affected by gas phase disturbances of adjacent storage tanks, pressure changes in shared gas phase pipelines and changes in the operating status of the tail gas recovery system. Therefore, there is no simple one-to-one correspondence between abnormal storage tank pressure and abnormal operation of the breather valve.

[0068] In actual tank farm operation, breather valves are typically in a passive response state, and their opening, closing, reseating, and sealing status are indirectly reflected by changes in tank pressure. For single-tank independent operation scenarios, operators can make a preliminary judgment on whether the breather valve is abnormal based on pressure thresholds, pressure change trends, or on-site inspection results. However, in scenarios where multiple tanks share a gas phase system, the gas phase space has relationships that change with operating conditions, such as connectivity, isolation, unidirectional discharge, and recovery / suction. The same pressure fluctuation may propagate to multiple tank nodes along different gas phase paths, causing multiple tanks to experience pressure increases, decreases, or slow pressure recovery within a similar timeframe. In this case, if judgment is still based solely on whether the pressure of a single tank exceeds the limit, it is easy to misjudge normal operating disturbances as breather valve malfunctions, or to incorrectly attribute pressure disturbances caused by other tanks, shared pipelines, and exhaust gas recovery systems to the breather valve of this tank.

[0069] When diagnosing faults in storage tank breather valves, a common technical challenge for those skilled in the art is that breather valve malfunctions often lack directly measurable internal state quantities, while tank pressure is simultaneously affected by material receiving and dispatching, temperature changes, gas-air connection status, and the operating status of external recovery systems. In particular, when the breather valve exhibits delayed opening, incomplete closing, obstructed exhaust, obstructed intake, or abnormal reseating, the resulting pressure changes may not manifest as immediate exceedances, but rather as inconsistencies between the pressure change amplitude and normal predictions, abnormal pressure recovery processes, persistent residual pressure deviations, or certain propagation distribution characteristics along the gas-phase coupling path. Therefore, the key to breather valve fault diagnosis lies not in simply identifying whether the pressure is abnormal at a particular moment, but in separating the pressure residual propagation characteristics corresponding to candidate breather valve malfunctions from complex operating condition disturbances.

[0070] Based on the above understanding, this embodiment does not treat each storage tank as an independent monitoring object, nor does it use a fixed pressure threshold as the sole anomaly criterion. Instead, it understands the gas phase system of the tank area as a pressure disturbance propagation network composed of storage tanks, breather valves, and gas phase coupling pipelines. In this network, storage tanks correspond to pressure monitoring nodes, used to carry pressure response information; breather valves correspond to breather valve nodes, which can serve as candidate source nodes that may introduce abnormal disturbances. The connection relationship between storage tanks and breather valves, as well as the gas connection relationship between storage tanks, can serve as the path basis for pressure disturbance propagation. Since the connectivity, propagation direction, and propagation intensity of the gas phase system will change with valve status, pressure relationship, exhaust gas recovery operation status, and current production conditions, this embodiment further combines static topology relationships with real-time operating data, so that the basic graph structure forms a condition gating graph adapted to the current operating conditions in different monitoring cycles, and constitutes a condition gating spatiotemporal graph in chronological order.

[0071] It should be noted that the method described in this embodiment is not a simple extension of the single tank pressure alarm logic, but rather a state identification based on the correspondence between "anomaly source - propagation path - node residual response". That is, when the breather valve is in a normal response state, a reasonable expectation of the pressure change of each tank can be obtained based on the current operating data; when there is a deviation between the measured pressure change and this reasonable expectation, this deviation can be regarded as pressure residual information. If a breather valve experiences an anomaly such as delayed opening, incomplete closure, or obstructed exhaust, its impact will usually propagate along the currently effective gas connection path, forming a residual distribution with specific polarity, amplitude, and time duration at different tank nodes. By treating each breather valve as a candidate anomaly source, extrapolating its residual propagation response in the operating condition gating spatiotemporal diagram, and matching it with the actual pressure residual diagram sequence, the abnormal breather valve and its operating state that better match the actual residual distribution can be determined from multiple possible disturbance sources.

[0072] Therefore, this embodiment is applicable to tank farms characterized by multiple tanks operating in parallel, sharing gas phase pipelines, centralized exhaust gas recovery, shared venting pipelines, or local gas-phase interconnection states that switch according to operating conditions. Application scenarios can include liquid chemical raw material tank farms, volatile organic compound tank farms, oil tank farms, and other industrial tank systems that require breather valves to maintain gas phase pressure balance within the tanks. A common characteristic of these application scenarios is that pressure changes caused by breather valve malfunctions may be masked by normal production disturbances or propagate to non-faulty tanks through gas-phase coupling, making malfunction localization difficult. This embodiment maps the structural connectivity, real-time operating conditions, and pressure residual information of the tank farm onto a spatiotemporal diagram, providing a calculable and traceable implementation basis for subsequent breather valve malfunction localization and operating status identification.

[0073] Please see Figure 2 The present invention provides an embodiment of a method for diagnosing breathing valve faults based on pressure residual propagation matching, the specific steps of which are as follows:

[0074] S1: Obtain the topology information and operation data of the gas phase system in the tank area, and construct a basic graph structure based on the topology information, with the tanks and breather valves as nodes and the valve association relationship and gas phase coupling relationship as edges.

[0075] In this embodiment, the topology information can be derived from tank area pipeline layout data, process flow diagrams, equipment ledgers, breather valve installation location records, and on-site valve connection relationship records. The topology information is used at least to characterize the direct relationship between each tank and its corresponding breather valve, and whether different tanks are connected via gas phase balance pipes, tail gas recovery pipelines, vent branches, or shared manifolds. Operating data can be acquired based on existing on-site data collection conditions, such as tank pressure, liquid level, temperature, feed status, discharge status, gas phase pipeline valve status, tail gas recovery pressure, tail gas recovery flow rate, and breather valve opening / closing related signals. Those skilled in the art will understand that the specific types of operating data do not need to be completely consistent; they only need to be sufficient to determine the current gas phase connection status, pressure change status, and breather valve response status. This embodiment does not impose further limitations in this regard.

[0076] When constructing the basic graph structure, storage tanks are abstracted as tank nodes that reflect pressure response, and breather valves are abstracted as breather valve nodes that may introduce abnormal disturbances. Valve association edges are formed between tank nodes and their corresponding breather valve nodes to represent the direct release, intake, or sealing effect of the breather valve on the gas phase pressure of the tank. Gas phase coupling edges are formed between tank nodes with gas phase interconnections to represent the potential transmission of pressure disturbances between tanks. This approach transforms the complex pipeline connections, valve associations, and parallel tank relationships in the field into a computable graph structure, enabling subsequent anomaly analysis to move beyond simply judging the pressure threshold of a single tank and instead analyze the source and propagation range of breather valve anomalies within the overall gas phase coupling relationship of the tank area.

[0077] S2: Update the basic graph structure according to the operating data to generate a working condition gating spatiotemporal graph.

[0078] In this embodiment, the basic diagram reflects the static relationship of the tank area at the equipment connection level. However, in actual operation, gas phase pressure disturbances do not always propagate along all connection paths. For example, some gas phase branches may temporarily fail due to the closure of isolation valves, the suction status of the tail gas recovery device may change the direction of gas phase disturbances, and the tank inlet and outlet processes may also cause changes in the pressure difference relationship between nodes. Therefore, using only a fixed topology for anomaly detection can easily lead to misidentifying pipelines that do not meet the propagation conditions under the current operating conditions as valid propagation paths, thus affecting the anomaly source location results.

[0079] Based on this, in this embodiment, multiple monitoring cycles are divided according to the data collection time. Within each monitoring cycle, the edges in the basic graph structure are dynamically updated by combining valve status, pipeline connectivity, tank pressure relationship, exhaust gas recovery operation status, and breather valve direction information. For edges that currently have gas connection conditions, they are set to an effective connectivity state, and the pressure disturbance propagation direction is determined based on the pressure relationship between the nodes on both sides, the edge connection type, and the exhaust or intake direction of the breather valve. For edges that currently do not have connectivity conditions, they are set to a suppressed propagation state, so that they do not participate in the abnormal propagation analysis within this monitoring cycle. Furthermore, the pressure disturbance propagation weight can be determined based on conventional operating parameters such as pipeline pressure difference, gas flow rate, valve opening, recovery suction intensity, or historical pressure response speed, to characterize the degree of influence of the path on pressure disturbance transmission under the current operating conditions.

[0080] Therefore, each monitoring cycle corresponds to a condition gating chart adapted to the current operating state. After organizing multiple condition gating charts in chronological order, a condition gating spatiotemporal diagram is formed. This spatiotemporal diagram not only preserves the structural basis of the gas phase connection in the tank area, but also reflects the changes in connectivity, propagation direction, and propagation intensity in different monitoring cycles. This helps to distinguish between pressure fluctuations caused by normal operating condition switching and residual propagation effects caused by abnormal breather valves.

[0081] S3: Determine the pressure residual information of each storage tank based on the operating data, and write the pressure residual information into the corresponding storage tank node in the operating condition gating spatiotemporal diagram to generate a pressure residual diagram sequence.

