A method for identifying reliability events of hydropower equipment based on production monitoring data

Abstract

This invention discloses a method for identifying reliability events in hydropower equipment based on production monitoring data. First, a dataset for reliability event identification is constructed. Then, the data in the dataset is preprocessed and labeled to obtain labeled data. Next, a start-up and shutdown record of the hydropower generator unit is generated based on the labeled data, and the shutdown category is determined based on the start-up and shutdown record. Finally, a list of reliability events for the hydropower generator unit is generated based on the event switching mode, using the start-up and shutdown record and the maintenance record. This method applies power system operation data and power equipment monitoring data for analysis and evaluation. Under the premise of ensuring information security, it achieves automated collection, intelligent analysis, traceable management, and transparent supervision of reliability data, enabling real-time monitoring of power equipment reliability and ensuring the accuracy, timeliness, and completeness of reliability data.

Description

Technical Field

[0001] This invention relates to the field of power generation system reliability technology, and specifically to a method for identifying reliability events of hydropower equipment based on production monitoring data. Background Technology

[0002] Power generation equipment reliability refers to the equipment's ability to perform its intended functions under specified conditions and within a specified time interval. Existing power generation reliability management information systems require each power station to input the previous month's reliability data into the system at the beginning of each month. This input primarily references data from manually recorded start-up and shutdown logs and maintenance application forms within the power production management system. While this manually-reported reliability management system can effectively avoid misjudgments of key events, it suffers from low reporting timeliness, as data cannot be automatically collected, resulting in delays. Judgments of special events rely heavily on the management experience of on-site personnel. The accuracy of past event judgments is highly correlated with personnel experience, and the judgment basis is derived from human interpretation, lacking established reference rules and monitoring alarms, thus hindering transparent supervision and control. Furthermore, the logic of manual judgment varies from person to person and lacks systematic standardization, resulting in untraceable results and preventing transparent supervision. This also leads to the potential for overlooking abnormal operating conditions of hydroelectric generator units. Therefore, a method for identifying reliability events in hydroelectric power generation equipment based on production monitoring data is needed to address these issues. Summary of the Invention

[0003] The technical problem to be solved by this invention is to provide a method for identifying reliability events of hydropower equipment based on production monitoring data, so as to solve the problem that the reliability management of hydropower equipment in the prior art is highly dependent on management information systems and human experience, resulting in low timeliness and opaque supervision. This method enables power generation companies to grasp the reliability status of hydropower equipment in a timely and comprehensive manner through digital monitoring.

[0004] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows:

[0005] A method for identifying reliability events in hydropower equipment based on production monitoring data includes:

[0006] S1. Construct a dataset for reliability event identification. The dataset includes equipment monitoring data, hydro-generator maintenance records, and reliability events reported in the reliability system.

[0007] S2, performs data preprocessing and data labeling on the dataset to obtain labeled data;

[0008] S3, Generate the start-up and shutdown record of the hydro-generator unit based on the marked data;

[0009] S4, Determine the shutdown category based on the start-up and shutdown records of the hydro-generator unit;

[0010] S5. Based on the start-up and shutdown records and maintenance records of the hydro-generator unit in S1~S4, generate a list of reliability events for the hydro-generator unit based on the event switching mode.

[0011] S6 compares the generated list of reliability events for the hydro-generator unit with the records entered in the reliability system for supplementary verification.

[0012] Preferably, in step S1, the equipment monitoring data includes power data of the hydro-generator set and equipment status data;

[0013] Power data includes active power and reactive power; each hydro-generator unit is equipped with one power measurement point, and the power data is at a frequency of seconds during the operation of the hydro-generator unit; during shutdown, each data point is approximately 20 minutes long.

[0014] The equipment status refers to the status of the hydro-generator unit, including eight equipment statuses: shutdown, idling, no-load, grid-connected, transition, water system operation, maintenance, and unknown status. Each status corresponds to one equipment status measurement point, and each hydro-generator unit has eight equipment status measurement points. When the unit is in a certain status, the corresponding measurement point value is 1; otherwise, it is 0. The data frequency is not synchronized with the power measurement points, and it is approximately 10-20 minutes per data point.

