A supercritical combined heat and power unit equipment fault diagnosis method

Abstract

The present application relates to the technical field of power generator set equipment fault diagnosis, and discloses a supercritical combined heat and power unit equipment fault diagnosis method, comprising the following steps: S1, obtaining relevant information of a supercritical combined heat and power unit equipment, and identifying the working condition of the unit equipment running state based on the relevant information to form a corresponding working condition label; S2, under the constraint of the working condition label, identifying anchor point events, and constructing an event window around the anchor point events to analyze the information in the event window; S3, based on the response sequence relationship of the information in the event window, constructing a running structure reflecting the running influence relationship of the unit equipment. By time aligning the measuring point data, control signals and heat and power load distribution related information and forming a relevant information sequence under a unified time reference, the working condition features are extracted and discretely combined and coded to generate a working condition label, which improves the comparability and stability of the diagnosis results of different unit equipment in different running stages.

Description

Technical Field

[0001] This invention relates to the field of fault diagnosis technology for thermal power generating units, specifically a fault diagnosis method for supercritical combined heat and power generating units. Background Technology

[0002] Supercritical combined heat and power (CHP) units refer to coal-fired power units where the main steam parameters are in a supercritical state and simultaneously undertake power generation and heating tasks. Their equipment systems typically include a boiler and its combustion and soot blowing systems, a steam turbine and extraction steam heating systems, a condenser and circulating water system (cold end systems), and associated valve actuators and control systems. These units operate long-term under the coordinated constraints of grid load and heating demand, with frequent switching of operating modes and complex thermoelectric coupling relationships. Abnormalities in key equipment can easily lead to efficiency decline, fluctuations in heating capacity, or safety risks. Therefore, it is necessary to conduct fault diagnosis across the entire process of the unit equipment to support the timely detection, location, and handling of abnormalities.

[0003] Existing technologies typically rely on measurement data from unit equipment or partial control signals for status assessment and anomaly detection. The diagnostic process often involves directly analyzing features or roughly grouping historical operating conditions after data acquisition. Consequently, when load switching or changes in operating mode occur in cogeneration units, it is difficult to maintain consistency in the time reference and status meaning of data from different sources. This leads to inconsistent operational semantics corresponding to diagnostic inputs at the same time, making it easy for subsequent feature analysis to misidentify normal fluctuations caused by operating condition switching as anomalies. This affects the comparability and stability of diagnostic results across different operating stages. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a fault diagnosis method for supercritical combined heat and power (CHP) units, which solves the problem that the lack of unified operating condition constraints in diagnostic inputs under CHP load switching and operating mode fluctuations easily leads to misjudgments and insufficient comparability and stability of diagnostic results.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a method for diagnosing equipment faults in supercritical combined heat and power units, comprising the following steps: S1. Obtain relevant information about the supercritical cogeneration unit equipment, and identify the operating status of the unit equipment based on the relevant information to form corresponding operating condition tags; S2. Under the constraints of the operating condition label, identify anchor point events, construct an event window around the anchor point events, and analyze the information within the event window; S3. Based on the response order relationship of the information in the event window, construct an operation structure that reflects the impact relationship of unit equipment operation, which is used to describe the operation structure status of unit equipment under the corresponding operating condition label; S4. Based on historical operation data, construct a trusted domain for each operating condition label. The trusted domain for the operating structure is used to limit the operating structure states that are allowed to exist under the operating condition label. S5. Compare the currently constructed operating structure with the operating structure trusted domain under the corresponding working condition label. When the operating structure exceeds the operating structure trusted domain, generate operating structure out-of-bounds information and drive the irreversible diagnostic state machine to perform state evolution based on the operating structure out-of-bounds information. S6. When the irreversible diagnostic state machine enters the preset structural confirmation abnormality state, it outputs the unit equipment fault diagnosis result and records the diagnostic evidence corresponding to the fault diagnosis result.

[0006] Preferably, in step S1, the relevant information of the supercritical cogeneration unit equipment includes measurement data, control signals, and information related to the distribution of heat and power loads during the operation of the unit equipment.