[0082] In this embodiment, the tank pressure is affected by various factors such as feeding, discharging, temperature changes, material volatilization, exhaust gas recovery and extraction, and normal breather valve operation. Therefore, a rise or fall in pressure does not necessarily indicate a breather valve malfunction. To avoid directly using normal operating fluctuations as a basis for anomaly assessment, this embodiment does not directly use measured pressure values ​​for judgment. Instead, it first determines the pressure residual information of each tank under the current operating conditions. The pressure residual information can be understood as the deviation between the measured pressure change and the predicted pressure change under normal response conditions. It is used to characterize the portion of pressure change that cannot be explained by normal operating conditions after considering conventional operating factors.

[0083] Specifically, a corresponding time window can be selected for each residual calculation moment, and the actual pressure change trend of the storage tank can be determined based on the measured pressure data within that time window. Simultaneously, by combining the pressure regulation patterns under conditions of liquid level change, temperature change, inlet / outlet status, tail gas recovery status, and normal breather valve response, the predicted pressure change that the storage tank should exhibit under normal operating conditions can be determined. Predicted pressure changes can be obtained through mechanistic models, empirical models, historical data fitting models, or field calibration curves. Those skilled in the art can choose the appropriate model based on the storage tank type, medium characteristics, and the completeness of field data, as long as a reasonable pressure change benchmark under the current operating conditions can be provided.

[0084] After obtaining the pressure residual information, the pressure residual values ​​of each tank are written into the tank node at the corresponding time position in the operating condition gating spatiotemporal diagram, resulting in a pressure residual map. Organizing multiple pressure residual maps in chronological order forms a pressure residual map sequence. Through this processing method, the anomaly analysis object is transformed from the original pressure curve into a residual distribution with topological location and time order, which can more accurately reflect the response characteristics of breather valve anomalies at different tank nodes and reduce the interference of feed / discharge, temperature disturbances, and tail gas recovery fluctuations on anomaly judgment.

[0085] S4: Take each breathing valve node as a candidate anomaly source, and determine the residual propagation response corresponding to each candidate anomaly source based on the working condition gating spatiotemporal diagram.

[0086] In this embodiment, when multiple storage tanks share a gas phase pipeline, the pressure residual observed at a certain storage tank is not necessarily caused by the tank's own breather valve, but may also originate from an indirect response formed by the propagation of an anomaly from other breather valves through the gas phase coupling path. Therefore, this embodiment treats each breather valve node as a candidate anomaly source, rather than assuming in advance that the anomaly must occur at the breather valve corresponding to the storage tank with the largest pressure residual.

[0087] For any candidate anomaly source, the corresponding breathing valve node is used as the propagation starting point, and response simulation is performed along the effective propagation path within the current monitoring period in the working condition-gated spatiotemporal diagram. During the simulation, edges in the suppressed propagation state do not participate in the path calculation, while effective connected edges are used to determine the tank nodes that the pressure residual may reach and the corresponding propagation intensity based on their propagation direction and propagation weight. If a propagation path crosses multiple tank nodes or multiple gas-phase coupling edges, the contribution of the path to the residual of the target tank node can be determined based on the propagation weight, directional consistency, and temporal sequence of each edge on the path.

[0088] Through the above processing, the residual propagation response that each candidate breather valve may generate for each tank node at different time locations can be obtained. This residual propagation response does not represent the actual measured pressure residual, but rather the residual distribution pattern that should appear according to the current operating condition gating relationship when the breather valve is assumed to be the anomaly source. Therefore, it can provide a basis for comparison with the actual pressure residual map sequence, transforming anomaly localization from finding pressure anomaly points to finding the anomaly source that best matches the actual residual propagation pattern.

[0089] S5: Based on the matching results between the residual propagation response and the pressure residual diagram sequence, determine the abnormal breathing valve node and its fault type, and generate the breathing valve fault diagnosis result.

[0090] In this embodiment, the residual propagation response corresponding to each candidate anomaly source is aligned with the pressure residual map sequence in terms of time and nodes, so that the projected response at the same time location and the same tank node corresponds to the actual pressure residual information. During matching, the consistency of residual polarity, the similarity of residual amplitude changes, the coincidence of residual occurrence times, and the conformity of residual distribution across multiple tank nodes to the propagation path relationship can be compared. If the residual propagation response corresponding to a candidate breather valve has high consistency with the actual pressure residual map sequence in the above aspects, it indicates that the candidate breather valve is more likely to be the source of the current pressure anomaly propagation phenomenon.

[0091] After identifying the abnormal breather valve node, the fault type can be further determined based on the time persistence, residual polarity, and node distribution relationship between the residual propagation response and the pressure residual diagram sequence. For example, when the pressure residual corresponding to the exhaust direction is consistently positive and diffuses outward along the effective gas phase path, it can be used as a criterion for diagnosing exhaust obstruction or delayed opening; when the breather valve-associated storage tank exhibits a long-term, inexplicable pressure release trend, and the residual persists after reseating, it can be used as a criterion for diagnosing incomplete closure or abnormal reseating; when the negative pressure direction residual persists under intake conditions, it can be used as a criterion for diagnosing intake obstruction or delayed intake response. Specific fault characteristic rules can be pre-configured based on the breather valve type, set pressure, storage tank medium, and on-site operation and maintenance standards.

[0092] Through the aforementioned matching and fault type determination process, this embodiment can combine the abnormal location of the breather valve with the current operating condition propagation path in an operating environment where multiple storage tanks are coupled in the gas phase and pressure disturbances are mutually transmitted. This avoids false alarms and mislocations caused by relying solely on single tank pressure exceeding limits. Simultaneously, pressure residual information can mitigate the impact of normal production conditions on pressure changes, and the propagation response of candidate anomaly sources can reflect the spatial propagation law of anomaly effects. The combination of these two aspects improves the reliability of breather valve fault location and fault type identification under shared gas phase conditions.

[0093] Please see Figure 3 The present invention provides a schematic diagram of the basic structure of the gas phase system in the tank area, wherein T1, T2, and T3 represent tank nodes, V1, V2, and V3 represent breather valve nodes associated with the corresponding tanks, solid lines represent valve association edges between tank nodes and breather valve nodes, and dashed lines represent gas phase coupling edges formed between tank nodes due to shared gas phase pipelines. Figure 3 The left side shows the relationship between multiple storage tanks connected by gas phase pipes in the field. Figure 3 The right side shows the basic graph structure obtained from the transformation of on-site object relationships. Based on the topology information, a basic graph structure is constructed with storage tanks and breather valves as nodes, and valve association relationships and gas-phase coupling relationships as edges, including:

[0094] S1.1: Determine the valve association between each storage tank and each breather valve, as well as the gas phase coupling relationship between each storage tank through the gas phase pipeline, based on the topology information.

[0095] S1.2: Based on the valve association relationship, set corresponding breather valve nodes for the breather valves associated with each storage tank.

[0096] S1.3: Form valve association edges between tank nodes with valve association relationships and breather valve nodes, and form gas phase coupling edges between tank nodes with gas phase coupling relationships.

[0097] S1.4: Construct the basic graph structure based on the tank node, the breather valve node, the valve associated edge, and the gas phase coupling edge.

[0098] Topology information can be obtained from tank farm process flow diagrams, pipeline layout diagrams, equipment ledgers, breather valve installation records, gas phase pipeline valve lists, and field instrument tag tables. To enable calculation and processing of on-site object relationships, the topology information should first be organized into a structured topology table. This structured topology table should at least include tank identifiers, breather valve identifiers, breather valve installation locations, gas phase pipeline start and end points, shared pipe section identifiers, isolation valve identifiers, check valve direction, exhaust gas recovery access location, and vent branch access location. If a DCS, PLC, or historical database is available on-site, the equipment tag numbers and real-time status corresponding to the above objects can also be read from it. If some connection relationships only exist in paper drawings or equipment ledgers, a topology table with the same fields can be generated through manual input or drawing parsing. This embodiment does not limit the source of topology information, as long as the connection relationships between tanks, breather valves, and gas phase pipelines can be determined.

[0099] Preferably, when organizing topology information, the gas phase space of the storage tank is used as the physical benchmark for determining connection relationships. For breather valves installed on the top of the storage tank, on the tank top gas phase branch pipe, or on a branch pipe directly connected to the gas phase space of the storage tank, as long as there is no normally closed isolation component between the breather valve and the corresponding storage tank, or if there is a maintenance isolation component but it remains connected under normal operating conditions, a valve association relationship can be identified between the storage tank and the breather valve. For cases where a storage tank is equipped with multiple breather valves, an independent breather valve node can be set for each breather valve, and these multiple breather valve nodes can be connected to the same storage tank node. For cases where a breather valve acts simultaneously on the gas phase spaces of multiple storage tanks through a manifold, the breather valve node can be connected to the corresponding multiple storage tank nodes based on the connectivity between the manifold and the gas phase spaces of each storage tank. This approach is compatible with various field layouts, including single-tank single-valve, single-tank multi-valve, and multi-tank shared valve groups.

[0100] The determination of gas-phase coupling is based on whether gas-phase disturbances can be transmitted between storage tanks. If two storage tanks have a gas-phase balancing pipe, a shared tail gas recovery manifold, a shared vent manifold, or other pipeline paths that can cause gas-phase pressure changes to affect each other, then a gas-phase coupling relationship can be identified between them. For cases where multiple storage tanks are connected to the same gas-phase manifold, storage tank nodes connected to the same manifold and potentially affecting each other in terms of process can be paired to form gas-phase coupling edges. If there are multiple parallel gas-phase paths between two storage tanks, they can be merged into one gas-phase coupling edge, and the corresponding multiple pipeline path identifiers can be recorded in the attributes of this edge. If a check valve, pressure regulating valve, or tail gas recovery suction device is installed in a certain gas-phase path, its directional restrictions and operating status should not be directly deleted from the basic diagram structure, but rather recorded as edge attributes for later determination based on the current operating conditions whether the edge allows pressure disturbance propagation and its propagation direction.