[0015] Preferably, in step S1, the maintenance record of the hydro-generator unit includes a maintenance application form, which includes the equipment name, maintenance level, work content, dispatch order time, and completion report time. The maintenance application form belongs to the category of power production management data, and the submitted maintenance plan takes effect after dispatch approval.

[0016] Preferably, the reliability events reported in the reliability system include:

[0017] Reliability event types include operational events and downtime events;

[0018] Operational events consist of full-output operation and reduced-output operation; shutdown events consist of standby, planned shutdown, unplanned shutdown, and outage events; reliability events are reported based on start-up and shutdown records and maintenance application forms in the power production management system.

[0019] Furthermore, reliability events are reported monthly by designated personnel; operational events consist of full-output operation and reduced-output operation. Full-output operation events are generally not reported, but are manually reported only if no other events occur in the unit during the month. The mandatory information for event reporting includes the unit number, the start and end time of the event, and the event status. Reduced-output operation events cannot currently be technically determined, and this status event is not yet reported in the reliability system.

[0020] Furthermore, shutdown events consist of standby, planned shutdown, unplanned shutdown, and outage events. The reference basis is the start-up and shutdown records and maintenance application forms in the power production management system. Unplanned shutdown events do not currently have system data support and need to be manually judged and then filled in the reliability system.

[0021] Preferably, the data in the dataset undergoes data preprocessing and data labeling to obtain labeled data, including:

[0022] The active power data in the dataset is processed from second-level data to minute-level data, and the minute node where the power-on or power-off operation is located is identified by marking.

[0023] Set minimum operating threshold for hydro-generator units Based on minute-level data, the active power of the hydro-generator unit is considered to be greater than this threshold. The data is recorded as the operating status and marked as S. The active power of the hydro-generator unit in the operating status is less than this threshold. The data record is in a shutdown state and is marked as R.

[0024] Furthermore, the minimum operating threshold can be selected based on unit capacity and experience, or grid search can be used to compare the output results corresponding to the threshold with the reported results.

[0025] Preferably, in step S3, generating the start-up and shutdown record of the hydro-generator unit based on the marked data includes:

[0026] Based on the data markers in step S2, the generation rules are as follows:

[0027] The first record time node of N consecutive data labeled R is the shutdown start time of the hydro-generator unit;

[0028] The first record time node of N consecutive data marked as S is the start time of the hydro-generator unit;

[0029] Correct the marked data.

[0030] Preferably, in step S4, determining the shutdown category based on the start-up and shutdown records of the hydro-generator unit includes:

[0031] Based on the shutdown termination time identified in step S3, the system determines whether the shutdown process is a normal or abnormal shutdown based on the characteristics of the raw data of the active power of the hydro-generator unit. The raw data is the data collected in step S1 with a frequency on the order of seconds. The specific determination method is as follows:

[0032] S301, Set the baseline variation range and interval duration threshold ;

[0033] S302, using the shutdown termination time as the base time point, and forming a time period T by moving t minutes forward from the base time point, and cyclically calculating the time nodes within the time period T. Active power data and the previous time node Active power data range of change and interval duration Regarding the magnitude of change and interval duration Execution decision:

[0034] If the magnitude of change Exceeding the benchmark change range And the interval length Not less than the interval duration threshold If the above conditions are met, the shutdown is determined to be a data loss-related abnormal shutdown, calculated using the following formula:

[0035] ;

[0036] If the magnitude of change Exceeding the benchmark change range And the interval length Less than the interval duration threshold If so, this shutdown is determined to be a general abnormal shutdown, and the calculation formula is as follows;

[0037] ;

[0038] If the magnitude of change Not exceeding the benchmark change range If the shutdown is normal, then the shutdown is determined to be a normal shutdown. The calculation formula is as follows:

[0039] ;

[0040] S303, Set the loop stop condition:

[0041] The loop stops when the following condition is met for the first time, and the time node of the data is reassigned to the shutdown start time of the hydro-generator unit.