[0007] Preferably, in step S1, forming the corresponding operating condition label includes the following steps: (1) Perform time alignment processing on the relevant information to obtain a relevant information sequence under the same time base, and select operating condition feature items to characterize the operating status of the unit equipment from the relevant information sequence to form an operating condition feature set; (2) Discretize the set of operating conditions to obtain the operating condition status corresponding to each operating condition feature item, combine and encode the operating condition status to generate an operating condition label that corresponds one-to-one with the operating status of the unit equipment, and update the current operating condition label when the operating condition label changes.

[0008] Preferably, in step S2, identifying the anchor point event includes the following steps: (1) Monitor the control signals and changes in operating status in the relevant information; (2) When equipment switching, valve action, soot blowing operation or change of operating mode is detected, the corresponding change in operating status is identified as an anchor event, and the occurrence time and type of the anchor event are recorded.

[0009] Preferably, in step S2, constructing the event window around the anchor point event includes the following steps: (1) Taking the time of the anchor event as the center, determine the preset time range before and after the anchor event, and extract the relevant information within the preset time range to form an event window; (2) Use the relevant information in the event window as input data for anchor event impact analysis.

[0010] Preferably, in step S3, constructing the operating structure that reflects the operational impact of unit equipment includes the following steps: (1) Within the event window corresponding to the same anchor point event, analyze the order in which different related information generates responses, and determine the direction of the operational influence between related information based on the order of the responses; (2) Based on the aforementioned operational influence direction, construct an operational structure to describe the operational influence relationship of the unit equipment; (3) Determine the event type and make timely corrections based on parameter matching similarity.

[0011] Preferably, in step S4, constructing the trusted domain of the runtime structure includes the following steps: (1) Filter out the running segments with stable operating condition labels in the historical operating data, and summarize the operating structure formed within the running segments; (2) Based on the summary results, construct the trusted domain of the operating structure that is allowed to exist under the corresponding operating condition label.

[0012] Preferably, the trusted domain of the operational structure includes a set of operational structures that are allowed to exist under the corresponding operating condition label, a set of anchor event types that are allowed to occur, and a set of event propagation paths that are allowed to be established.

[0013] Preferably, the irreversible diagnostic state machine in step S5 includes a normal structural state, a structurally interpretable evolutionary state, a structurally abnormal candidate state, and a structurally confirmed abnormal state.

[0014] Preferably, in step S5, driving the irreversible diagnostic state machine to undergo state evolution includes the following steps: (1) Compare the currently constructed operating structure with the operating structure trust domain under the corresponding operating condition label. When the operating structure exceeds the operating structure trust domain, generate operating structure out-of-bounds information. (2) Based on the out-of-bounds information of the running structure, the irreversible diagnostic state machine evolves from the normal structural state to the candidate state of structural anomaly; (3) When the out-of-bounds information of the running structure is repeatedly detected during continuous operation under the same working condition label constraint, the irreversible diagnostic state machine is made to enter the structural confirmation abnormal state. (4) When a preset reverse anchor event sequence is detected, the irreversible diagnostic state machine is allowed to revert to the structurally interpretable evolution state.

[0015] This invention provides a method for diagnosing equipment faults in supercritical combined heat and power (CHP) units. It has the following beneficial effects: 1. This invention aligns the measurement point data, control signals, and information related to the distribution of cogeneration loads in time to form a sequence of relevant information under a unified time reference. Then, it extracts operating condition features from these sequences and discretizes and combines them into codes to generate operating condition labels. This allows the entire subsequent diagnostic process to be carried out under the same operating condition constraints, thereby reducing the risk of misjudgment caused by load switching and fluctuations in operating modes of cogeneration from the source. It also improves the comparability and stability of diagnostic results for different units and equipment at different operating stages.

[0016] 2. Under the constraint of operating condition labels, this invention uses changes in operating status such as equipment switching, valve action, soot blowing operation, or change of operating mode as anchor events, and constructs an event window centered on the time of occurrence of the anchor event. The relevant information in the event window is used as the input for impact analysis, so that the diagnosis focuses on the local dynamic process directly related to operation and state transition. This can effectively suppress the interference of long-term drift and irrelevant fluctuations on feature extraction, thereby improving the ability to distinguish between early fault anomalies and operational disturbances and the efficiency of location.