[0101] For example, Figure 3 In this diagram, T1, T2, and T3 represent three tank nodes, and V1, V2, and V3 represent breather valve nodes associated with their respective tanks. When V1 is installed on the gas phase branch at the top of T1 and used to regulate the pressure inside T1, a valve-associated edge is formed between T1 and V1. When T1 and T2 are connected via a gas phase pipeline, and this pipeline may transmit pressure disturbances under operating conditions, a gas phase coupling edge is formed between T1 and T2. If T2 and T3 are both connected to the tail gas recovery manifold, even if there is no direct pipe section between T2 and T3, a gas phase coupling edge can be formed between them based on the gas phase influence relationship formed by the shared manifold. Thus, the arrangement of tanks, pipe sections, manifolds, and valves in the field is transformed into a basic diagram structure composed of tank nodes, breather valve nodes, and edges.

[0102] The attributes of a tank node may include tank identifier, tank volume, type of stored medium, pressure measuring point number, level measuring point number, temperature measuring point number, normal pressure range, list of associated breather valves, and list of connected gas phase pipelines. The attributes of a breather valve node may include breather valve identifier, installed tank identifier, exhaust set pressure, intake set pressure, rated ventilation capacity, valve type, direction of action, and maintenance status. The attributes of a valve-associated edge may include edge identifier, associated tank node, associated breather valve node, breather valve direction of action, direct connection status, and whether a maintenance isolation component exists. The attributes of a gas phase coupling edge may include edge identifier, both end tank nodes, pipeline path identifier, whether it passes through a shared manifold, isolation valve identifiers along the route, check direction, pipe diameter, pipe length, and exhaust gas recovery connection relationship. Not all of the above attributes are required to participate in the calculation, but at least the necessary fields that can distinguish between valve-associated edges and gas phase coupling edges, identify endpoint objects, and support operational condition gating judgments should be retained.

[0103] The basic graph structure can be stored using conventional data structures such as adjacency lists, adjacency matrices, or attribute graphs. When using an adjacency list, each tank node can record its connected breather valve nodes and adjacent tank nodes, and each edge can record its type and attributes. When using an adjacency matrix, matrix elements can represent whether a connection exists between two nodes, and an attached attribute table records edge types, pipeline paths, and valve constraints. When using an attribute graph, tank nodes, breather valve nodes, valve-associated edges, and gas-phase coupling edges can be directly stored as graph objects. All of the above storage methods are conventional graph data representation methods in this field. This embodiment does not limit the specific implementation form, as long as it supports traversal, querying, and attribute reading by nodes and edges.

[0104] The physical meaning of the basic graph structure lies in fixing the relationships between objects where pressure disturbances may occur, act, and propagate within the tank area. Tank nodes carry pressure residual information, breather valve nodes represent candidate locations that may cause abnormal pressure disturbances, valve-associated edges represent the direct impact of the breather valve on the corresponding tank's gas phase pressure, and gas-phase coupling edges represent the indirect impact between tanks due to gas phase connections. Through this structured representation, pressure disturbances directly caused by the breather valve of a given tank can be distinguished from those propagated from adjacent tanks or shared gas phase pipelines during subsequent pressure residual analysis, thus providing a clear topological basis for locating anomaly sources under shared gas phase conditions.

[0105] Based on the operational data, the basic graph structure is updated using operational condition gating to generate an operational condition gating spatiotemporal graph, including:

[0106] S2.1: Determine multiple monitoring cycles according to the collection time of the aforementioned operational data.

[0107] S2.2: For each monitoring cycle, determine the current operating status of each side in the basic graph structure based on the operating data within that monitoring cycle.

[0108] S2.3: Based on the current operating condition, set the effective connectivity state, pressure disturbance propagation direction, and pressure disturbance propagation weight of each side in the basic graph structure to obtain the operating condition gating graph for the corresponding monitoring period.

[0109] S2.4: Organize the gating diagrams of each operating condition according to the time sequence of the multiple monitoring cycles, and generate the spatiotemporal diagram of the operating condition gating.

[0110] In this embodiment, the operational data, after being unified to the same time base according to the collection time, is divided into continuous monitoring cycles. The monitoring cycle is used to define the time range corresponding to one operating condition gating update, and its length is determined according to the tank pressure change rate, pressure sampling cycle, valve status update cycle, and breather valve action response time. For example, the monitoring cycle is 30s, 60s, or 5min. When there are multiple pressure sampling data within the same monitoring cycle, the pressure data is time-aligned, and the representative pressure value corresponding to the monitoring cycle is obtained according to preset statistical rules. When the sampling frequencies of valve status, exhaust gas recovery status, and pressure data are inconsistent, resampling is performed according to a unified timestamp or the most recent valid state is maintained, so that the tank pressure, liquid level, temperature, valve status, exhaust gas recovery status, and breather valve related status within the same monitoring cycle can correspond to each other.

[0111] For each monitoring cycle, the current operating condition is determined by the edges in the basic graph structure. For valve-related edges, the current operating condition includes the breather valve branch connectivity status, maintenance isolation status, associated tank pressure, breather valve exhaust setting pressure, breather valve intake setting pressure, and breather valve direction information. For gas-phase coupling edges, the current operating condition includes the gas-phase path connectivity status, the status of isolation valves along the path, the operating status of the shared manifold, check valve direction, tail gas recovery and suction status, and the pressure relationship between the two end tanks. Thus, each edge generates a corresponding edge operating condition record within the current monitoring cycle. This edge operating condition record is used to characterize whether the edge has the conditions for pressure disturbance propagation, and, if the conditions are met, in what direction and intensity should it participate in the pressure residual propagation analysis.

[0112] After obtaining the current operating condition, operating condition gating attributes are written to each edge in the basic graph structure. These attributes include at least the effective connectivity state, pressure disturbance propagation direction, and pressure disturbance propagation weight. The effective connectivity state indicates whether pressure disturbances are allowed to pass through the edge during the current monitoring period; the pressure disturbance propagation direction indicates the allowed propagation direction of the pressure disturbance on that edge; and the pressure disturbance propagation weight indicates the degree of influence of the edge on pressure disturbance transmission under the current operating condition. This process only updates the edge attributes in the basic graph structure, without deleting tank nodes, breather valve nodes, valve-associated edges, or gas-phase coupling edges themselves, thus allowing the same basic graph structure to exhibit different propagation constraint states in different monitoring periods.

[0113] After setting the gating attributes for each edge condition within the current monitoring period, the corresponding gating chart for that monitoring period is obtained. Each gating chart represents the effective propagation topology of the gas phase system in the tank area within the corresponding time range. Multiple gating charts are arranged in chronological order according to the monitoring period, and a corresponding time position is set for each chart, thus generating a spatiotemporal map of the gating conditions. This spatiotemporal map of the gating conditions allows for the preservation of the basic connectivity relationships of the gas phase system in the tank area while reflecting changes in the gas connection status, pressure disturbance propagation direction, and propagation weight within different monitoring periods. This provides a dynamic topological basis for subsequent pressure residual information mapping and the determination of candidate anomaly source propagation responses.

[0114] Based on the current operating conditions, the effective connectivity state, pressure disturbance propagation direction, and pressure disturbance propagation weight of each edge in the basic graph structure are set, including:

[0115] S2.3.1: Based on the current working condition, determine whether the corresponding edge is a valid connected edge in the current monitoring period.

[0116] S2.3.2: For the effective connected edge, the pressure disturbance propagation direction of the effective connected edge is determined according to the pressure relationship between the nodes on both sides of the corresponding edge, the edge type, and the breather valve direction information.

[0117] S2.3.3: Determine the pressure disturbance propagation weight of the effective connected edge based on the pressure disturbance propagation direction and the corresponding edge's operating data.

[0118] S2.3.4: Set the suppression propagation state for non-effective connected edges, and write the effective connected state, pressure disturbance propagation direction and pressure disturbance propagation weight of each edge into the basic graph structure to obtain the working condition gating graph for the corresponding monitoring period.

[0119] The current operating condition is used to determine whether each edge in the basic graph structure has the conditions for pressure disturbance propagation within the current monitoring period. It assesses the edges in the basic graph structure, rather than directly determining whether the tank or breather valve is abnormal. By setting the effective connectivity state, pressure disturbance propagation direction, and pressure disturbance propagation weight for each edge, the static connection relationships in the basic graph structure can be transformed into dynamic propagation relationships under the current operating condition.

[0120] For valve-associated edges, their validity is determined based on the branch connectivity status between the breather valve and the corresponding tank's gas phase space. When the breather valve branch is connected and there are no maintenance isolation barriers, shutdown markers, or engaged isolation components, the valve-associated edge is determined as a valid connection edge. When the breather valve branch is closed, the breather valve is marked for maintenance and shutdown, or there are engaged isolation components between the branch and the tank's gas phase space, the valve-associated edge is determined as an invalid connection edge. For valves capable of providing continuous opening value feedback, an effective opening threshold can be set as the basis for connection judgment. This effective opening threshold is determined based on the minimum stable venting opening and valve position feedback error in the valve commissioning record, for example, by adding the minimum stable venting opening to the valve position feedback error margin. This avoids misjudging valve slight opening, valve position fluctuations, or feedback errors as valid connections.