[0042] ;

[0043] in For the current time node active power, The minimum operating threshold for the hydro-generator unit set in step S2; The set baseline variation range;

[0044] The loop stops after all data in time period T has been cycled through;

[0045] If no data is found during the T-period, the system will be automatically shut down due to data loss.

[0046] Furthermore, when marking data and determining the shutdown category, the data used is the raw data at the second level. However, it does not use all the second-level data within the T-period. Instead, it uses a portion of the second-level data within a fixed interval before and after each data node of the minute-level data. This ensures high accuracy by using second-level data while avoiding the problem of slow processing speed caused by processing a large amount of second-level data.

[0047] Preferably, the baseline change range is the difference between the active power data of two adjacent time points during a normal shutdown; the data for the normal shutdown process is obtained by collecting historical data of planned outages from the shutdown events in the dataset.

[0048] Preferably, in step S5, the method for generating a list of reliability events for the hydro-generator unit based on the start-up and shutdown records and maintenance records of the hydro-generator unit in S1~S4 includes:

[0049] S501, based on historical data and unit operating conditions, summarizes three common event switching modes, including:

[0050] First mode: Running, Standby, Running mode;

[0051] Second mode: Operation, Standby, Planned shutdown, Standby, Operation mode;

[0052] Third mode: Operation, Unplanned shutdown, Planned shutdown, Standby, Operation mode;

[0053] S502, generate a list of reliability events for the hydro-generator unit according to the start-up and shutdown records and maintenance records of the hydro-generator unit in S1-S4, following these steps:

[0054] a. All time periods from "power-on to power-off" are classified as running events;

[0055] b. For the "downtime to startup" period and the downtime type is normal downtime, if there are no maintenance records during this period, it is judged as a standby event; if there are maintenance records during this period, the reliability events during this period are, in order, standby, planned shutdown, and standby.

[0056] c. For the "downtime to startup" period and the downtime type is general abnormal downtime, search for the maintenance records within this period. The reliability events within this period are, in order, unplanned downtime, planned downtime, and standby. If no maintenance records are found, it indicates that maintenance records are missing, and the maintenance records should be added to the dataset.

[0057] d. For the "downtime to startup" period and the downtime type is data loss abnormal downtime, if there are maintenance records during this period, the reliability events during this period are, in order, unplanned downtime due to data loss, planned downtime, and standby; if there are no maintenance records during this period, the reliability event during this period is unplanned downtime due to data loss.

[0058] The "Power-on to Power-off" time period refers to the time between the power-on start time and the power-off start time; the "Power-off to Power-on" time period refers to the time between the power-off start time and the power-on start time.

[0059] Furthermore, the first mode corresponds to the vast majority of normal standby scenarios, where the unit enters standby status after a normal shutdown and resumes operation after the standby period ends; the second mode corresponds to scenarios with planned maintenance, where the unit first enters standby status after a normal shutdown, which can be determined by analyzing the shutdown time and the start time of the maintenance plan, and then enters maintenance status according to the plan, and returns to standby after the maintenance is completed, and finally resumes operation; the third mode corresponds to scenarios with abnormal shutdowns, where if there is an unplanned shutdown, planned maintenance is usually arranged after the event to find out the cause of the unplanned shutdown, and then the unit returns to standby and resumes operation.

[0060] Preferably, the marked data is corrected, including real-time data verification based on grid connection status for data marked as R. The verification method is as follows:

[0061] Starting from the time node corresponding to the data marked R, find the first grid connection status record in the past or future M time period. If the grid connection status is grid connection, the numerical expression is 1, then the mark R is corrected to S; otherwise, it remains unchanged.