[0017] 3. This invention constructs an operational structure that reflects the direction of operational impact based on the response sequence relationship within the event window, and establishes a reliable domain for the operational structure using historical stable operating condition data. When the current operational structure goes out of bounds, it generates out-of-bounds information to drive the irreversible diagnostic state machine to evolve step by step, and outputs diagnostic results and evidence after confirming the anomaly. This distinguishes between one-off occasional deviations and persistent structural deviations, reducing false alarms and enhancing the interpretability and traceability of diagnostic conclusions. It also facilitates operators to quickly verify and take appropriate measures in conjunction with the out-of-bounds propagation link. Attached Figure Description

[0018] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0019] The technical solution of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] Please see the appendix Figure 1 This invention provides a method for diagnosing equipment faults in supercritical combined heat and power units, comprising the following steps: S1. Obtain relevant information about the supercritical cogeneration unit equipment, and identify the operating status of the unit equipment based on the relevant information to form corresponding operating condition labels. Furthermore, in step S1, the relevant information of the supercritical cogeneration unit equipment includes the measurement data, control signals, and information related to the distribution of heat and power loads during the operation of the unit equipment. Furthermore, in step S1, generating the corresponding operating condition label includes the following steps: (1) Time alignment processing is performed on the relevant information to obtain the relevant information sequence under the same time base. The operating condition feature items used to characterize the operating status of the unit equipment are selected from the relevant information sequence to form the operating condition feature set. (2) Discretize the set of operating condition features to obtain the operating condition feature status corresponding to each operating condition feature item, combine and encode the operating condition feature status to generate an operating condition label that corresponds one-to-one with the operating status of the unit equipment, and update the current operating condition label when the operating condition label changes. By adopting the above technical solution, step S1 first collects relevant information about the supercritical combined heat and power (CHP) unit from its operating environment. This relevant information includes at least measurement point data, control signals, and information related to CHP load distribution during unit operation. Measurement point data corresponds to the time-series sampled values ​​of the unit's operating conditions and status variables during operation. Control signals correspond to the time-series sampled values ​​of the unit's control and execution mechanisms' commands or status variables during operation. Information related to CHP load distribution corresponds to load commands or operating mode indications generated by the scheduling or load distribution module during the unit's coordinated power generation and heating operation. To ensure that data from different sources can be processed consistently within the same operating condition identification logic, time alignment processing is performed on the relevant information. This maps the measurement point data, control signals, and CHP load distribution information to a relevant information sequence under the same time reference. The time-aligned relevant information sequence is denoted as... ,in, Indicates the sampling time on the time base. It consists of corresponding elements of the measurement point data sequence, control signal sequence, and thermoelectric load distribution related information sequence at the same sampling time: (The sentence is incomplete and requires further context.) Then, operating condition feature items that characterize the operating status of the unit equipment are selected from the relevant information sequence to form an operating condition feature set, denoted as . Each of these Each condition is composed of one or more quantities from a relevant information sequence and is directly related to the operating status of the unit equipment. For example, it is composed of key operating quantities from measurement point data, key command quantities from control signals, and operating mode indication quantities from information related to heat and power load distribution, enabling it to characterize the changes in the operating conditions of the unit equipment under conditions of coordinated power generation and heating. Subsequently, the set of operating condition features is discretized to obtain the operating condition feature state corresponding to each feature item. Discretization can be achieved through a pre-defined state partitioning mapping, dividing each feature... Mapped to state values ​​in a finite set of states This mapping is denoted as in, Indicates that for the first Discretization mapping rules for each working condition feature item This indicates that the characteristic of this operating condition is at time [time]. The working condition characteristics; in obtaining the state vector Then, the operating condition characteristics are combined and encoded to generate operating condition tags corresponding to the operating status of the unit equipment. This combination encoding can be implemented using a deterministic encoding function. The operating condition tags are denoted as follows: It is determined by the state vector and satisfies ,in, For combined encoding functions, The output of the operating condition label identifies the operating condition category of the unit equipment at that moment, enabling subsequent steps to perform consistency analysis on anchor events and operating structure under the constraints of the operating condition label. When the relevant information sequence changes during operation, causing a change in the operating condition feature state vector, the current operating condition label is updated accordingly. That is, the operating condition feature extraction, discretization, and combination encoding are repeatedly performed for each sampling moment to obtain a new label. The updated operating condition labels are used as input constraints for subsequent steps, thereby enabling the continuous generation of operating condition label outputs consistent with the operating status of the unit equipment during operating condition transitions caused by changes in the power generation and heating loads of the unit equipment, switching of control signals, or adjustments to the heat and power load distribution strategy, and supporting subsequent fault diagnosis processes.