[0121] For a gas-phase coupling edge, its validity is determined based on the actual connectivity of the gas-phase path between the two tanks. If the two tanks are actually connected via a gas-phase pipeline, gas-phase balance pipe, or shared manifold, and the valves along the path meet the connectivity requirements, then the gas-phase coupling edge is designated as a valid connection edge. When the critical isolation valve along the path is closed, or the gas-phase path is under maintenance, disconnected, out of service, or blind-plate isolated, the gas-phase coupling edge is determined as an invalid connection edge. The critical isolation valve is determined based on the pipeline layout diagram and valve list, and is a valve or isolation component capable of cutting off the gas-phase connection path between the two tanks. If the same gas-phase coupling edge corresponds to multiple parallel gas-phase paths, then the gas-phase coupling edge is determined as a valid connection edge when at least one path meets the connectivity requirements; when all paths do not meet the connectivity requirements, the gas-phase coupling edge is determined as an invalid connection edge.

[0122] Preferably, when the valve status on-site is a continuous opening value, an effective opening threshold can be set. For example, a valve opening of not less than 10% is considered to meet the basic connectivity conditions, while an opening less than 10% is considered to not meet the effective connectivity conditions. For valves that can only report on / off status, connectivity is directly determined by the open and closed states. If there are conflicts between different data sources, such as the DCS showing a valve open while the maintenance record shows a branch isolation, the safety side can be prioritized for determination, setting the corresponding edge as an ineffective connectivity edge to avoid including actual non-propagating paths in the pressure residual propagation analysis.

[0123] The propagation direction of pressure disturbances is determined based on the edge type. For gas-phase coupled edges, the propagation direction is determined according to the pressure relationship between the two tank nodes. When the pressure difference between the two tanks exceeds the pressure dead zone, the direction of pressure disturbance propagation is taken as the direction from the tank node with the higher pressure value to the tank node with the lower pressure value. The pressure dead zone is determined based on the allowable error of the pressure transmitter and the normal pressure fluctuation amplitude during stable tank operation, for example, 0.1 kPa to 0.5 kPa. This setting can avoid frequent switching of the propagation direction due to instrument noise or normal short-term fluctuations. When the pressure difference between the two tanks does not exceed the pressure dead zone, it is considered that the edge does not have a clear unidirectional propagation trend in the current monitoring period, and it can be set to a bidirectional restricted propagation state or not participate in directional propagation calculations.

[0124] For the valve-associated edge, the propagation direction is determined based on the breather valve direction information and the position of the associated tank pressure relative to the breather valve's set pressure. When the associated tank pressure reaches the exhaust trigger boundary, the propagation direction is set from the tank node to the breather valve node; when the associated tank pressure reaches the intake trigger boundary, the propagation direction is set from the breather valve node to the tank node. The exhaust trigger boundary is determined based on the breather valve's exhaust set pressure and allowable deviation, while the intake trigger boundary is determined based on the breather valve's intake set pressure and allowable deviation. The allowable deviation is derived from the breather valve calibration record or equipment technical parameters and reflects the permissible offset range between the actual opening pressure and the set pressure of the breather valve, thus ensuring that the propagation direction determination conforms to the actual operating characteristics of the breather valve.

[0125] The pressure disturbance propagation weight is determined based on the operational data corresponding to the propagation direction. It characterizes the ability of an effective connected edge to transmit pressure disturbances within the current monitoring period, with a value ranging from 0 to 1. For an effective connected edge, the propagation weight can be determined based on the pressure difference between the two ends, the valve opening, and the edge type. Specifically, the pressure difference between the two ends reflects the driving degree of the pressure disturbance, the valve opening reflects the ventilation capacity of the gas phase channel, and the edge type distinguishes the propagation meaning of valve-associated edges and gas-phase coupled edges. The pressure difference can be determined based on the pressure fluctuation range during stable tank operation, historical pressure response records, and the breather valve's operating pressure setting; the valve opening can be determined based on valve commissioning records or valve flow characteristics. This establishes a mapping relationship between pressure difference, valve opening, and propagation weight, ensuring that the weight setting is based on field data.

[0126] For example, the pressure difference can be divided into a first pressure difference interval, a second pressure difference interval, and a third pressure difference interval, and the valve opening can be divided into a first opening interval, a second opening interval, and a third opening interval, with the propagation weights determined through a mapping table. The weights corresponding to the first pressure difference interval and the first opening interval are 0.2 to 0.3, the weights corresponding to the second pressure difference interval and the second opening interval are 0.4 to 0.6, and the weights corresponding to the third pressure difference interval and the third opening interval are 0.7 to 0.9. The boundaries of these intervals are not arbitrarily set, but are calibrated based on the pressure response amplitude under different combinations of pressure difference and opening in historical operating data. When the pressure response under a certain interval combination can be stably identified within a monitoring period, the corresponding weight is taken as the median or upper limit of the interval; when the pressure response is close to the normal fluctuation range, the corresponding weight is taken as the lower limit of the interval.

[0127] For non-effective connected edges, a propagation suppression state is set, with the effective connectivity state set to 0, the pressure disturbance propagation direction set to non-propagable, and the pressure disturbance propagation weight set to 0. For effective connected edges, the effective connectivity state is set to 1, and the determined pressure disturbance propagation direction and weight are written in. This writing process only updates the edge attributes in the basic graph structure, without deleting the original nodes and edges, allowing the same basic graph structure to form different condition-gated graphs based on actual operating conditions in different monitoring periods. After writing all edge attributes, the condition-gated graph for the corresponding monitoring period is obtained.

[0128] Based on operational data, the pressure residual information of each storage tank is determined, and this pressure residual information is mapped to the corresponding storage tank node in the operating condition gating spatiotemporal diagram to generate a pressure residual diagram sequence, including:

[0129] S3.1: Determine multiple residual calculation times according to the acquisition time of the running data.

[0130] S3.2: For each residual calculation time, determine the pressure residual information of each storage tank based on the operating data corresponding to that residual calculation time.

[0131] S3.3: Assign the pressure residual information of each storage tank to the storage tank node at the corresponding time position in the working condition gating spatiotemporal diagram to obtain the pressure residual diagram at the corresponding residual calculation time.

[0132] S3.4: Organize the pressure residual diagrams according to the time sequence of the multiple residual calculation times to generate the pressure residual diagram sequence.

[0133] The residual calculation time can be uniformly determined based on the data acquisition time. If the tank pressure data is acquired at the second or minute level, the end time of each monitoring cycle can be used as the residual calculation time, or the residual calculation time can be extracted from the continuous operating data according to a preset step size. For example, when the pressure data is acquired every 10 seconds, a residual calculation time can be determined every 60 seconds, and the operating data 3 to 5 minutes before that time can be selected as the calculation window. The residual calculation time should correspond to the time position in the operating condition gating spatiotemporal diagram so that the pressure residual information can be mapped to the tank node at the same time position.

[0134] The pressure residual information is not the measured pressure value of the storage tank itself, but rather a quantity used to characterize the degree to which the actual pressure change deviates from the normal response. The normal response can be determined based on the pressure change pattern of the storage tank when no breather valve malfunction occurs. This pattern can be obtained from historical normal operation data, field calibration data, empirical pressure change curves, or conventional process models. In practical applications, the measured pressure change can be determined based on the measured tank pressure data within the current calculation window. This, combined with changes in liquid level, temperature, feeding status, discharging status, tail gas recovery status, and normal breather valve operation conditions, determines the corresponding predicted pressure change. The deviation between the measured pressure change and the predicted pressure change is thus considered the pressure residual information of the storage tank at the current residual calculation moment.

[0135] To facilitate comparisons between different storage tanks, pressure residual information can include pressure residual values, residual polarity, and residual amplitude. Pressure residual values ​​can be expressed in pressure units, such as kPa; alternatively, they can be normalized to dimensionless values ​​based on the allowable range of normal pressure variation for the storage tank. The normalized value range can be set to -1 to 1. Residual polarity indicates the direction of deviation: a positive residual indicates that the measured pressure change is higher than expected under normal conditions, typically corresponding to insufficient venting, delayed venting, or the introduction of external pressure disturbances; a negative residual indicates that the measured pressure change is lower than expected under normal conditions, typically corresponding to abnormal venting, abnormal suction, or enhanced suction effects. The residual amplitude indicates the degree of deviation; the larger the value, the more difficult it is to explain the current pressure change under normal operating conditions.

[0136] Preferably, a residual dead zone can be set to prevent pressure sensor noise or small normal fluctuations from being mistaken for valid residuals. The residual dead zone can be determined based on the pressure sensor accuracy, the normal fluctuation range of the storage tank, and the sensitivity of the on-site alarm, for example, set to 0.1 kPa to 0.3 kPa; when using normalized pressure residual values, it can also be set to 0.05 to 0.1. When the pressure residual value is within the residual dead zone, the pressure residual value of that storage tank node can be set to 0, indicating that there is no pressure deviation at the current moment; when the pressure residual value exceeds the residual dead zone, its polarity and amplitude are retained to represent the spatial distribution of abnormal pressure response.