[0062] When performing the above verification method, the data marked as R must meet the following conditions:

[0063] ;

[0064] in, The current time point active power, The minimum operating threshold for the unit determined in step S2; for The grid connection status of the generating unit at any given moment; The minimum time to satisfy the following conditions:

[0065] ;

[0066] If no data is found within time M, the original identifier is maintained.

[0067] Furthermore, the identification method of the present invention is regarded as supervised learning. The reliability events historically reported in the centralized reliability system are used as the result reference to set and adjust the relevant thresholds in the identification method of the present invention. Conversely, the reliability event list generated by the identification method of the present invention is also used to fill in the gaps in the reporting results in the reliability system.

[0068] The beneficial effects of the hydropower equipment reliability event identification method based on production monitoring data provided by this invention are as follows:

[0069] 1. This invention utilizes power system operation data and power equipment monitoring data for analysis and evaluation, promotes the construction of a power reliability management system based on real-time data, breaks the traditional post-event management model for reliability events, and enables timely and comprehensive monitoring and management of production conditions. It achieves automated data collection, intelligent analysis, traceable management, and transparent supervision of reliability data, laying the foundation for reliability data to assist in operation and maintenance, equipment selection, construction and renovation, and other work.

[0070] 2. The identification method proposed in this invention efficiently utilizes real-time data from device detection. By processing data at the minute level and using raw data at the second level, it effectively improves the data processing speed without sacrificing accuracy.

[0071] 3. The identification method proposed in this invention proposes an application method for time series data of different frequencies. By integrating and using multiple measurement points, it effectively solves the problem caused by the instability of a single data source.

[0072] 4. The identification method proposed in this invention can support and meet the management needs of different levels. For example, several instances of abnormal shutdown characteristics found during the historical data identification process are key management concerns for power generation companies. In the past, such cases were not reported during manual data reporting and were easily overlooked. On the other hand, the identification results provide sufficient basis for manual verification work, avoiding the characteristics of untimely reporting and easy omissions in traditional methods. The event results verified by manual verification can better meet the China Electricity Council's requirements for reliability management. Attached Figure Description

[0073] The present invention will be further described below with reference to the accompanying drawings and embodiments:

[0074] Figure 1 This is a schematic diagram of the process framework of the present invention;

[0075] Figure 2 This is a schematic diagram of the three switching modes of reliability events in this invention. Detailed Implementation

[0076] Example 1:

[0077] like Figure 1 As shown, a method for identifying reliability events of hydropower equipment based on production monitoring data includes:

[0078] S1. Construct a dataset for reliability event identification. The dataset includes equipment monitoring data, hydro-generator maintenance records, and reliability events reported in the reliability system.

[0079] S2, performs data preprocessing and data labeling on the dataset to obtain labeled data;

[0080] S3, Generate the start-up and shutdown record of the hydro-generator unit based on the marked data;

[0081] S4, Determine the shutdown category based on the start-up and shutdown records of the hydro-generator unit;

[0082] S5. Based on the start-up and shutdown records and maintenance records of the hydro-generator unit in S1~S4, generate a list of reliability events for the hydro-generator unit based on the event switching mode.

[0083] S6 compares the generated list of reliability events for hydro-generator units with the records entered in the reliability system to improve the rules.

[0084] Preferably, in step S1, the equipment monitoring data includes power data of the hydro-generator set and equipment status data;

[0085] Power data includes active power and reactive power; each hydro-generator unit is equipped with one power measurement point, and the power data is at a frequency of seconds during the operation of the hydro-generator unit; during shutdown, each data point is approximately 20 minutes long.

[0086] The equipment status refers to the status of the hydro-generator unit, including eight equipment statuses: shutdown, idling, no-load, grid-connected, transition, water system operation, maintenance, and unknown status. Each status corresponds to one equipment status measurement point, and each hydro-generator unit has eight equipment status measurement points. When the unit is in a certain status, the corresponding measurement point value is 1; otherwise, it is 0. The data frequency is not synchronized with the power measurement points, and it is approximately 10-20 minutes per data point.