[0021] S2. Under the constraints of the working condition label, identify the anchor point event, construct an event window around the anchor point event, and analyze the information in the event window; Furthermore, in step S2, identifying anchor events includes the following steps: (1) Monitor the control signals and changes in operating status in the relevant information; (2) When equipment switching, valve action, soot blowing operation or change of operating mode is detected, the corresponding change in operating status is identified as an anchor event, and the occurrence time and type of the anchor event are recorded. Furthermore, in step S2, constructing the event window around the anchor event includes the following steps: (1) Taking the time of the anchor event as the center, determine the preset time range before and after the anchor event, and extract the relevant information within the preset time range to form an event window; (2) Use the relevant information in the event window as input data for anchor event impact analysis; By adopting the above technical solution, under the constraint of operating condition labels, the relevant information collected by the distributed control system of the power plant is filtered according to the label function. Anchor event identification is performed only on the time series that meet the value of the label function. The relevant information includes at least the control signal sequence and the operating state sequence. The control signal is represented by a vector function using a set of symbols. and The components correspond to observable measurements such as valve opening commands, valve opening feedback, or equipment start / stop commands. The operating state is represented by a vector function using a set of symbols. The components correspond to discrete state quantities such as equipment activation / deactivation status, valve opening / closing status, soot blowing permission and execution status, or operating mode indicators. Anchor point determination is achieved through joint monitoring of control signal changes and state jump variables. The control signal change is defined as... Its Let δ be the sampling time, δ be the sampling interval, and the running state transition be defined as follows: When the triggering criterion is met Right now This moment will be identified as the moment the anchor event occurs. And obtain the anchor event type based on the trigger source mapping. Subsequently An event window is constructed around a central location as input data for the impact analysis; the event window uses a set. . indicates and ,in, The relevant information vector is summarized at time t and is... ,and Composition and The preset time range parameter is used to cover the dynamic response process before and after the anchor point event, so that the relevant information in the event window can form a comparable local time sequence segment under the same working condition label constraint and serve as the input for the anchor point event impact analysis. This enables focused extraction of the linkage changes between control commands and equipment status before and after the anchor point event, reduces the interference of load fluctuations and long-term drift on the judgment, and improves the ability and efficiency of subsequent fault diagnosis to distinguish between transient anomalies and operational disturbances.

[0022] S3. Based on the response order relationship of information within the event window, construct an operation structure that reflects the impact relationship of unit equipment operation, which is used to describe the operation structure status of unit equipment under the corresponding operating condition label; Furthermore, in step S3, constructing an operational structure that reflects the operational impact relationships of unit equipment includes the following steps: (1) Within the event window corresponding to the same anchor point event, analyze the order in which different related information generates responses, and determine the direction of the operational influence between related information based on the order of responses; (2) Based on the direction of operational influence, construct an operational structure to describe the operational influence relationship of unit equipment; (3) Determine the event type and make timely corrections based on parameter matching similarity.