[0137] For missing or abnormal sampled data, conventional data quality processing methods can be used. If pressure data is missing for a short period of time, and the missing duration does not exceed one residual calculation step, interpolation using adjacent valid sampled values ​​can be used. If the missing duration is long, or if the pressure sensor has abnormal markers such as exceeding its range, disconnection, or sudden jump, the pressure residual information of the corresponding storage tank at that residual calculation time can be marked as invalid and not included in the residual matching at that time. This processing can avoid false residuals caused by instrument failure interfering with the abnormal judgment of the breather valve.

[0138] When mapping pressure residual information to the operating condition gating spatiotemporal diagram, the operating condition gating diagram corresponding to the current residual calculation time is first selected. Then, based on the correspondence between tank identifiers and tank nodes, the pressure residual values ​​of each tank are written into the corresponding tank node. Breather valve nodes do not directly carry pressure residual values ​​and can be left as null or default values. The effective connectivity states, pressure disturbance propagation directions, and pressure disturbance propagation weights already set in the operating condition gating diagram remain unchanged. The resulting pressure residual diagram simultaneously contains node residual information and edge propagation constraints, enabling it to express "where the pressure deviation occurs" and "what path this deviation might propagate along."

[0139] Multiple pressure residual maps are arranged in chronological order of residual calculation time, forming a pressure residual map sequence. Each frame in this sequence corresponds to a time position, with nodes recording the pressure residual information of each tank at that moment, and edges retaining the operating condition gating attributes at that moment. Through this time-series representation, a single-point pressure anomaly can be transformed into a residual change process with temporal continuity and spatial distribution characteristics, providing a data foundation for determining whether pressure deviation originates from an anomaly in a breather valve and whether it propagates along a gas-phase coupling path.

[0140] For each residual calculation time, the pressure residual information of each storage tank is determined based on the operating data corresponding to that residual calculation time, including:

[0141] S3.2.1: For any storage tank, determine the measured pressure change of the storage tank based on the measured pressure data within the time window corresponding to the current residual calculation time.

[0142] S3.2.2: Based on the operating data within the time window corresponding to the current residual calculation time, determine the predicted pressure change of the storage tank when the breather valve is in normal response state.

[0143] S3.2.3: Based on the deviation between the measured pressure change and the predicted pressure change, determine the pressure residual information of the storage tank at the current residual calculation time.

[0144] For any storage tank, a time window can be selected before the current residual calculation time as the pressure change observation interval. The length of the time window can be determined based on the tank volume, pressure change rate, and sampling frequency, for example, set to 3 min, 5 min, or 10 min. Specifically, the pressure change of the tank per unit time can be statistically analyzed based on historical normal operation data, and the number of effective sampling points to be included in the window can be determined in combination with the sampling frequency. When the pressure change per unit time reaches the rapid pressure change judgment threshold, or the tank volume is lower than the tank area volume classification threshold, a 3 min or 5 min time window can be used; when the pressure change per unit time does not reach the rapid pressure change judgment threshold, or the tank volume reaches the tank area volume classification threshold, a 5 min or 10 min time window can be used. The rapid pressure change judgment threshold is determined based on the historical pressure change rate distribution and the breather valve action response time, and the tank area volume classification threshold is determined based on the tank volume distribution in the tank area equipment ledger. The pressure data within the window is first subjected to outlier removal and filtering, such as removing out-of-range values, disconnected values, and instantaneous peak values. Then, the pressure data used for calculation is obtained by using a moving average, median filtering, or the average of the first and last effective sampling segments.

[0145] The measured pressure change is used to represent the actual pressure change that occurs in the storage tank within the specified time window. Preferably, the average of several effective pressure samples at the beginning of the time window can be taken as the starting pressure, and the average of several effective pressure samples at the end of the time window can be taken as the ending pressure. The measured pressure change is then determined based on the difference between the two. For example, when pressure data is collected every 10 seconds and the time window is 5 minutes, the average effective pressure within 30 seconds after the start of the window and within 30 seconds before the end of the window can be taken as the starting pressure and ending pressure, respectively.

[0146] Predicted pressure changes represent the theoretically or empirically expected pressure changes that a storage tank should experience within the same time window when the breather valve is in normal response mode. This prediction can be determined based on tank operating data and a pre-established normal response baseline. The normal response baseline can be obtained through training with historical normal operating data, or determined using field calibration data, process experience curves, or conventional process models. Input data may include liquid level changes, temperature changes, feed status, discharge status, tail gas recovery status, tank vapor space size, volatilization characteristics of the stored medium, and the breather valve's exhaust and intake set pressures.

[0147] For example, when the storage tank is in the feeding state, the increase in liquid level will compress the gas phase space inside the tank, and the pressure will normally tend to increase. If the pressure approaches the exhaust set pressure, the breather valve should generate an exhaust response to limit the pressure increase. When the storage tank is in the discharging state, the decrease in liquid level will expand the gas phase space inside the tank, and the pressure will normally tend to decrease. If the pressure approaches the intake set pressure, the breather valve should generate an intake response to limit the pressure decrease. When the temperature rises, the gas phase pressure inside the tank usually tends to rise; when the temperature falls, the gas phase pressure inside the tank usually tends to fall. When the tail gas recovery is in the suction state, the tank pressure is also affected by the suction intensity.

[0148] Preferably, the normal response benchmark can be implemented using a condition-specific mapping table. This mapping table can classify the tank status into categories such as feeding, discharging, static, temperature rise, temperature drop, and tail gas recovery / suction, and configure corresponding normal pressure change ranges for different categories. Specifically, the static state can be determined based on the feeding / discharging flow rate being zero or below the flow dead zone, and the liquid level change rate being below the liquid level stability threshold; the temperature rise and temperature drop states can be determined based on whether the temperature change per unit time exceeds the temperature change threshold; and the feeding and discharging states can be determined based on the direction of liquid level change and whether the liquid level change rate exceeds the feeding / discharging judgment threshold. The flow dead zone, liquid level stability threshold, temperature change threshold, and feeding / discharging judgment threshold can be set based on instrument measurement errors, the normal fluctuation range in the stable operation data of the tank, and on-site feeding / discharging records.

[0149] For example, when the tank is stationary and the temperature change does not exceed the temperature change threshold, the normal pressure change range can be set to pressure fluctuation within the zero residual dead zone; when feeding and the liquid level rise rate exceeds the feeding judgment threshold, the normal pressure change range can be set to positive pressure change, with a maximum increase limited when the tank pressure enters the venting trigger boundary; when discharging and the liquid level fall rate exceeds the discharging judgment threshold, the normal pressure change range can be set to negative pressure change, with a maximum decrease limited when the tank pressure enters the suction trigger boundary. The above mapping table can be statistically set based on stable operating data from 7 to 30 days, or calibrated using pressure response records from the on-site commissioning phase.

[0150] Once the measured pressure change and the predicted pressure change are obtained, the deviation between the two serves as the basis for the pressure residual information. If the measured pressure change is higher than the predicted pressure change, forming a positive residual, it indicates that the pressure increase in the storage tank exceeds the expected normal response, which may correspond to delayed venting, venting obstruction, or the introduction of external high-pressure disturbances. If the measured pressure change is lower than the predicted pressure change, forming a negative residual, it indicates that the pressure decrease in the storage tank exceeds the expected normal response, which may correspond to incomplete closure, abnormal venting, obstructed air intake, or abnormally enhanced tail gas suction effects.

[0151] The step of assigning the pressure residual information of each storage tank to the storage tank node at the corresponding time position in the working condition gating spatiotemporal diagram to obtain the pressure residual diagram at the corresponding residual calculation time includes:

[0152] S3.3.1: Extract the working condition gating diagram corresponding to the current residual calculation time from the working condition gating spatiotemporal diagram.

[0153] S3.3.2: Based on the correspondence between each storage tank and each storage tank node, the pressure residual information of each storage tank at the current residual calculation time is written as the node residual value into the corresponding storage tank node.

[0154] S3.3.3: Retain the effective connectivity state, pressure disturbance propagation direction, and pressure disturbance propagation weight of each edge in the working condition gating diagram, and determine the working condition gating diagram after writing the node residual values ​​as the pressure residual diagram corresponding to the current residual calculation time.

[0155] When writing pressure residual information into the operating condition gating spatiotemporal diagram, the correspondence between the current residual calculation time and the time position in the operating condition gating spatiotemporal diagram should be determined first. If the residual calculation time coincides with the end time of a monitoring cycle of a certain operating condition gating diagram, then that operating condition gating diagram should be selected directly; if the residual calculation time falls within a certain monitoring cycle, then the operating condition gating diagram corresponding to that monitoring cycle should be selected; if there is a sampling delay or timestamp deviation, a time matching tolerance range can be set, such as not exceeding half a monitoring cycle or not exceeding 30 seconds. If the tolerance range is exceeded, the residual calculation time can be marked as a time mismatch to avoid writing pressure residual information into the wrong operating condition gating diagram.

[0156] The correspondence between storage tanks and storage tank nodes can be determined through a pre-established storage tank node mapping table. This mapping table includes at least the tank identifier, storage tank node identifier, pressure measuring point identifier, and the graph structure identifier where the node is located. For example, storage tank T-101 corresponds to storage tank node T1, storage tank T-102 corresponds to storage tank node T2, and storage tank T-103 corresponds to storage tank node T3. If a storage tank is configured with multiple pressure measuring points, the representative pressure residual value of the storage tank can be determined first when calculating the pressure residual information, and then the representative pressure residual value can be written into the corresponding storage tank node; the representative pressure residual value can be the residual of the main pressure measuring point, or the median value of the residuals of multiple valid measuring points, or a weighted value after on-site calibration.

