A method and system for recalling and associating abnormal events of a hydroelectric generating unit and a storage medium

Abstract

The application discloses a kind of hydroelectric generating set abnormal event recollecting association method, system and storage medium, belongs to hydroelectric generating set state monitoring, data fusion, abnormal event analysis and fault diagnosis technical field, method includes and divides hydroelectric generating set operating state data into four levels of original sampling data, transient state data, monitoring state data and summary state data, establishes different storage strategy and unified index;Identify operating condition and establish the correlation between condition event and multilevel state data;When abnormal event occurs, determine the recollection window and establish the correlation between event and data;Then generate a step-by-step recollection path, and fill in the original sampling data that is not stored through the calling mechanism;Finally generate abnormal event data evidence package.The application realizes the automatic association and recollection of multilevel state data before, during and after abnormal event, and provides complete and traceable data support for alarm review, fault diagnosis, trend warning and emergency disposal.

Description

Technical Field

[0001] This invention relates to a method, system, and storage medium for recalling and associating abnormal events in hydropower units, belonging to the technical fields of hydropower unit status monitoring, data fusion, abnormal event analysis, and fault diagnosis. Background Technology

[0002] Hydropower unit operating status data exhibits distinct multi-level, multi-timescale, and multi-source characteristics. Raw sampled data is characterized by high frequency, high capacity, and rich detail, making it suitable for fault diagnosis and characteristic identification, but it is difficult to store for long periods. Transient status data reflects dynamic processes and is suitable for analyzing changes in operating conditions and transient responses. Monitoring status data reflects short-term trends and is suitable for operational monitoring. Summary status data reflects long-term trends and is suitable for long-term statistical analysis.

[0003] Existing hydropower station monitoring systems typically focus on only one level of data. For example, monitoring systems focus on data at the second or ten-second level, condition monitoring systems focus on vibration, sway, or waveform data, and production management systems focus on minute or hourly statistical data. The lack of a unified index and correlation between these data points means that after an anomaly occurs, operators often need to manually search for data in multiple systems, making it difficult to promptly reconstruct the operational status before and after the event.

[0004] Especially in scenarios with centralized intelligent alarm systems, abnormal events may involve multiple plants, systems, devices, and measurement points. If events cannot be correlated with status data at different levels, it becomes difficult to achieve alarm verification, event recall, fault diagnosis, and handling guidance. Therefore, a method is needed that can hierarchically organize raw sampled data, transient status data, monitoring status data, and summary status data, and establish recall correlations around operating conditions and abnormal events. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of the prior art and provide a method, system and storage medium for recalling and associating abnormal events of hydropower units. This is to solve the problem that it is difficult to uniformly organize, quickly trace and comprehensively analyze the status data of multiple sources, multiple time scales and multiple data levels around abnormal events during the operation of hydropower units.

[0006] To achieve the above objectives, the present invention is implemented using the following technical solution:

[0007] In a first aspect, the present invention provides a method for recalling and associating abnormal events in hydropower units, comprising:

[0008] Acquire the status data generated during the operation of the hydropower unit, and divide the status data into multiple levels including raw sampling data, transient status data, monitoring status data and summary status data according to the data formation mechanism and time scale. Establish storage strategies and unified indexes for each level of status data.

[0009] Based on the pre-acquired unit operating parameters, the operating conditions are identified and operating condition events are generated. Based on the time range of the operating condition events, the operating condition events are associated with the status data of each level within the corresponding time period.

[0010] When an abnormal event is detected, an abnormal event object is generated and a recall window is determined;

[0011] Based on the abnormal event object, the recall window, and the unified index, relevant status data is retrieved, and an abnormal event-data association relationship is established between the abnormal event and the status data.

[0012] Based on the type of abnormal event, a hierarchical recall path is generated from summary status data to monitoring status data, from monitoring status data to transient status data, and from transient status data to raw sampled data. Status data at each level is extracted according to the hierarchical recall path. If the required raw sampled data is not stored, a data retrieval request is issued and the successfully retrieved raw sampled data is bound to the abnormal event identifier.

[0013] An abnormal event data evidence package is generated based on the abnormal event object, the operating condition-data association, the abnormal event-data association, and the hierarchical recall path for alarm review, fault diagnosis, trend warning, or emergency response.

[0014] Furthermore, the original sampling data is the continuous output signal sampling data of the sensor, the transient state data is the data that can reflect the dynamic process of the state extracted based on the original sampling data, the monitoring state data is the short-cycle state data formed according to the monitoring cycle, and the summary state data is the maximum value, minimum value, average value or other statistical value formed according to the statistical cycle.

[0015] And / or, the storage strategy includes: raw sampling data is stored using short-cycle waveform recording or recall-based storage, transient state data and monitoring state data are stored using a real-time database, and summary state data is stored using a relational database or analytical database;

[0016] And / or, the unified index is represented as:

[0017] I=(S,U,D,C,L i ,t,Δt s (,Q,P);

[0018] Where I represents the unified index of a certain status data, S represents the plant code, U represents the unit code, D represents the equipment code, C represents the unified measurement point code, and L represents the station code. i Δt represents the data level, t represents the data time, and Δt represents the data time. s Q represents the sampling period, Q represents the data quality identifier, and P represents the data storage path.

[0019] Furthermore, the unit operating parameters include at least one or more of the following: active power, unit speed, guide vane opening, head, flow rate, and start-up / shutdown status; the operating condition events include at least the operating condition identifier, plant code, unit code, operating condition type, start time, and end time.