[0087] Preferably, in step S1, the maintenance record of the hydro-generator unit includes a maintenance application form, which includes the equipment name, maintenance level, work content, dispatch order time, and completion report time. The maintenance application form belongs to the category of power production management data, and the submitted maintenance plan takes effect after dispatch approval.

[0088] Preferably, the reliability events reported in the reliability system include:

[0089] Reliability event types include operational events and downtime events;

[0090] Operational events consist of full-output operation and reduced-output operation; shutdown events consist of standby, planned shutdown, unplanned shutdown, and outage events; reliability events are reported based on start-up and shutdown records and maintenance application forms in the power production management system.

[0091] Furthermore, reliability events are reported monthly by designated personnel; operational events consist of full-output operation and reduced-output operation. Full-output operation events are generally not reported, but are manually reported only if no other events occur in the unit during the month. The mandatory information for event reporting includes the unit number, the start and end time of the event, and the event status. Reduced-output operation events cannot currently be technically determined, and this status event is not yet reported in the reliability system.

[0092] Furthermore, shutdown events consist of standby, planned shutdown, unplanned shutdown, and outage events. The reference basis is the start-up and shutdown records and maintenance application forms in the power production management system. Unplanned shutdown events do not currently have system data support and need to be manually judged and then filled in the reliability system.

[0093] Preferably, the data in the dataset undergoes data preprocessing and data labeling to obtain labeled data, including:

[0094] The data in the dataset is processed into minute-level data, and the minute node where the power-on or power-off operation occurred is identified by marking.

[0095] Set minimum operating threshold for hydro-generator units Based on minute-level data, the active power of the hydro-generator unit is considered to be greater than this threshold. The data is recorded as the operating status and marked as S. The active power of the hydro-generator unit in the operating status is less than this threshold. The data record is in a shutdown state and is marked as R.

[0096] Furthermore, the minimum operating threshold can be selected based on unit capacity and experience, or grid search can be used to compare the output results corresponding to the threshold with the reported results.

[0097] Preferably, in step S3, generating the start-up and shutdown record of the hydro-generator unit based on the marked data includes:

[0098] Based on the data markers in step S2, the generation rules are as follows:

[0099] The first record time node of N consecutive data labeled R is the shutdown start time of the hydro-generator unit;

[0100] The first record time node of N consecutive data marked as S is the start time of the hydro-generator unit;

[0101] Correct the marked data.

[0102] Preferably, in step S4, determining the shutdown category based on the start-up and shutdown records of the hydro-generator unit includes:

[0103] Based on the shutdown termination time identified in step S3, the system determines whether the shutdown process is a normal or abnormal shutdown based on the characteristics of the raw data of the active power of the hydro-generator unit. The raw data is the data collected in step S1 with a frequency on the order of seconds. The specific determination method is as follows:

[0104] S301, Set the baseline variation range and interval duration threshold ;

[0105] S302, using the shutdown termination time as the base time point, and forming a time period T by moving t minutes forward from the base time point, and cyclically calculating the time nodes within the time period T. Active power data and the previous time node Active power data range of change and interval duration Regarding the magnitude of change and interval duration Execution decision:

[0106] If the magnitude of change Exceeding the benchmark change range And the interval length Not less than the interval duration threshold If the above conditions are met, the shutdown is determined to be a data loss-related abnormal shutdown, calculated using the following formula:

[0107] ;

[0108] If the magnitude of change Exceeding the benchmark change range And the interval length Less than the interval duration threshold If so, this shutdown is determined to be a general abnormal shutdown, and the calculation formula is as follows;

[0109] ;

[0110] If the magnitude of change Not exceeding the benchmark change range If the shutdown is normal, then the shutdown is determined to be a normal shutdown. The calculation formula is as follows:

[0111] ;

[0112] S303, Set the loop stop condition:

[0113] The loop stops when the following condition is met for the first time, and the time node of the data is reassigned to the shutdown start time of the hydro-generator unit.