[0157] The node residual values ​​written to the tank node can include pressure residual values, residual polarity, residual magnitude, and a validity flag. The pressure residual value represents the deviation of the measured pressure change from the predicted pressure change; the residual polarity indicates the direction of deviation; the residual magnitude indicates the degree of deviation; and the validity flag indicates whether the node has available pressure residual information at the current residual calculation time. If the pressure residual values ​​are normalized, the node residual values ​​can be values ​​in the range of -1 to 1, where positive values ​​represent positive residuals, negative values ​​represent negative residuals, and the absolute value represents the residual magnitude.

[0158] For tank nodes where no valid pressure residual information is available at the current residual calculation time, it is not advisable to directly write a zero value to avoid misinterpreting missing data as no pressure deviation. Preferably, the valid flag for the tank node can be set to invalid, and the pressure residual value can be set to null. If there is only a short-term sampling gap and interpolation has been completed using adjacent valid samples, the interpolated pressure residual value can be written, while retaining the valid flag. For tank nodes where the pressure residual value is within the residual dead zone, a zero value can be written, and the valid flag can be set to valid to indicate that the tank has available data at the current time but no pressure deviation.

[0159] Pressure residual information is typically not directly written to the breather valve node; its node attributes can retain the original equipment attributes and candidate anomaly source identifiers. This process ensures that pressure residual information is distributed only on the tank nodes, making the pressure residual map clearly express the spatial distribution of pressure deviations in each tank. The breather valve node then serves as the starting point for subsequent candidate anomaly source propagation analysis. If breather valve opening, vibration, or motion feedback data is provided on-site, this data can be retained as auxiliary attributes of the breather valve node, but it does not replace the pressure residual information on the tank nodes.

[0160] When writing node residual values, the existing edge attributes in the operating condition gating graph are not changed. That is, the effective connectivity state, pressure disturbance propagation direction, and pressure disturbance propagation weight of valve-associated edges and gas-phase coupling edges remain unchanged at that time. In this way, the pressure residual graph simultaneously contains two types of information: the pressure residual information at the tank nodes represents the current actual pressure deviation, and the operating condition gating attributes on the edges represent the current propagation paths and propagation strength. Since both are located at the same time point, it ensures that the residual distribution and propagation constraints are consistent in time.

[0161] Preferably, the pressure residual map can be generated as a copy of the operating condition gating map. That is, without altering the original operating condition gating spatiotemporal map, the nodes and edges at the current time position are copied, and then the node residual values ​​are written to the corresponding tank nodes. This copy must at least save the graph timestamp, tank node residual attributes, breather valve node attributes, and edge gating attributes. After all tank nodes corresponding to the current residual calculation time have been assigned values, the pressure residual map for that time is formed. Multiple pressure residual maps, arranged in timestamp order, can be used to observe the temporal evolution and spatial distribution of pressure residuals in the gas phase pipeline network of the tank area.

[0162] The step of using each breathing valve node as a candidate anomaly source and determining the residual propagation response corresponding to each candidate anomaly source based on the working condition gating spatiotemporal diagram includes:

[0163] S4.1: For any candidate anomaly source, the breathing valve node corresponding to the candidate anomaly source is taken as the propagation starting point in the working condition gating spatiotemporal diagram.

[0164] S4.2: Based on the aforementioned working condition gating spatiotemporal diagram, determine the propagable paths and propagation intensity from the propagation starting point to each tank node at each time location.

[0165] S4.3: Based on the propagable path and path propagation intensity corresponding to each time location, determine the residual propagation contribution of the candidate anomaly source to each tank node at each time location, and organize them in chronological order as the residual propagation response corresponding to the candidate anomaly source.

[0166] Candidate anomaly sources are used to represent breather valve nodes that may cause pressure residual propagation. Each breather valve node within the tank area can be used as the starting point for a propagation simulation, without pre-excluding other breather valve nodes based on the magnitude of the pressure residual of a particular tank. This approach avoids directly identifying the tank with the largest pressure residual as the anomaly location in a shared gas phase pipeline network, as the pressure residual may be transmitted from anomalies in the breather valves of adjacent tanks via gas phase coupling edges.

[0167] When any breather valve node is used as the propagation starting point, it is first determined whether the valve-associated edge corresponding to that breather valve node is in a valid connected state at the current time position. If the valve-associated edge is not a validly connected edge, it means that the breather valve branch does not have the conditions to effectively affect the gas phase pressure of the storage tank within the current monitoring period. In this case, the residual propagation contribution of the candidate anomaly source to each storage tank node at that time position can be set to zero. If the valve-associated edge is a validly connected edge, the anomaly effect of the breather valve node is first transmitted to its associated storage tank node. The associated storage tank node serves as the initial response node for the candidate anomaly source affecting the gas phase pipeline network of the storage tank area.

[0168] It should be noted that the pressure disturbance propagation direction on the valve-associated edge reflects the exhaust or intake relationship between the breather valve and the storage tank under the current operating conditions. When performing candidate anomaly source deduction, the breather valve node is considered the location of the fault, and its influence can be applied to the associated storage tank node through the effective valve-associated edge. In other words, the valve-associated edge not only represents the actual gas flow direction but also the direct effect of the breather valve fault on the pressure residual of the associated storage tank. For exhaust obstruction or delayed opening faults, the associated storage tank node usually exhibits a positive residual tendency; for faults related to incomplete closure, abnormal release, or intake, the associated storage tank node usually exhibits a negative residual tendency. The specific fault type is not finalized at this stage, but the positive and negative residual propagation components can be retained for matching with the actual pressure residual map sequence.

[0169] When expanding from the initial response node to other tank nodes, propagable paths are only sought along the effectively connected gas-phase coupling edges at the current time position. The propagation direction of pressure disturbances along the gas-phase coupling edges should be consistent with the propagation direction of the candidate residuals; if the edge is set to bidirectional weak propagation, propagation from either end to the other is allowed, but the path propagation intensity should be lower than that of edges with unidirectional explicit propagation. Edges set to suppress propagation do not participate in path finding. To avoid forming duplicate paths in the ring-shaped gas-phase network, paths can be set to not repeatedly pass through the same tank node, and a maximum propagation level can be limited, for example, crossing a maximum of 3 to 5 gas-phase coupling edges; paths exceeding the maximum propagation level can be excluded from response calculation.

[0170] The path propagation intensity represents the degree to which the residual influence of a candidate anomaly source is retained when it reaches the target tank node along a propagable path. Its value can be set from 0 to 1. The initial propagation intensity of directly associated tank nodes can be 1, or the pressure disturbance propagation weight of the valve-associated edge. When propagating through gas-phase coupling edges, the path propagation intensity decreases segment by segment according to the pressure disturbance propagation weight of each edge along the path. For example, if the initial intensity from the candidate breather valve node to the associated tank node is 1, and the propagation weight of the first gas-phase coupling edge is 0.8, then the path propagation intensity to the next tank node can be 0.8. If it passes through another gas-phase coupling edge with a propagation weight of 0.6, the path propagation intensity to more distant tank nodes can be further reduced to below 0.5. To reduce weak path interference, an intensity cutoff threshold can be set, for example, stopping further propagation when the path propagation intensity is below 0.05.

[0171] When multiple propagable paths exist from the same candidate anomaly source to the same tank node, the path with the highest propagation intensity can be selected as the main propagation path from the candidate anomaly source to the tank node. If the superposition effect of multiple parallel gas phase paths needs to be considered, the propagation intensities of multiple paths can be accumulated, and the accumulated result can be limited to within 1. Preferably, to reduce the risk of double counting caused by multiple paths in a shared manifold, using the maximum path propagation intensity as the residual propagation contribution of the tank node is more conducive to maintaining the comparability between candidate anomalies.

[0172] The residual propagation contribution at each time point can be represented as the theoretical residual distribution formed by the candidate breather valve node to each tank node. This theoretical residual distribution includes at least the target tank node identifier, residual propagation polarity, path propagation intensity, and main propagation path identifier. Residual propagation polarity indicates whether the candidate anomaly source may result in a positive or negative residual under the corresponding failure mode. Path propagation intensity indicates the relative strength of the impact after reaching the target tank node. The main propagation path identifier records which effective edges the failure impact is transmitted through. For unreachable tank nodes, their path propagation intensity is set to 0.

[0173] Arranging the theoretical residual distributions of the same candidate breather valve node at multiple time locations in chronological order yields the residual propagation response corresponding to that candidate anomaly source. This residual propagation response is not the actual measured pressure residual information, but rather the influence pattern of the candidate anomaly source derived from the spatiotemporal diagram of the operating condition gating system. Since the propagable path and path propagation intensity at each time location are constrained by the current operating condition gating attributes, the same candidate breather valve can generate different residual propagation ranges and intensities in different monitoring periods, thus reflecting the effect of changes in the gas phase pipeline connectivity on the distribution of anomaly influences.

[0174] Based on the matching results between the residual propagation response and the pressure residual map sequence, abnormal breathing valve nodes and their fault types are determined, and breathing valve fault diagnosis results are generated, including:

[0175] S5.1: Align the residual propagation response corresponding to any candidate anomaly source with the pressure residual information of the same time location and the same tank node in the pressure residual map sequence.