[0020] For the operating condition event G j and state data fragment X k The degree of temporal overlap is expressed as:

[0021] ;

[0022] Among them, O j,k t represents the time overlap between the j-th condition event and the k-th state data segment. g,s and t g,e Let t represent the start and end times of the operating condition event, respectively. x,s and t x,e These represent the start and end times of the state data segment, respectively.

[0023] When O j,k When the value is greater than a preset threshold O0, it is determined that the state data segment has a valid correlation with the operating condition event. For point value state data, if its data time t satisfies t g,s ≤t≤t g,e Then, the point value class status data will be associated with the corresponding working condition event;

[0024] And / or, the operating condition-data association relationship includes at least the operating condition identifier, measurement point code, equipment code, data level, data start time, data end time, data storage path, and association status.

[0025] Furthermore, the abnormal events include at least one of the following: status exceeding limit events, performance index exceeding limit events, trend abnormal events, protection action events, alarm signal triggering events, and data quality abnormal events; the abnormal event object includes at least an event identifier, plant code, unit code, equipment code, event type, event level, event start time, and event end time.

[0026] The recall window for the m-th anomalous event at the i-th data level is represented as follows:

[0027] Wm,i =[t e,s −τ pre (m,i),t e,e +τ post (m,i)];

[0028] Among them, W m,i t represents the recall window for the m-th anomalous event at the i-th data level. e,s Indicates the start time of the event, t e,e τ represents the end time of the event. pre (m,i) represents the duration of recollection before the event, τ post (m,i) represents the duration of post-event recollection;

[0029] For temperature-related anomalies, a long window is used; for vibration and swing-related anomalies, a short-time, high-frequency window is used; and for protection action anomalies, a recording window before and after the action is used.

[0030] Furthermore, the event correlation degree is expressed as:

[0031] A m,k =λ1O m,k +λ2R m,k +λ3Q k +λ4H k ;

[0032] Among them, A m,k This represents the correlation between the m-th anomalous event and the k-th state data segment, O m,k R represents the degree of temporal overlap. m, k represents the correlation score of the measuring point or equipment, Q k H represents the data quality score. k The data source reliability score is represented by λ1, λ2, λ3, and λ4, which represent the weights of each item and satisfy λ1+λ2+λ3+λ4=1.

[0033] When A m,k When the value is greater than a preset association threshold A0, a strong association is established between the state data fragment and the abnormal event; when A... m,k When the value is below A0 but above the candidate threshold A1, the state data fragment is listed as candidate associated data; when A... m,k If the value is below A1, it will not be included in the event recall data range; where A0 and A1 are preset thresholds, and A0>A1;

[0034] And / or, the abnormal event-data association relationship includes at least the event identifier, data level, measurement point code, device code, data time range, data storage path, association degree, data quality identifier, and association status.

[0035] Furthermore, the step-by-step recall path is represented as follows:

[0036] Path(E m =L3→L2→L1→L0;

[0037] Among them, Path(E m L1 represents the recall path of the mm-th abnormal event, L2 represents the summary state data layer, L1 represents the monitoring state data layer, L0 represents the transient state data layer, and L0 represents the raw sampled data layer.

[0038] The step-by-step recall process includes: first, using summary state data to determine the macro-level time period of the anomaly; then, using monitoring state data to determine the start and end times of the anomaly; then, using transient state data to analyze the dynamic change process; and finally, using raw sampled data to analyze the details of the anomaly. For events that do not require calling all levels of data, the recall path is automatically truncated based on the event type and correlation.

[0039] During the step-by-step recall process, if the required original sampling data is not stored on the platform for a long time, a data recall request is sent to the special monitoring unit according to the abnormal event identifier and the recall window. After the recall is successful, the original sampling data is bound with the abnormal event identifier and the abnormal event-data association relationship is updated.

[0040] Furthermore, the data retrieval request includes at least an event identifier, plant code, unit code, equipment code, measurement point code, request start time, request end time, data type, sampling frequency, and return format. After the data source returns the data, the returned data undergoes integrity verification, time continuity verification, unit verification, and data quality identification, and is bound to an abnormal event identifier. If the data retrieval is successful, the retrieved data is written to the event temporary data warehouse and the abnormal event-data association is updated. If the data retrieval fails, the reason for the failure is recorded and the missing original sampled data is indicated in the event evidence package.

[0041] And / or, the abnormal event data evidence package includes at least the following: basic event information, event triggering information, event-related operating conditions, event-related measurement points, pre-event data, in-event data, post-event data, summary status data, monitoring status data, transient status data, original sampling data index, data quality information, recall path, data missing status, correlation calculation results, and storage path information.

[0042] Furthermore, the method also includes a data transfer step: after an abnormal event is confirmed, reviewed, or a report is generated, the corresponding data evidence package is transferred from the temporary data warehouse to the permanent data warehouse; during the transfer process, the event identifier, data index, data level, original storage path, transfer time, transfer personnel or transfer task, data verification results, and version information are retained; for the original sampled data, the file path, summary information, and index information are saved.

[0043] And / or, the method further includes a recall rule update step: receiving confirmation information from operators or professionals regarding the results of abnormal event analysis; updating the correspondence between event types and recall windows, the correspondence between events and associated measurement points, event correlation weights, recall path templates, and data retrieval strategies based on the confirmation information; for event types that occur repeatedly but for which recall data is insufficient, increasing the recall window or raising the priority of original sampled data retrieval; and for measurement points that are mistakenly associated, reducing their measurement point correlation weights.

[0044] Secondly, the present invention provides a hydropower unit abnormal event tracing and correlation system for implementing the hydropower unit abnormal event tracing and correlation method described in any of the preceding claims, comprising:

[0045] The data acquisition and segmentation module is used to acquire the status data generated during the operation of the hydropower unit, and to divide the status data into multiple levels including raw sampling data, transient status data, monitoring status data and summary status data according to the data formation mechanism and time scale, and to establish storage strategies and unified indexes for each level of status data.