[0023] By adopting the above technical solution, an event window for each anchor point event is created under the corresponding working condition label. Related information sequence within The operational structure state is established by extracting the response sequence to determine the impact of unit equipment operation. From the control signal vector With running state vector The components are composed of valve opening commands or feedback, as well as actual data of the unit equipment such as equipment activation / deactivation, valve opening / closing, soot blowing execution, and operating mode indicators. First, each component signal... Determine and define the response time within the event window. The response time is obtained by exceeding the response amplitude criterion. in, The baseline value of the event window's starting segment and can be determined by... Obtain and For the time span used to estimate the baseline and The response threshold of the signal is used. The magnitude comparison yields the response order and determines the direction of the impact. Furthermore, the consistency of the changes in the two signals within the window is satisfied. When determining whether there is an influence of the direction of execution pointing to j. in The correlation coefficient between two signal change sequences within the event window is defined as follows: and for The sequence of changes relative to the baseline within the window and Its standard deviation, A consistency threshold is set, and then an operational structure is constructed based on all directed relations that satisfy the judgment conditions. To characterize the operational structural state under this working condition label, where, It is a set of relevant information nodes, with each node corresponding to a specific measurement point of the unit equipment. It is a set of directed edges consisting of the directions of influence of the operation and can be represented by an adjacency matrix. Indicates and If and only if there exists otherwise This operational structure enables the explicit and solidified response chain of control and state variables caused by valve action equipment switching, soot blowing, or changes in operating mode after the same anchor point event is triggered. This results in a comparable structured state description, allowing subsequent fault diagnosis to identify abnormal response sequence and abnormal influence direction under the same operating condition label constraints and associate them with the actual measurement points of the unit equipment, thereby improving the interpretable location capability of deviations in operational influence relationships.

[0024] S4. Based on historical operation data, construct a trusted domain for the operation structure for each operating condition label. The trusted domain for the operation structure is used to limit the operating structure states that are allowed to exist under the operating condition label. Furthermore, in step S4, constructing the trusted domain of the runtime structure includes the following steps: (1) Filter out the running segments with stable operating condition labels in the historical operating data, and summarize the operating structure formed within the running segments; (2) Based on the summary results, construct a trusted domain of the operating structure that is allowed to exist under the corresponding operating condition label; Furthermore, the trusted domain of the operational structure includes the set of operational structures that are allowed to exist under the corresponding operating condition label, the set of anchor event types that are allowed to occur, and the set of event propagation paths that are allowed to be established; By adopting the above technical solution, when constructing the trusted domain of the operating structure for each operating condition label based on historical operating data, the first step is to select operating segments in the historical data where the operating condition labels remain stable. This ensures that the statistical results correspond to the same operating mode and boundary conditions. The stable operating segments can be determined by the label function. Constant determination within a time period, that is, for any candidate time period Satisfy all belong All Thus, the label was determined to be Stable running dataset Subsequently Internally, by repeatedly executing the runtime structure on each anchor point event and its event window, a set of runtime structure samples is obtained. in, The runtime structure corresponding to an anchor event and the adjacency matrix is ​​available. It indicates that its elements This indicates whether the direction of operational influence from one measuring point to another is valid, and includes actual data on the unit equipment such as the valve opening command or feedback equipment activation / deactivation status, valve opening / closing status, soot blowing execution status, and operating mode indicators corresponding to the measuring point. Based on this, operational structure samples are summarized to form the river information domain boundary and avoid interference from occasional disturbances. The summary can be defined by the edge occurrence frequency. in For the tag is The proportion of cases that were established in the historical samples. The number of samples is determined by the set of allowed edges. Defined as ,in, The confidence threshold is used to limit the allowed operational influence directions under the corresponding operating condition label, thereby obtaining the set of allowed operational structures. For Ren like but belong ,in, For any candidate execution structure and The adjacency matrix is ​​used as the constraint of the operational structure's credibility domain on the operational structure's state. Simultaneously, the anchor event types from historical samples are summarized to obtain the set of allowed anchor event types. exist Among them The event type identifier is used to switch valve actions, perform soot blowing operations, or change operating modes, and is set in the allowed edge group. The above calculation yields the set of allowed event propagation paths. For the reason A directed path is formed by connecting directed edges in the path. This represents the propagation path from a source measurement point to an affected measurement point, thus operating the structural trust domain. and and Commonly limited labels are By establishing reasonable structured operating state boundaries for the equipment, the control response and state linkage of the supercritical cogeneration unit under specific operating conditions are solidified into a verifiable and credible range in a structural pattern. This enables subsequent diagnostics to compare the real-time constructed operating structure with the credible domain and effectively identify structural state event types or propagation paths that are not allowed by the credible domain. This improves the sensitivity to abnormal linkages and fault symptoms and reduces misjudgments caused by changes in operating conditions.