[0176] S5.2: Based on the alignment results, compare the consistency between the residual propagation response and the pressure residual information in terms of residual polarity and residual amplitude to obtain the response consistency value corresponding to the candidate anomaly source.

[0177] S5.3: The breathing valve node corresponding to the candidate anomaly source whose response consistency value meets the preset anomaly judgment condition is determined as the abnormal breathing valve node.

[0178] S5.4: Based on the correspondence between the residual propagation response corresponding to the abnormal breathing valve node and the pressure residual diagram sequence in terms of time persistence, residual polarity, and node distribution, a preset fault feature rule is matched to determine the fault type of the abnormal breathing valve node and generate the breathing valve fault diagnosis result.

[0179] During matching, candidate anomaly sources are aligned one by one with the pressure residual map sequence corresponding to each candidate breather valve node. Alignment relationships include time location alignment and tank node alignment. Time location alignment refers to selecting the theoretical residual propagation distribution and actual pressure residual map at the same residual calculation time. Tank node alignment refers to comparing the theoretical residual propagation contribution and actual pressure residual information at the same tank node. If the pressure residual information of a tank node at the current time location is marked as invalid, that node will not participate in the consistency calculation at the current time location. If the theoretical residual propagation contribution is zero while the actual pressure residual amplitude exceeds the residual dead zone, this situation can be considered a factor reducing consistency.

[0180] Residual polarity consistency is used to determine whether the residual direction derived from candidate anomalies is the same as the actual pressure residual direction. When both the theoretical residual propagation response and the actual pressure residual information are positive, they are considered to have consistent polarity; similarly, when both are negative, they are also considered to have consistent polarity. If the theoretical residual propagation response and the actual pressure residual information have opposite polarities, they are considered to have inconsistent polarity. Nodes whose actual pressure residual amplitude is within the residual dead zone can be considered as having no residual and are not used as the primary basis for polarity consistency or inconsistency, to avoid small noise affecting the matching results.

[0181] Residual amplitude consistency is used to determine whether the propagation strength inferred from candidate anomalies matches the actual pressure residual amplitude distribution. Since the path propagation intensity in the residual propagation response is typically a relative value between 0 and 1, while the actual pressure residual value can be a normalized value, the actual pressure residual amplitude of each tank node at the same time location can be normalized to the range of 0 to 1 before being compared with the path propagation intensity of the corresponding node. For example, when the difference between the normalized amplitude of the actual pressure residual and the path propagation intensity does not exceed 0.2, it can be determined that the amplitudes are highly consistent; when the difference is between 0.2 and 0.5, it can be determined that the amplitudes are partially consistent; and when the difference exceeds 0.5, it can be determined that the amplitude consistency is low. These difference thresholds are determined based on historical normal operation data and the amplitude deviation distribution of known anomaly samples. Specifically, historical monitoring segments with clear operating status of the breathing valve can be selected first, and the differences between the residual propagation response of the candidate abnormal source and the normalized amplitude of the actual pressure residual can be calculated respectively. Among them, the concentrated distribution interval of differences in the normal matching samples can be used as the basis for setting the threshold of high amplitude consistency, and the boundary interval between the normal matching samples and the non-matching samples can be used as the boundary for partial consistency and low consistency.

[0182] The response consistency value is used to comprehensively represent the degree of matching between the residual propagation response of a candidate anomaly source and the actual pressure residual map sequence. The value can be set from 0 to 1. Preferably, residual polarity consistency can be used as the primary criterion, and residual amplitude consistency as the secondary criterion, and both can be aggregated according to preset weights. For example, the polarity consistency weight can be set to 0.6, and the amplitude consistency weight can be set to 0.4. These weights are determined based on the discrimination contribution in historical anomaly samples; when historical samples indicate that the residual direction dominates the anomaly source differentiation contribution, the polarity consistency weight is set to be greater than the amplitude consistency weight. For tank nodes with path propagation intensity reaching the strong propagation range, the matching results are assigned a strong propagation node influence weight; for tank nodes with path propagation intensity in the weak propagation range, the matching results are assigned a weak propagation node influence weight, and the strong propagation node influence weight is greater than the weak propagation node influence weight. The strong propagation interval and weak propagation interval are determined based on the normalized distribution of path propagation intensity. For example, a path propagation intensity of not less than 0.6 is classified as a strong propagation interval, and a path propagation intensity of not more than 0.3 is classified as a weak propagation interval. The corresponding boundary values ​​are calibrated based on the range of propagation intensity that can be stably identified in historical pressure response records. The response consistency value obtained in this way can reflect the overall matching degree of candidate anomaly sources in terms of time, nodes, and residual morphology.

[0183] Preset anomaly detection criteria can be set based on historical operating data of the tank farm and on-site maintenance requirements. For example, when the response consistency value of a candidate anomaly source is not less than 0.7, and the difference between its response consistency value and other candidate anomaly sources is not less than 0.1, the breather valve node corresponding to that candidate anomaly source can be identified as an abnormal breather valve node. Here, 0.7 can be used as the anomaly location threshold, and 0.1 can be used as the candidate source differentiation threshold; both can be calibrated based on historical pressure residual matching results. If the response consistency values ​​of multiple candidate anomaly sources all exceed the anomaly location threshold, and the difference between the highest response consistency value and other response consistency values ​​exceeding the threshold is less than the candidate source differentiation threshold, then the breather valve node with the highest response consistency value is retained as the primary abnormal breather valve node, and the other candidate breather valve nodes are marked as nodes to be reviewed. If the response consistency values ​​of all candidate anomaly sources are less than the anomaly location threshold, then the current pressure residual map sequence is determined not to meet the breather valve anomaly location criteria.

[0184] After identifying the abnormal breathing valve node, preset fault characteristic rules can be matched based on time persistence, residual polarity, and node distribution. Time persistence indicates whether the abnormal residual occurs within multiple consecutive residual calculation times. For example, if pressure residuals with the same polarity and amplitude exceeding the residual dead zone exist for two to three consecutive monitoring cycles, they can be considered persistent. The residual dead zone is determined based on the pressure sensor's allowable error, the pressure residual fluctuation range under stable operating conditions, and the modeling error under normal operating conditions, and is used to exclude instrument noise and normal short-term fluctuations. Node distribution indicates whether pressure residuals with amplitudes exceeding the residual dead zone are concentrated at the abnormal breathing valve-associated tank node, or whether they decrease in distribution towards adjacent tank nodes along the effective gas-phase coupling edge in the operating condition gating diagram. If the residual distribution is consistent with the propagation path of the candidate anomaly source, it meets the characteristics of pressure disturbance caused by breathing valve failure; if the residual distribution deviates from the propagation path of the candidate anomaly source, it should not be directly attributed to the breathing valve.

[0185] For example, when the tank pressure is close to or exceeds the exhaust set pressure, if the abnormal breather valve associated tank node shows a continuous positive residual, and adjacent tank nodes show positive residuals of the same polarity that gradually weaken along the effective gas phase coupling path, this can be matched as exhaust obstruction or opening hysteresis. If the positive residual only appears for a short period of time and weakens rapidly after several monitoring cycles, and the corresponding pressure range is close to the exhaust action threshold, it is more suitable to determine it as opening hysteresis; if the positive residual persists and does not decay over time, it is more suitable to determine it as exhaust obstruction.

[0186] When the tank pressure is lower than or close to the set intake pressure, if a persistent negative residual occurs at the abnormal breather valve-associated tank node, and this negative residual does not fully correspond to normal factors such as material discharge, temperature drop, or tail gas extraction, it can be identified as intake obstruction or intake opening lag. If the negative residual disappears after reaching the intake action range, it can be determined as intake opening lag; if the negative residual persists for multiple monitoring cycles and its amplitude exceeds the intake obstruction judgment threshold, it can be determined as intake obstruction. The intake obstruction judgment threshold is determined based on the pressure residual range under normal breather valve intake response, the allowable error of the pressure sensor, and the negative residual amplitude in historical intake obstruction samples. It is used to distinguish between short-term pressure deviations during normal intake and persistent negative pressure deviations caused by intake channel obstruction.

[0187] If, under conditions where the tank pressure has returned to the normal range or the breather valve should be closed, a residual pressure in the same direction as the release or intake continues to appear at the associated tank node of the abnormal breather valve, and this residual pressure is mainly concentrated at the associated tank node and its neighboring nodes, it can be matched as either incomplete closure or abnormal reseating. If the residual pressure persists for a short period after the breather valve operation ends but gradually decreases, it can be judged as abnormal reseating; if the residual pressure persists for a long time and repeatedly appears under static conditions, it can be judged as incomplete closure. The final output breather valve fault diagnosis result can include the abnormal breather valve node identifier, fault type, response consistency value, matching time range, and main matched tank nodes for verification of the diagnosis results.

[0188] Example 2

[0189] Please see Figure 4 The present invention provides an embodiment of a breathing valve fault diagnosis system based on pressure residual propagation matching, the system comprising:

[0190] The graph structure construction module is used to acquire the topology information and operating data of the gas phase pressure monitoring system, and construct a basic graph structure based on the topology information, with pressure monitoring nodes and breathing valve nodes as graph nodes and valve association relationships and gas phase coupling relationships as edges;

[0191] The gating graph generation module is used to update the basic graph structure with working condition gating based on the operating data and generate a working condition gating spatiotemporal graph.