[0046] The first association establishment module is used to identify operating conditions and generate operating condition events based on the pre-acquired unit operating parameters, and establish operating condition-data association relationships between the operating condition events and the status data of each level within the corresponding time period based on the time range of the operating condition events.

[0047] The first generation module is used to generate an abnormal event object and determine the recall window when an abnormal event is detected.

[0048] The second association establishment module is used to retrieve relevant status data based on the abnormal event object, the recall window and the unified index, and establish an abnormal event-data association relationship between the abnormal event and the status data.

[0049] The recall path generation module is used to generate a hierarchical recall path from summary status data to monitoring status data, from monitoring status data to transient status data, and from transient status data to original sampled data according to the type of abnormal event. The module extracts status data at each level according to the hierarchical recall path. If the required original sampled data is not stored, a data retrieval request is issued and the successfully retrieved original sampled data is bound to the abnormal event identifier.

[0050] The second generation module is used to generate an abnormal event data evidence package for alarm review, fault diagnosis, trend warning or emergency response based on the abnormal event object, the working condition-data association relationship, the abnormal event-data association relationship and the hierarchical recall path.

[0051] Thirdly, the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of any of the methods described above.

[0052] Fourthly, the present invention provides an electronic device, comprising:

[0053] Memory, used to store computer programs / instructions;

[0054] A processor for executing the computer program / instructions to implement the steps of any of the methods described above.

[0055] Fifthly, the present invention provides a computer program product, including a computer program / instructions that, when executed by a processor, implement the steps of any of the methods described above.

[0056] Compared with the prior art, the beneficial effects achieved by the present invention are as follows:

[0057] This invention provides a method, system, and storage medium for reconstructing and associating abnormal events in hydropower units. It establishes a multi-level data model comprising raw sampled data, transient state data, monitoring state data, and summary state data, and configures differentiated storage strategies. A unified index containing data level identifiers, time information, and storage location information is established. The system establishes operating condition-data associations between operating events and multi-level state data. Furthermore, when an abnormal event occurs, a reconstructing window is determined, and an abnormal event-data association is established between the abnormal event and the state data. This process generates data from summary state data to monitoring state data, from monitoring state data to transient state data, and from transient state data to raw sampled data. The system traces the path of sample data step by step and extracts state data at each level according to this path. Finally, it generates an abnormal event data evidence package based on the abnormal event object, the working condition-data correlation, the abnormal event-data correlation, and the step-by-step tracing results. This realizes the automatic association, automatic tracing, and unified encapsulation of multi-level state data before, during, and after an abnormal event. It solves the problem of data at different time scales being unable to be analyzed collaboratively. This allows data to be stored not only by measurement point but also by operating condition and abnormal event, realizing an automated process of abnormal analysis from coarse to fine and from macro to micro. It provides complete, reliable, and traceable data support for alarm review, fault diagnosis, trend early warning, and emergency response. Attached Figure Description

[0058] Figure 1 This is a flowchart of a method for recalling and associating abnormal events of hydropower units provided in an embodiment of the present invention. Detailed Implementation

[0059] The present invention will be further described below with reference to the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and should not be used to limit the scope of protection of the present invention.

[0060] Example 1: This example introduces a method for recalling and associating abnormal events in hydropower units, including:

[0061] Acquire the status data generated during the operation of the hydropower unit, and divide the status data into multiple levels including raw sampling data, transient status data, monitoring status data and summary status data according to the data formation mechanism and time scale. Establish storage strategies and unified indexes for each level of status data.

[0062] Based on the pre-acquired unit operating parameters, the operating conditions are identified and operating condition events are generated. Based on the time range of the operating condition events, the operating condition events are associated with the status data of each level within the corresponding time period.

[0063] When an abnormal event is detected, an abnormal event object is generated and a recall window is determined;

[0064] Based on the abnormal event object, the recall window, and the unified index, relevant status data is retrieved, and an abnormal event-data association relationship is established between the abnormal event and the status data.

[0065] Based on the type of abnormal event, a hierarchical recall path is generated from summary status data to monitoring status data, from monitoring status data to transient status data, and from transient status data to raw sampled data. Status data at each level is extracted according to the hierarchical recall path. If the required raw sampled data is not stored, a data retrieval request is issued and the successfully retrieved raw sampled data is bound to the abnormal event identifier.

[0066] An abnormal event data evidence package is generated based on the abnormal event object, the operating condition-data association, the abnormal event-data association, and the hierarchical recall path for alarm review, fault diagnosis, trend warning, or emergency response.

[0067] like Figure 1 As shown in the figure, the application process of the hydropower unit abnormal event recall and association method provided in this embodiment involves the following steps:

[0068] Step S1: Establish a hierarchical model for the state data of hydropower units;

[0069] First, multi-source status data generated during the operation of the hydropower unit is acquired. Based on the data formation mechanism, time scale, data accuracy, storage period, and application purpose, the status data is divided into four levels: raw sampling data, transient status data, monitoring status data, and summary status data.

[0070] Raw sampling data refers to high-frequency raw data directly collected by sensors, condition monitoring devices, waveform recording devices, etc., including vibration waveforms, swing waveforms, current and voltage waveforms, pressure pulsation waveforms, etc. This type of data contains rich details and is suitable for fault mechanism analysis, feature extraction and abnormal detail verification, but the data volume is large and it is not suitable for long-term continuous full storage.