[0025] S5. Compare the currently constructed operating structure with the operating structure trusted domain under the corresponding working condition label. When the operating structure exceeds the operating structure trusted domain, generate operating structure out-of-bounds information and drive the irreversible diagnostic state machine to perform state evolution based on the operating structure out-of-bounds information. Furthermore, the irreversible diagnostic state machine in step S5 includes normal structural state, structurally interpretable evolution state, structurally anomalous candidate state, and structurally confirmed anomalous state. Furthermore, in step S5, driving the irreversible diagnostic state machine to undergo state evolution includes the following steps: (1) Compare the currently constructed operating structure with the operating structure trust domain under the corresponding operating condition label. When the operating structure exceeds the operating structure trust domain, generate operating structure out-of-bounds information. (2) Based on the out-of-bounds information of the running structure, the irreversible diagnostic state machine evolves from the normal structural state to the candidate state of structural anomaly; (3) When the out-of-bounds information of the running structure is repeatedly detected during continuous operation under the same working condition label constraint, the irreversible diagnostic state machine is made to enter the structural confirmation abnormal state. (4) When a preset reverse anchor event sequence is detected, the irreversible diagnostic state machine is allowed to revert to a structurally interpretable evolutionary state; By adopting the above technical solution, the resulting runtime structure is constructed within the current anchor event window. With adjacency matrix Characterize and match with the corresponding working condition label When performing consistency verification on the trusted domain of the running structure, the first step is to base it on the set of allowed edges in the trusted domain. The ref of the equivalent adjacent matrix unit equipment completes the boundary judgment, among which, This indicates whether the direction of the operational influence from the measuring point to the measuring point is valid within the current window, and whether the actual data of the unit equipment, such as the valve opening command or feedback equipment activation / deactivation status, valve opening / closing status, soot blowing execution status, and operating mode flag, are valid. This indicates that the operating condition label is Whether the direction of influence is allowed to exist, the out-of-bounds information of the running structure is defined as a set. and And its base It reflects the degree of out-of-bounds access of the current operating structure relative to the trusted domain, and will Similar to the type of anchor event in this case Simultaneously serving as an irreversible diagnostic state machine as the driving input, the current state of the state machine is denoted as... ,and The normal structural state can be used to interpret the evolutionary state, the structural anomaly candidate state, or the structural confirmation anomaly state, and an out-of-bounds trigger function can be used to trigger the function. Implement state evolution determination, where, This is an indicator of whether a running structure out of bounds has occurred at the current anchor point event. For indicator functions, when When an out-of-bounds information is written into the diagnostic memory, the state machine transitions from a normal structural state to a structural anomaly candidate state. This identifies unacceptable event propagation paths or unacceptable operational influence directions under the same operating condition label constraints, preventing direct confirmation of anomalies due to a single, occasional disturbance. Subsequently, out-of-bounds checks are repeatedly performed during continuous operation under the same operating condition label constraints, and irreversible confirmation is achieved through the cumulative amount of repeated checks. The cumulative amount is defined as... ,and When the operating condition label changes, it is reset to zero. For the tag is The cumulative number of times that out-of-bounds errors are detected within a continuous running segment and For the sampling interval, when When the preset repetition threshold is exceeded, the state machine enters a structurally confirmed abnormal state to distinguish between occasional and persistent out-of-bounds errors. Persistent out-of-bounds errors are considered suspicious signals consistent with structural deviations caused by equipment degradation, jamming, or logic mismatch. Simultaneously, to ensure the interpretability of rollbacks to normal operation, upon detecting a preset reverse anchor event sequence, the state machine is allowed to roll back to a structurally interpretable evolution state. The reverse anchor event sequence is represented by a sequence... This indicates that its elements are anchor event type identifiers and require that they form a reversible operational closed loop in chronological order with the anchor event type that previously triggered the boundary crossing. This allows the structural transient offset caused by switching, valve action, soot blowing, or change of operating mode to be interpreted as a recoverable evolution rather than being misjudged as a fault. Through the above comparison and irreversible state evolution mechanism, the operating structure boundary crossing can be transformed into an executable diagnostic feather drive signal under the constraint of operating condition label, and a stable confirmation of abnormal output can be given when the boundary crossing continues, thereby improving the robustness and consistency of fault diagnosis of supercritical cogeneration unit equipment under complex operational disturbance backgrounds.