[0192] The residual map generation module is used to determine the pressure residual information of each pressure monitoring node based on the operating data, and write the pressure residual information into the corresponding pressure monitoring node in the working condition gated spatiotemporal map to generate a pressure residual map sequence.

[0193] The propagation response module is used to take each breathing valve node as a candidate anomaly source and determine the residual propagation response corresponding to each candidate anomaly source based on the working condition gating spatiotemporal diagram.

[0194] The diagnostic result generation module is used to determine the abnormal breathing valve node and its fault type based on the matching result of the residual propagation response and the pressure residual diagram sequence, and generate the breathing valve fault diagnosis result.

[0195] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention.

Claims

1. A method for diagnosing breather valve faults based on pressure residual propagation matching, characterized in that, The method includes: Obtain the topology information and operating data of the gas phase pressure monitoring system, and construct a basic graph structure based on the topology information, with pressure monitoring nodes and breathing valve nodes as graph nodes and valve association relationships and gas phase coupling relationships as edges; Based on the operational data, the basic graph structure is updated with operational condition gating to generate an operational condition gating spatiotemporal graph. Based on the operating data, the pressure residual information of each pressure monitoring node is determined, and the pressure residual information is written into the corresponding pressure monitoring node in the working condition gating spatiotemporal diagram to generate a pressure residual diagram sequence. Each breathing valve node is used as a candidate anomaly source, and the residual propagation response corresponding to each candidate anomaly source is determined based on the working condition gating spatiotemporal diagram. Based on the matching results between the residual propagation response and the pressure residual diagram sequence, abnormal breathing valve nodes and their fault types are determined, and breathing valve fault diagnosis results are generated.

2. The method for diagnosing a breather valve fault based on pressure residual propagation matching according to claim 1, characterized in that, Based on the aforementioned topology information, a basic graph structure is constructed, with pressure monitoring nodes and breathing valve nodes as graph nodes and valve association relationships and gas phase coupling relationships as edges, including: Based on the topology information, determine each pressure monitoring node, the valve association between each pressure monitoring node and each breathing valve, and the gas phase coupling relationship between each pressure monitoring node through the gas phase pipeline. Based on the valve association relationship, a corresponding breathing valve node is set for each breathing valve; A valve-associated edge is formed between pressure monitoring nodes with valve association and breather valve nodes, and a gas-phase coupling edge is formed between pressure monitoring nodes with gas-phase coupling. The basic graph structure is constructed based on the pressure monitoring node, the breathing valve node, the valve associated edge, and the gas phase coupling edge.

3. The method for diagnosing a breather valve fault based on pressure residual propagation matching according to claim 1, characterized in that, Based on the operational data, the basic graph structure is updated using operational condition gating to generate an operational condition gating spatiotemporal graph, including: Multiple monitoring cycles are determined based on the collection time of the aforementioned operational data; For each monitoring cycle, the current operating status of each edge in the basic graph structure is determined based on the operating data within that monitoring cycle; Based on the current operating condition, the effective connectivity state, pressure disturbance propagation direction, and pressure disturbance propagation weight of each edge in the basic graph structure are set to obtain the operating condition gating graph for the corresponding monitoring period. Organize the gating diagrams for each operating condition according to the time sequence of the multiple monitoring cycles, and generate a spatiotemporal diagram of the operating condition gating.

4. The method for diagnosing a breather valve fault based on pressure residual propagation matching according to claim 3, characterized in that, Based on the current operating condition, the effective connectivity state, pressure disturbance propagation direction, and pressure disturbance propagation weight of each edge in the basic graph structure are set, including: Based on the current working condition and the connection status of each edge, determine whether the corresponding edge is a valid connection edge in the current monitoring period; For the effective connected edge, the pressure disturbance propagation direction of the effective connected edge is determined according to the pressure relationship between the nodes on both sides of the edge, the edge type, and the breathing valve direction information. Based on the propagation direction of the pressure disturbance and the corresponding edge's operational data, determine the pressure disturbance propagation weight of the effective connected edge; Suppress propagation states are set for non-effective connected edges, and the effective connected states, pressure disturbance propagation directions, and pressure disturbance propagation weights corresponding to each edge are written into the basic graph structure to obtain the working condition gating graph for the corresponding monitoring period.

5. The method for diagnosing a breather valve fault based on pressure residual propagation matching according to claim 1, characterized in that, Based on the operational data, the pressure residual information of each pressure monitoring node is determined, and the pressure residual information is written into the corresponding pressure monitoring node in the operating condition gating spatiotemporal diagram to generate a pressure residual diagram sequence, including: Multiple residual calculation times are determined based on the data collection time of the operation; For each residual calculation time, the pressure residual information of each pressure monitoring node is determined based on the operating data corresponding to that residual calculation time; The pressure residual information of each pressure monitoring node is assigned to the pressure monitoring node at the corresponding time position in the working condition gating spatiotemporal diagram to obtain the pressure residual diagram at the corresponding residual calculation time. The pressure residual diagrams are organized according to the time sequence of the multiple residual calculation times to generate the pressure residual diagram sequence.

6. The method for diagnosing a breather valve fault based on pressure residual propagation matching according to claim 5, characterized in that, For each residual calculation time, the pressure residual information of each pressure monitoring node is determined based on the operating data corresponding to that residual calculation time, including: For any pressure monitoring node, the measured pressure change of the pressure monitoring node is determined based on the measured pressure data within the time window corresponding to the current residual calculation time. Based on the operating data within the time window corresponding to the current residual calculation time, the predicted pressure change of the pressure monitoring node when the associated breathing valve is in a normal response state is determined; Based on the deviation between the measured pressure change and the predicted pressure change, the pressure residual information of the pressure monitoring node at the current residual calculation time is determined.

7. The method for diagnosing a breather valve fault based on pressure residual propagation matching according to claim 5, characterized in that, The step of assigning the pressure residual information of each pressure monitoring node to the pressure monitoring node at the corresponding time position in the working condition gating spatiotemporal diagram to obtain the pressure residual diagram at the corresponding residual calculation time includes: Extract the working condition gating diagram corresponding to the current residual calculation time from the working condition gating spatiotemporal diagram; Write the pressure residual information of each pressure monitoring node at the current residual calculation time as the node residual value into the corresponding pressure monitoring node. The effective connectivity state, pressure disturbance propagation direction, and pressure disturbance propagation weight of each edge in the working condition gating diagram are retained, and the working condition gating diagram after writing the node residual values ​​is determined as the pressure residual diagram corresponding to the current residual calculation time.

8. The method for diagnosing a breather valve fault based on pressure residual propagation matching according to claim 1, characterized in that, The step of using each breathing valve node as a candidate anomaly source and determining the residual propagation response corresponding to each candidate anomaly source based on the working condition gating spatiotemporal diagram includes: For any candidate anomaly source, the breathing valve node corresponding to the candidate anomaly source is taken as the propagation starting point in the working condition gating spatiotemporal diagram; Based on the aforementioned working condition gating spatiotemporal diagram, the propagation paths from the propagation starting point to each pressure monitoring node and the propagation intensity of the paths are determined at each time location. Based on the propagable paths and propagation intensity corresponding to each time location, the residual propagation contribution of the candidate anomaly source to each pressure monitoring node at each time location is determined, and then organized in chronological order as the residual propagation response corresponding to the candidate anomaly source.

9. A method for diagnosing a breather valve fault based on pressure residual propagation matching according to claim 8, characterized in that, Based on the matching results between the residual propagation response and the pressure residual map sequence, abnormal breathing valve nodes and their fault types are determined, and breathing valve fault diagnosis results are generated, including: Align the residual propagation response corresponding to any candidate anomaly source with the pressure residual information of the same pressure monitoring node at the same time position in the pressure residual map sequence; Based on the alignment results, the consistency between the residual propagation response and the pressure residual information in terms of residual polarity and residual amplitude is compared to obtain the response consistency value corresponding to the candidate anomaly source. The breathing valve node corresponding to the candidate anomaly source whose response consistency value meets the preset anomaly judgment condition is determined as the abnormal breathing valve node. Based on the correspondence between the residual propagation response corresponding to the abnormal breathing valve node and the pressure residual map sequence in terms of time persistence, residual polarity, and node distribution, a preset fault feature rule is matched to determine the fault type of the abnormal breathing valve node and generate the breathing valve fault diagnosis result.

10. A breathing valve fault diagnosis system based on pressure residual propagation matching, used to implement the breathing valve fault diagnosis method based on pressure residual propagation matching as described in any one of claims 1 to 9, characterized in that, The system includes: The graph structure construction module is used to acquire the topology information and operating data of the gas phase pressure monitoring system, and construct a basic graph structure based on the topology information, with pressure monitoring nodes and breathing valve nodes as graph nodes and valve association relationships and gas phase coupling relationships as edges; The gating graph generation module is used to update the basic graph structure with working condition gating based on the operating data and generate a working condition gating spatiotemporal graph. The residual map generation module is used to determine the pressure residual information of each pressure monitoring node based on the operating data, and write the pressure residual information into the corresponding pressure monitoring node in the working condition gated spatiotemporal map to generate a pressure residual map sequence. The propagation response module is used to take each breathing valve node as a candidate anomaly source and determine the residual propagation response corresponding to each candidate anomaly source based on the working condition gating spatiotemporal diagram. The diagnostic result generation module is used to determine the abnormal breathing valve node and its fault type based on the matching result of the residual propagation response and the pressure residual diagram sequence, and generate the breathing valve fault diagnosis result.

Images