[0071] Transient state data refers to dynamic process data extracted or calculated based on the original sampled data according to a shorter period, including short-period frequency, guide vane opening, speed change, peak-to-peak vibration, peak-to-peak swing, pressure pulsation characteristic value, etc. This type of data is used to reflect the dynamic response characteristics of the unit during start-up and shutdown, load adjustment, operating condition switching, and abnormal disturbances.

[0072] Monitoring status data refers to operational monitoring data generated according to the sampling cycle of computer monitoring systems, condition monitoring systems, or real-time databases. This includes active power, reactive power, head, flow rate, bearing temperature, cooling water pressure, cooling water flow rate, unit vibration, unit swing, and switch status. This type of data is used to support operational monitoring, alarm judgment, trend analysis, and short-cycle status identification.

[0073] Summary status data refers to statistical data calculated from monitoring status data according to statistical periods such as minutes, quarter minutes, hours, and days. It includes maximum value, minimum value, average value, standard deviation, duration, number of times the limit is exceeded, and statistical values ​​of change rate. This type of data is used to support long-term trend overview, operation evaluation, deterioration analysis, and historical comparison.

[0074] The above data hierarchy can be represented as:

[0075] L={L0,L1,L2,L3}

[0076] Where L represents the set of hydropower unit status data levels, L0 represents the raw sampled data layer, L1 represents the transient status data layer, L2 represents the monitoring status data layer, and L3 represents the summary status data layer.

[0077] Through the aforementioned hierarchical model, the system can switch orderly between different time scales and different data precisions according to the needs of abnormal event analysis.

[0078] Step S2: Establish a multi-level hierarchical storage strategy for state data;

[0079] Establish differentiated storage strategies based on the data volume, call frequency, retention period, and business value of different levels of status data.

[0080] For raw sampling data, a combination of short-cycle caching, event-triggered saving, on-demand retrieval, and file indexing is used for storage. Under normal operating conditions, raw sampling data can be stored in dedicated monitoring devices or edge acquisition units, retaining only the data from the most recent period. When an abnormal event, alarm event, or manual analysis task occurs, a data retrieval request is initiated to the corresponding device according to the event time, transferring the raw sampling data within the specified time range before and after the event to the platform.

[0081] For transient state data, a real-time database or a high-performance time-series database is used for storage to retain a high time resolution, which is used to support the analysis of operating condition changes, abnormal dynamic processes, and response processes before and after events.

[0082] For monitoring status data, a real-time database is used for storage, and indexes are established according to measurement point code, equipment object, plant, and data time to support real-time monitoring, alarm judgment, short-cycle trend analysis, and event replay.

[0083] For summary status data, relational databases, analytical databases, or data warehouses are used for storage. Indexes are built according to plant, unit, equipment, measurement point, and statistical period to support long-term trend queries, report analysis, and rapid location of abnormal time periods.

[0084] In practical applications, each data record should at least retain the unified measurement point code, equipment code, plant code, data level, data time, access time, data value, unit, data source, data quality identifier, storage location, and index version number.

[0085] Step S3: Establish a unified index for multi-level status data;

[0086] To enable rapid association between data from different levels, sources, and storage locations, a unified index model is established. Each piece of status data is described through a unified index, which includes at least the plant / station code, unit code, equipment code, measurement point code, data level, data time, sampling period, data source, operating condition identifier, event identifier, data quality identifier, and storage path.

[0087] A unified index can be represented as:

[0088] I=(S,U,D,C,L i ,t,Δt s (Q,P)

[0089] Where I represents the unified index of a certain status data, S represents the plant code, U represents the unit code, D represents the equipment code, C represents the unified measuring point code, and L represents the station code. i Δt represents the data level, t represents the data time, and Δt represents the data time. s Q represents the sampling period, Q represents the data quality identifier, and P represents the data storage path.

[0090] With a unified index, even if the original sampling data, transient status data, monitoring status data, and summary status data are stored in different databases or file systems, they can still be queried, located, and correlated around the same device, the same measurement point, the same operating condition, or the same abnormal event.

[0091] Step S4: Identify the unit's operating conditions and generate operating condition events;

[0092] The system identifies the unit's operating conditions based on the unit's operating parameters and generates operating condition events. These operating parameters include unit start-up / shutdown status, active power, reactive power, speed, guide vane opening, head, flow rate, and load change rate.

[0093] Operating events include at least the operating condition identifier, plant code, unit code, operating condition type, start time, end time, duration, average load, average head, load change rate, and operating condition stability identifier.

[0094] Operating condition types can include shutdown conditions, startup conditions, grid connection conditions, stable operation conditions, low load conditions, medium load conditions, high load conditions, rapid load change conditions, and shutdown conditions. The system can be configured with operating condition classification rules according to the actual needs of the project.

[0095] For any operating condition event G j Its time range can be expressed as:

[0096] G j =[t g,s ,t g,e ]

[0097] Among them, G j Let t represent the j-th operating condition event. g,s Indicates the start time of this operating condition event, t g,e This indicates the end time of the event in this condition.

[0098] By using operating condition events, the unit status data can be analyzed within specific operating conditions, avoiding the bias in anomaly judgment caused by mixing data from different operating conditions.

[0099] Step S5: Establish the working condition-data relationship;

[0100] Based on the time range of the operating condition event and the timestamp of the status data, establish a correlation between the operating condition event and the multi-level status data within the corresponding time period.

[0101] For a certain working condition event G j and a certain state data fragment X k The degree of temporal overlap can be expressed as:

[0102]

[0103] Among them, O j,k t represents the time overlap between the j-th condition event and the k-th state data segment. g,s and t g,e Let t represent the start and end times of the operating condition event, respectively. x,s and t x,e These represent the start and end times of the state data segment, respectively. When O j,k When the value is greater than the preset threshold O0, it is determined that the state data segment is effectively associated with the working condition event.