[0026] S6. When the irreversible diagnostic state machine enters the preset structural confirmation abnormality state, it outputs the unit equipment fault diagnosis result and records the diagnostic evidence corresponding to the fault diagnosis result.

[0027] By adopting the above technical solution, when the irreversible diagnostic state machine enters a structurally confirmed abnormal state, it uses the out-of-bounds information of the operating structure under the current operating condition label constraints as the core basis to output the unit equipment fault diagnosis results and simultaneously solidify traceable diagnostic evidence. The fault diagnosis results are defined as the union of the set of affected equipment or measuring points most relevant to the out-of-bounds structure, with the out-of-bounds boundary set as the basis. The input is obtained by endpoint aggregation, i.e. exist belong belong ,in, This refers to the set of equipment or measuring points indicated by the fault diagnosis result, where each equipment or measuring point corresponds one-to-one with the actual data of the unit equipment. It can point to valve actuators and their opening feedback measuring points, equipment engagement / disengagement loop status measuring points, soot blowing execution loop status measuring points, or operating mode indicator measuring points, etc. Simultaneously, diagnostic evidence corresponding to the fault diagnosis result is recorded to support verifiability and engineering implementation. Diagnostic evidence is defined as a set of evidence triplets. ,in, To trigger the confirmation of abnormal operating conditions, The time when the most recent out-of-bounds anchor event occurred. This is the set of operational influence directions that are not allowed in the relatively trusted domain at that moment. This corresponds to the relevant information sequence within the event window, and is controlled by the signal vector. With running state vector This structure binds and stores the valve opening command and feedback process, equipment state transition process, anchor point event type, and out-of-bounds propagation direction within the same time window. Furthermore, it can quantify the out-of-bounds intensity. Used to characterize the degree of structural deviation, and output together with F and EVd, so that operators can directly locate the impact link and corresponding measurement point curve that caused the boundary crossing on a unified platform. With the help of the irreversible mechanism of structural confirmation of abnormal state, the output result is generated only after continuous boundary crossing is repeatedly detected, thereby reducing false alarms caused by single operation disturbances and improving the stability and sufficiency of evidence in the judgment of unit equipment failure.

[0028] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for diagnosing equipment faults in a supercritical combined heat and power unit, characterized in that, Includes the following steps: S1. Obtain relevant information about the supercritical cogeneration unit equipment, and identify the operating status of the unit equipment based on the relevant information to form corresponding operating condition tags. S2. Under the constraints of the operating condition label, identify anchor point events, construct an event window around the anchor point events, and analyze the information in the event window; S3. Based on the response order relationship of the information in the event window, construct an operation structure that reflects the impact relationship of unit equipment operation, which is used to describe the operation structure status of unit equipment under the corresponding operating condition label; S4. Based on historical operation data, construct a trusted domain for each operating condition label. The trusted domain for the operating structure is used to limit the operating structure states that are allowed to exist under the operating condition label. S5. Compare the currently constructed operating structure with the operating structure trusted domain under the corresponding working condition label. When the operating structure exceeds the operating structure trusted domain, generate operating structure out-of-bounds information and drive the irreversible diagnostic state machine to perform state evolution based on the operating structure out-of-bounds information. S6. When the irreversible diagnostic state machine enters the preset structural confirmation abnormality state, it outputs the unit equipment fault diagnosis result and records the diagnostic evidence corresponding to the fault diagnosis result.