[0114] ;

[0115] in For the current time node active power, The minimum operating threshold for the hydro-generator unit set in step S2; The set baseline variation range;

[0116] The loop stops after all data in time period T has been cycled through;

[0117] If no data is found during the T-period, the system will be automatically shut down due to data loss.

[0118] Furthermore, when marking data and determining the shutdown category, the data used is the raw data at the second level. However, it does not use all the second-level data within the T-period. Instead, it uses a portion of the second-level data within a fixed interval before and after each data node of the minute-level data. This ensures high accuracy by using second-level data while avoiding the problem of slow processing speed caused by processing a large amount of second-level data.

[0119] Preferably, the baseline change range is the difference between the active power data of two adjacent time points during a normal shutdown; the data for the normal shutdown process is obtained by collecting historical data of planned outages from the shutdown events in the dataset.

[0120] Preferably, in step S5, the method for generating a list of reliability events for the hydro-generator unit based on the start-up and shutdown records and maintenance records of the hydro-generator unit in S1~S4 includes:

[0121] S501, based on historical data and unit operating conditions, summarizes three common event switching modes, including:

[0122] First mode: Running, Standby, Running mode;

[0123] Second mode: Operation, Standby, Planned shutdown, Standby, Operation mode;

[0124] Third mode: Operation, Unplanned shutdown, Planned shutdown, Standby, Operation mode;

[0125] S502, generate a list of reliability events for the hydro-generator unit according to the start-up and shutdown records and maintenance records of the hydro-generator unit in S1-S4, following these steps:

[0126] a. All time periods from "power-on to power-off" are classified as running events;

[0127] b. For the "downtime to startup" period and the downtime type is normal downtime, if there are no maintenance records during this period, it is judged as a standby event; if there are maintenance records during this period, the reliability events during this period are, in order, standby, planned shutdown, and standby.

[0128] c. For the "downtime to startup" period and the downtime type is general abnormal downtime, search for the maintenance records within this period. The reliability events within this period are, in order, unplanned downtime, planned downtime, and standby. If no maintenance records are found, it indicates that maintenance records are missing, and the maintenance records should be added to the dataset.

[0129] d. For the "downtime to startup" period and the downtime type is data loss abnormal downtime, if there are maintenance records during this period, the reliability events during this period are, in order, unplanned downtime due to data loss, planned downtime, and standby; if there are no maintenance records during this period, the reliability event during this period is unplanned downtime due to data loss.

[0130] The "Power-on to Power-off" time period refers to the time between the power-on start time and the power-off start time; the "Power-off to Power-on" time period refers to the time between the power-off start time and the power-on start time.

[0131] Furthermore, the first mode corresponds to the vast majority of normal standby scenarios, where the unit enters standby status after a normal shutdown and resumes operation after the standby period ends; the second mode corresponds to scenarios with planned maintenance, where the unit first enters standby status after a normal shutdown, which can be determined by analyzing the shutdown time and the start time of the maintenance plan, and then enters maintenance status according to the plan, and returns to standby after the maintenance is completed, and finally resumes operation; the third mode corresponds to scenarios with abnormal shutdowns, where if there is an unplanned shutdown, planned maintenance is usually arranged after the event to find out the cause of the unplanned shutdown, and then the unit returns to standby and resumes operation.

[0132] Preferably, the marked data is corrected, including real-time data verification based on grid connection status for data marked as R. The verification method is as follows:

[0133] Starting from the time node corresponding to the data marked R, find the first grid connection status record in the past or future M time period. If the grid connection status is grid connection, the numerical expression is 1, then the mark R is corrected to S; otherwise, it remains unchanged.

[0134] When performing the above verification method, the data marked as R must meet the following conditions:

[0135] ;

[0136] in, The current time point active power, The minimum operating threshold for the unit determined in step S2; for The grid connection status of the generating unit at any given moment; The minimum time to satisfy the following conditions:

[0137] ;

[0138] If no data is found within time M, the original identifier is maintained.