[0104] For point-value state data, if its data time t satisfies:

[0105] t g,s ≤t≤t g,e

[0106] Then, the point value data is associated with the corresponding operating condition event.

[0107] The operating condition-data association relationship should include at least the operating condition identifier, measurement point code, equipment code, data level, data start time, data end time, data storage path, and association status. This association relationship enables the automatic extraction of status data under the same operating condition during subsequent anomaly analysis, facilitating data recall and comparative analysis based on operating conditions.

[0108] Step S6: Identify or receive abnormal events and generate abnormal event objects;

[0109] It can receive or identify abnormal events during the operation of hydropower units in real time. The sources of abnormal events include alarm signal triggering, analog quantity exceeding limits, trend warning, protection action, sudden state change, data quality anomaly, and manually marked events.

[0110] Each abnormal event generates an abnormal event object, which includes at least the event identifier, plant code, unit code, equipment code, event type, event level, event source, event start time, event end time, triggering measurement point, associated measurement point, trigger value, alarm setpoint, event status, and processing status.

[0111] The time range of abnormal events can be represented as:

[0112] E m =[t e,s ,t e,e ]

[0113] Among them, E m Let t represent the m-th abnormal event. e,s Indicates the start time of the abnormal event, t e,e This indicates the end time of the abnormal event. For transient events, if the system only obtains the trigger time t... e Then the event time range is initially set to:

[0114] Em=[t e ,t e ]

[0115] The event analysis window is expanded based on the event type and changes in related data during subsequent recollection.

[0116] Step S7: Determine the abnormal event dynamic recall window;

[0117] The recall window for an abnormal event is dynamically determined based on the type of abnormal event, the device object, the type of measurement point, the data level, and the purpose of analysis. The recall window includes a pre-event window, an event-in-event window, and a post-event window.

[0118] The recall window for the m-th anomalous event at the i-th data level can be represented as:

[0119] W m,i =[t e,s −τ pre (m,i),t e,e +τ post (m,i)]

[0120] Among them, W m,i t represents the recall window for the m-th anomalous event at the i-th data level. e,s Indicates the start time of the event, t e,e τ represents the end time of the event. pre (m,i) represents the duration of recollection before the event, τ post (m,i) represents the duration of the recollection after the event.

[0121] For temperature-related anomalies, the pre-event recall duration can be set to be longer to analyze the slow temperature rise process; for vibration, sway, and pressure pulsation anomalies, the pre-event recall duration can be set to be shorter but the data level can be finer to analyze transient fluctuations and dynamic responses; for protection action and waveform recording anomalies, the recall window should cover the original sampling data or waveform recording data before and after the action; for trend warning anomalies, the recall window should cover a longer time range to analyze gradual trends.

[0122] By using a dynamic recall window, the problems of insufficient or redundant data caused by using a fixed window for all events can be avoided.

[0123] Step S8: Establish abnormal events - multi-level data association;

[0124] The system automatically retrieves multi-level status data related to events based on the abnormal event object, the recall window, and the unified index, and establishes an abnormal event-data association relationship.

[0125] For a certain abnormal event E m and a certain state data fragment X k The correlation degree of events is calculated from four aspects: time overlap, correlation of measurement points, data quality, and reliability of source. The correlation degree of events can be expressed as:

[0126] A m,k =λ1O m,k +λ2R m,k +λ3Q k +λ4H k

[0127] Among them, A m,k This represents the correlation between the m-th anomalous event and the k-th state data segment, O m,k R represents the degree of temporal overlap. m, k represents the correlation score of the measuring point or equipment, Q k H represents the data quality score. k The data source reliability score is represented by λ1, λ2, λ3, and λ4, which represent the weights of each item and satisfy λ1+λ2+λ3+λ4=1.

[0128] When A m,k When the value is greater than the preset association threshold A0, a strong association is established between the state data fragment and the abnormal event; when A... m,k When the value is below A0 but above the candidate threshold A1, the state data segment is listed as candidate associated data; when A... m,k If the value is below A1, it will not be included in the event recall data. Here, A0 and A1 are preset thresholds, and A0 > A1.

[0129] An abnormal event-data association relationship should include at least the event identifier, data level, measurement point code, device code, data time range, data storage path, correlation degree, data quality identifier, and association status.

[0130] Step S9: Generate a progressively detailed recall path from coarse to fine;

[0131] Based on the type of abnormal event and the associated data, a hierarchical recall path is generated. The hierarchical recall path proceeds in the following order: from summary status data to monitoring status data, from monitoring status data to transient status data, and from transient status data to raw sampled data.

[0132] For long-term trend anomalies, first call the summary status data to locate the macro time period in which the anomaly occurred, then call the monitoring status data to determine the start and end times and change process of the anomaly, and if necessary call the transient status data or raw sampled data to analyze the anomaly formation process.

[0133] For short-term mutation anomalies, first call the monitoring status data to determine the time of the mutation, then call the transient status data to analyze the mutation process, and finally call the original sampling data or waveform data as needed.

[0134] For anomalies related to protection actions, priority is given to calling protection action records and waveform data. At the same time, the monitoring status data and transient status data before and after the action are correlated to analyze the changes in the operating status before and after the protection action.

[0135] The path of recollection can be represented as:

[0136] Path(E m =L3→L2→L1→L0

[0137] Among them, Path(E m L1 represents the recall path of the mm-th abnormal event, L2 represents the summary state data layer, L1 represents the monitoring state data layer, L2 represents the transient state data layer, and L0 represents the raw sampled data layer.