2. The method for fault diagnosis of supercritical combined heat and power unit equipment according to claim 1, characterized in that, In step S1, the relevant information of the supercritical cogeneration unit equipment includes measurement data, control signals, and information related to the distribution of heat and power loads during the operation of the unit equipment.

3. The method for fault diagnosis of supercritical combined heat and power unit equipment according to claim 1, characterized in that, In step S1, forming the corresponding operating condition label includes the following steps: (1) Perform time alignment processing on the relevant information to obtain a relevant information sequence under the same time base, and select operating condition feature items to characterize the operating status of the unit equipment from the relevant information sequence to form an operating condition feature set; (2) Discretize the set of operating conditions to obtain the operating condition status corresponding to each operating condition feature item, combine and encode the operating condition status to generate an operating condition label that corresponds one-to-one with the operating status of the unit equipment, and update the current operating condition label when the operating condition label changes.

4. The method for fault diagnosis of supercritical combined heat and power unit equipment according to claim 1, characterized in that, In step S2, identifying anchor point events includes the following steps: (1) Monitor the control signals and changes in operating status in the relevant information; (2) When equipment switching, valve action, soot blowing operation or change of operating mode is detected, the corresponding change in operating status is identified as an anchor event, and the occurrence time and type of the anchor event are recorded.

5. The method for fault diagnosis of supercritical combined heat and power unit equipment according to claim 1, characterized in that, In step S2, constructing the event window around the anchor point event includes the following steps: (1) Taking the time of the anchor event as the center, determine the preset time range before and after the anchor event, and extract the relevant information within the preset time range to form an event window; (2) Use the relevant information in the event window as input data for anchor event impact analysis.

6. The method for fault diagnosis of supercritical combined heat and power unit equipment according to claim 1, characterized in that, In step S3, constructing an operational structure that reflects the operational impact relationships of unit equipment includes the following steps: (1) Within the event window corresponding to the same anchor point event, analyze the order in which different related information generates responses, and determine the direction of the operational influence between related information based on the order of the responses; (2) Based on the aforementioned operational influence direction, construct an operational structure to describe the operational influence relationship of the unit equipment; (3) Determine the event type and make timely corrections based on parameter matching similarity.

7. The method for fault diagnosis of supercritical combined heat and power unit equipment according to claim 1, characterized in that, In step S4, constructing the trusted domain of the runtime structure includes the following steps: (1) Filter out the running segments with stable operating condition labels in the historical operating data, and summarize the operating structure formed within the running segments; (2) Based on the summary results, construct the trusted domain of the operating structure that is allowed to exist under the corresponding operating condition label.

8. The method for fault diagnosis of supercritical combined heat and power unit equipment according to claim 7, characterized in that, The trusted domain of the operational structure includes the set of operational structures that are allowed to exist under the corresponding operating condition label, the set of anchor event types that are allowed to occur, and the set of event propagation paths that are allowed to be established.

9. The method for fault diagnosis of supercritical combined heat and power unit equipment according to claim 1, characterized in that, The irreversible diagnostic state machine in step S5 includes normal structural state, structurally interpretable evolution state, structurally abnormal candidate state, and structurally confirmed abnormal state.

10. A method for diagnosing equipment faults in a supercritical combined heat and power unit according to claim 1, characterized in that, In step S5, driving the irreversible diagnostic state machine to undergo state evolution includes the following steps: (1) Compare the currently constructed running structure with the running structure trust domain under the corresponding working condition label. When the running structure exceeds the running structure trust domain, generate running structure out-of-bounds information. (2) Based on the out-of-bounds information of the running structure, the irreversible diagnostic state machine evolves from the normal structural state to the candidate state of structural anomaly; (3) When the out-of-bounds information of the running structure is repeatedly detected during continuous operation under the same working condition label constraint, the irreversible diagnostic state machine is made to enter the structural confirmation abnormal state. (4) When a preset reverse anchor event sequence is detected, the irreversible diagnostic state machine is allowed to regress to a structurally interpretable evolutionary state.

Images