[0138] For events that do not require access to all levels of data, the recall path is automatically truncated based on the event type and relevance. For example, for general temperature trend warning events, the recall path can be traced back to the monitoring status data layer; for vibration sudden changes or protection action events, the recall path can be traced back to the original sampled data layer.

[0139] Step S10: Perform raw sampling data retrieval and data completion;

[0140] When the abnormal event recall path requires access to the original sampling data, but the platform does not store the corresponding original sampling data for a long period of time, the system sends a data recall request to the special monitoring device, waveform recording device or edge data unit based on the event identifier, device code, measurement point code and recall window.

[0141] A data request must include at least the event identifier, plant / station code, unit code, equipment code, measurement point code, request start time, request end time, data type, sampling frequency, and return format. After the data source returns the data, the system performs integrity verification, time continuity verification, unit verification, and data quality identification on the returned data, and binds it to the abnormal event identifier.

[0142] If data retrieval is successful, the system will write the retrieved data to the event temporary data warehouse and update the abnormal event-data association. If data retrieval fails, the system will record the reason for the failure, including data non-existence, device offline, communication interruption, insufficient permissions, time range exceeding limits, etc., and mark the original sampled data as missing in the event evidence package.

[0143] Step S11: Generate an abnormal event data evidence package;

[0144] The system generates an abnormal event data evidence package based on the abnormal event object, the working condition-data correlation, the event-data correlation, and the hierarchical recall results. This data evidence package is a structured collection of data organized around the abnormal event, used to support alarm review, fault diagnosis, trend analysis, emergency response, and report generation.

[0145] An abnormal event data evidence package should include at least the following: basic event information, event triggering information, event-related operating conditions, event-related measurement points, pre-event data, in-event data, post-event data, summary status data, monitoring status data, transient status data, raw sampling data index, data quality information, recall path, data missing information, correlation calculation results, and storage path information.

[0146] For alarm review, the evidence package can be used to determine whether the alarm is true and valid; for fault diagnosis, the evidence package can be used to analyze the state evolution process before and after the fault occurs; for trend early warning, the evidence package can be used to collect trend anomaly samples; for emergency response, the evidence package can be used to match response plans and generate event reports.

[0147] Step S12: Establish a transfer mechanism between the temporary data warehouse and the permanent data warehouse;

[0148] Based on the status of the abnormal event and business needs, the system determines whether to transfer event-related data from the temporary data warehouse to the permanent data warehouse. When an abnormal event is confirmed by operators, generates an analysis report, triggers emergency response, is marked as a typical event, or is used as a model training sample, the system transfers the corresponding data evidence package from the temporary data warehouse to the permanent data warehouse.

[0149] During the transfer process, the system retains event identifiers, data indexes, data levels, original storage paths, transfer times, transfer personnel or tasks, data verification results, and version information. For large volumes of original sampled data, the system can save file paths, summary information, and index information without converting all data into relational data.

[0150] Through the transfer mechanism, the system can control the long-term storage pressure of high-frequency data, and ensure that typical abnormal events, effective alarm events and important fault events have long-term traceability capabilities.

[0151] Step S13: Update the event recall rules based on the results of manual confirmation;

[0152] The system receives confirmation information from operators, maintenance personnel, or professionals regarding the results of abnormal event analysis. Confirmation information includes whether the event is valid, the cause of the event, the accuracy of the associated measurement points, the adequacy of the recall window, whether the original sampling data needs to be supplemented, and whether the handling plan is appropriate.

[0153] The system updates the correspondence between event types and recall windows, the correspondence between events and associated measurement points, the event correlation weight, the recall path template, and the data retrieval strategy based on the results of manual confirmation. For event types that occur repeatedly but for which recall data is insufficient, the system increases the recall window or raises the priority of recalling original sampled data; for measurement points with many false associations, the system lowers their measurement point correlation weight; for confirmed typical events, the system includes their evidence packages in the event sample library.

[0154] Through the above feedback mechanism, the abnormal event recall method can be continuously optimized as operational experience and historical events accumulate.

[0155] Compared with the prior art, this embodiment has the following advantages and effects:

[0156] First, a four-layer data model is established, consisting of raw sampling, transient state, monitoring state, and summary state, to solve the problem of data from different time scales being unable to be analyzed collaboratively.

[0157] Second, a dual correlation mechanism of operating condition-data and event-data is proposed, so that data can be stored not only by measurement point, but also by operating condition and abnormal event.

[0158] Third, it proposes a step-by-step recall path from summary data to high-frequency raw data, realizing an automated process of anomaly analysis from coarse to fine and from macro to micro.

[0159] Example 2: This example provides a hydropower unit abnormal event tracking and correlation system, including:

[0160] The data acquisition and segmentation module is used to acquire the status data generated during the operation of the hydropower unit, and to divide the status data into multiple levels including raw sampling data, transient status data, monitoring status data and summary status data according to the data formation mechanism and time scale, and to establish storage strategies and unified indexes for each level of status data.

[0161] The first association establishment module is used to identify operating conditions and generate operating condition events based on the pre-acquired unit operating parameters, and establish operating condition-data association relationships between the operating condition events and the status data of each level within the corresponding time period based on the time range of the operating condition events.

[0162] The first generation module is used to generate an abnormal event object and determine the recall window when an abnormal event is detected.

[0163] The second association establishment module is used to retrieve relevant status data based on the abnormal event object, the recall window and the unified index, and establish an abnormal event-data association relationship between the abnormal event and the status data.

[0164] The recall path generation module is used to generate a hierarchical recall path from summary status data to monitoring status data, from monitoring status data to transient status data, and from transient status data to original sampled data according to the type of abnormal event. The module extracts status data at each level according to the hierarchical recall path. If the required original sampled data is not stored, a data retrieval request is issued and the successfully retrieved original sampled data is bound to the abnormal event identifier.

[0165] The second generation module is used to generate an abnormal event data evidence package for alarm review, fault diagnosis, trend warning or emergency response based on the abnormal event object, the working condition-data association relationship, the abnormal event-data association relationship and the hierarchical recall path.

[0166] The specific functions of each module described above are explained in the relevant content of the method in Embodiment 1, and will not be repeated here.

[0167] Example 3: This example provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of any of the methods described in Example 1.

[0168] Example 4: This example provides an electronic device, including:

[0169] Memory, used to store computer programs / instructions;

[0170] A processor for executing the computer program / instructions to implement the steps of any of the methods described in Embodiment 1.

[0171] Example 5: This example provides a computer program product, including a computer program / instructions, which, when executed by a processor, implement the steps of the method described in any one of Examples 1.

[0172] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the technical principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

[0173] Those skilled in the art will understand that embodiments of this disclosure can be provided as methods, systems, or computer program products. Therefore, this disclosure can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this disclosure can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0174] This disclosure is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this disclosure. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create a machine for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0175] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0176] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0177] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this disclosure and not to limit its protection scope. Although this disclosure has been described in detail with reference to the above embodiments, those skilled in the art should understand that after reading this disclosure, they can still make various changes, modifications or equivalent substitutions to the specific implementation of the invention, but these changes, modifications or equivalent substitutions are all within the protection scope of the pending claims.

Claims

1. A method for recalling and associating abnormal events in hydropower units, characterized in that, include: Acquire the status data generated during the operation of the hydropower unit, and divide the status data into multiple levels including raw sampling data, transient status data, monitoring status data and summary status data according to the data formation mechanism and time scale. Establish storage strategies and unified indexes for each level of status data. Based on the pre-acquired unit operating parameters, the operating conditions are identified and operating condition events are generated. Based on the time range of the operating condition events, the operating condition events are associated with the status data of each level within the corresponding time period. When an abnormal event is detected, an abnormal event object is generated and a recall window is determined; Based on the abnormal event object, the recall window, and the unified index, relevant status data is retrieved, and an abnormal event-data association relationship is established between the abnormal event and the status data. Based on the type of abnormal event, a hierarchical recall path is generated from summary status data to monitoring status data, from monitoring status data to transient status data, and from transient status data to raw sampled data. Status data at each level is extracted according to the hierarchical recall path. If the required raw sampled data is not stored, a data retrieval request is issued and the successfully retrieved raw sampled data is bound to the abnormal event identifier. An abnormal event data evidence package is generated based on the abnormal event object, the operating condition-data association, the abnormal event-data association, and the hierarchical recall path for alarm review, fault diagnosis, trend warning, or emergency response.

2. The method for recalling and associating abnormal events of hydropower units according to claim 1, characterized in that, The original sampling data is the continuous output signal sampling data of the sensor; the transient state data is the data extracted based on the original sampling data that can reflect the dynamic process of the state; the monitoring state data is the short-cycle state data formed according to the monitoring cycle; and the summary state data is the maximum value, minimum value, average value or other statistical value formed according to the statistical cycle. And / or, the storage strategy includes: raw sampling data is stored using short-cycle waveform recording or recall-based storage, transient state data and monitoring state data are stored using a real-time database, and summary state data is stored using a relational database or analytical database; And / or, the unified index is represented as: I=(S,U,D,C,L i ,t,Δt s ,Q,P); Where I represents the unified index of a certain status data, S represents the plant code, U represents the unit code, D represents the equipment code, C represents the unified measuring point code, and L represents the station code. i Δt represents the data level, t represents the data time, and Δt represents the data time. s Q represents the sampling period, Q represents the data quality identifier, and P represents the data storage path.

3. The method for recalling and associating abnormal events of hydropower units according to claim 1, characterized in that, The unit operating parameters include at least one or more of the following: active power, unit speed, guide vane opening, head, flow rate, and start / stop status; the operating condition events include at least the operating condition identifier, plant code, unit code, operating condition type, start time, and end time. For the operating condition event G j and state data fragment X k The degree of temporal overlap is expressed as: ； Among them, O j,k t represents the time overlap between the j-th condition event and the k-th state data segment. g,s and t g,e Let t represent the start and end times of the operating condition event, respectively. x,s and t x,e These represent the start and end times of the state data segment, respectively. When O j,k When the value is greater than a preset threshold O0, it is determined that the state data segment has a valid correlation with the operating condition event. For point value state data, if its data time t satisfies t g,s ≤t≤t g,e Then, the point value class status data will be associated with the corresponding working condition event; And / or, the operating condition-data association relationship includes at least the operating condition identifier, measurement point code, equipment code, data level, data start time, data end time, data storage path, and association status.

4. The method for recalling and associating abnormal events of hydropower units according to claim 1, characterized in that, The abnormal events include at least one of the following: status exceeding limit events, performance index exceeding limit events, trend abnormal events, protection action events, alarm signal triggering events, and data quality abnormal events; the abnormal event object includes at least the event identifier, plant code, unit code, equipment code, event type, event level, event start time, and event end time. The recall window for the m-th anomalous event at the i-th data level is represented as follows: W m,i =[t e,s −τ pre (m,i),t e,e +τ post (m,i)]; Among them, W m,i t represents the recall window for the m-th anomalous event at the i-th data level. e,s Indicates the start time of the event, t e,e τ represents the end time of the event. pre (m,i) represents the duration of recollection before the event, τ post (m,i) represents the duration of post-event recollection; For temperature-related anomalies, a long window is used; for vibration and swing-related anomalies, a short-time, high-frequency window is used; and for protection action anomalies, a recording window is used before and after the action.

5. The method for recalling and associating abnormal events of hydropower units according to claim 1, characterized in that, The event correlation degree is expressed as follows: A m,k =λ1O m,k +λ2R m,k +λ3Q k +λ4H k ; Among them, A m,k This represents the correlation between the m-th anomalous event and the k-th state data segment, O m,k R represents the degree of temporal overlap. m, k represents the correlation score of the measuring point or equipment, Q k H represents the data quality score. k The data source reliability score is represented by λ1, λ2, λ3, and λ4, which represent the weights of each item and satisfy λ1+λ2+λ3+λ4=1. When A m,k When the value is greater than a preset association threshold A0, a strong association is established between the state data fragment and the abnormal event; when A... m,k When the value is below A0 but above the candidate threshold A1, the state data fragment is listed as candidate associated data; when A... m,k If the value is below A1, it will not be included in the event recall data range; where A0 and A1 are preset thresholds, and A0>A1; And / or, the abnormal event-data association relationship includes at least the event identifier, data level, measurement point code, device code, data time range, data storage path, association degree, data quality identifier, and association status.

6. The method for recalling and associating abnormal events of hydropower units according to claim 1, characterized in that, The hierarchical recall path is represented as follows: Path(E m )=L3→L2→L1→L0; Among them, Path(E m L1 represents the recall path of the mm-th abnormal event, L2 represents the summary state data layer, L1 represents the monitoring state data layer, L0 represents the transient state data layer, and L0 represents the raw sampled data layer. The step-by-step recall process includes: first, using summary state data to determine the macro-level time period of the anomaly; then, using monitoring state data to determine the start and end times of the anomaly; then, using transient state data to analyze the dynamic change process; and finally, using raw sampled data to analyze the details of the anomaly. For events that do not require calling all levels of data, the recall path is automatically truncated based on the event type and correlation. During the step-by-step recall process, if the required original sampling data is not stored on the platform for a long time, a data recall request is sent to the special monitoring unit according to the abnormal event identifier and the recall window. After the recall is successful, the original sampling data is bound with the abnormal event identifier and the abnormal event-data association relationship is updated.

7. The method for recalling and associating abnormal events of hydropower units according to claim 1, characterized in that, The data retrieval request includes at least the event identifier, plant code, unit code, equipment code, measurement point code, request start time, request end time, data type, sampling frequency, and return format; after the data source returns the data, the returned data is checked for integrity, time continuity, unit, and data quality, and then bound to the abnormal event identifier. If the data retrieval is successful, the retrieved data will be written to the event temporary data warehouse and the abnormal event-data association relationship will be updated; If data retrieval fails, record the reason for the failure and mark the missing original sampled data in the event evidence package; And / or, the abnormal event data evidence package includes at least the following: basic event information, event triggering information, event-related operating conditions, event-related measurement points, pre-event data, in-event data, post-event data, summary status data, monitoring status data, transient status data, original sampling data index, data quality information, recall path, data missing status, correlation calculation results, and storage path information.

8. The method for recalling and associating abnormal events of hydropower units according to claim 1, characterized in that, The method also includes a data transfer step: after an abnormal event is confirmed, reviewed, or a report is generated, the corresponding data evidence package is transferred from the temporary data warehouse to the permanent data warehouse; during the transfer process, the event identifier, data index, data level, original storage path, transfer time, transfer personnel or transfer task, data verification results, and version information are retained; for the original sampled data, the file path, summary information, and index information are saved. And / or, the method further includes a recall rule update step: receiving confirmation information from operators or professionals regarding the results of abnormal event analysis; updating the correspondence between event types and recall windows, the correspondence between events and associated measurement points, event correlation weights, recall path templates, and data retrieval strategies based on the confirmation information; for event types that occur repeatedly but for which recall data is insufficient, increasing the recall window or raising the priority of original sampled data retrieval; and for measurement points that are mistakenly associated, reducing their measurement point correlation weights.

9. A hydropower unit abnormal event recall and correlation system, used to implement the hydropower unit abnormal event recall and correlation method according to any one of claims 1-6, characterized in that, include: The data acquisition and segmentation module is used to acquire the status data generated during the operation of the hydropower unit, and to divide the status data into multiple levels including raw sampling data, transient status data, monitoring status data and summary status data according to the data formation mechanism and time scale, and to establish storage strategies and unified indexes for each level of status data. The first association establishment module is used to identify operating conditions and generate operating condition events based on the pre-acquired unit operating parameters, and establish operating condition-data association relationships between the operating condition events and the status data of each level within the corresponding time period based on the time range of the operating condition events. The first generation module is used to generate an abnormal event object and determine the recall window when an abnormal event is detected. The second association establishment module is used to retrieve relevant status data based on the abnormal event object, the recall window and the unified index, and establish an abnormal event-data association relationship between the abnormal event and the status data. The recall path generation module is used to generate a hierarchical recall path from summary status data to monitoring status data, from monitoring status data to transient status data, and from transient status data to original sampled data according to the type of abnormal event. The module extracts status data at each level according to the hierarchical recall path. If the required original sampled data is not stored, a data retrieval request is issued and the successfully retrieved original sampled data is bound to the abnormal event identifier. The second generation module is used to generate an abnormal event data evidence package for alarm review, fault diagnosis, trend warning or emergency response based on the abnormal event object, the working condition-data association relationship, the abnormal event-data association relationship and the hierarchical recall path.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When executed by a processor, the computer program implements the steps of the method according to any one of claims 1-8.

Images