[0139] Furthermore, the identification method of the present invention is regarded as supervised learning. The reliability events historically reported in the centralized reliability system are used as the result reference to set and adjust the relevant thresholds in the identification method of the present invention. Conversely, the reliability event list generated by the identification method of the present invention is also used to fill in the gaps in the reporting results in the reliability system.

[0140] Example 2:

[0141] In the data preprocessing and data labeling steps, this embodiment takes Unit 16 of a certain power plant as an example, with an installed capacity of 700MW. The initial minimum threshold for active power in the operating state is set at 20MW. The sample data for processing and labeling are shown in Table 1 below:

[0142]

[0143] Table 1: Sample data for preprocessing and labeling;

[0144] In the step of generating unit start-up and shutdown records, the example data provided in this embodiment is shown in Table 2:

[0145]

[0146] Table 2: Start-stop record data;

[0147] In the shutdown category determination step, the shutdown time finally determined in this embodiment is 12:02. In this step, if we search backward for 30 minutes, the active power of 6,122,900 kW at 11:53:39 will drop to 0.012 kW at the next moment at 11:53:40. The change value exceeds the set threshold, and the time interval between the two points is 1 second, which meets the judgment rules for general abnormal shutdown.

[0148] This embodiment uses data from Unit 14 on January 6, 2024 as an example. Table 3 shows an example of abnormal shutdown due to missing data.

[0149]

[0150] Table 3: Example data of abnormal shutdowns due to missing data;

[0151] In Table 3, the determined downtime is 21:27. In this step, searching back 30 minutes, it can be found that the active power of 6,053,400 kW at 21:08:29 drops to -0.31 kW at the next moment 21:27:23. The change value exceeds the set threshold, but the time interval between the two points is about 20 minutes, which may be due to an abnormality or interruption in the data transmission process. Therefore, it is marked as an abnormal downtime due to missing data and handed over to manual review.

[0152] Example 3:

[0153] Step S5 of this embodiment provides example data:

[0154] Based on the start-up and shutdown records and maintenance records of the hydro-generator units in S1~S4, the sample data generated in the step of generating a reliability event list for the hydro-generator units based on the event switching mode is shown in Table 4 below:

[0155]

[0156] Table 4: Example data for the reliability event list;

[0157] Table 4 shows the generated list of reliability events for the hydro-generator set.

[0158] Example 4:

[0159] This embodiment provides data for correcting marked data, including real-time data verification marked as R based on grid connection status, as shown in Table 5 below. Taking the data from Unit 21 around 16:47-16:55 on March 28, 2023 as an example, the sample data is as follows:

[0160]

[0161] Table 5: Active power and identification, grid connection status data of generating units within the same time range;

[0162] Matching was performed on the data marked as R from 16:55 to 16:57. The most recent state timestamp corresponding to 16:55 was 16:54, at which time the grid connection status was 0, so the mark was kept as R. The most recent state timestamp corresponding to 16:56 was 16:54, at which time the grid connection status was 0, so the mark was kept as R. The most recent state timestamp corresponding to 16:57 was 16:57, at which time the grid connection status was 1, so the mark was corrected to S. The corrected data is shown in Table 6 below:

[0163]

[0164] Table 6: Data after review and correction;

[0165] As shown in Table 7 below, the data of all 32 generating units of a certain power plant from 00:00 on January 1, 2023 to 00:00 on March 31, 2024 were analyzed and tested according to the above method, involving more than 979 million data points.

[0166] As shown in Table 7 below, a total of 2,687 standby events, 9 unplanned outages (UOs), and 48 unplanned outages with missing data (UOs) were identified. The reliability management system recorded 2,750 standby events and 1 unplanned outage (UO). The system accurately identified the non-outage events that were the focus of reliability management and were reported. In addition, it monitored outage events with abnormal characteristics in the unit's operating conditions. The overall event identification results and reporting results had a consistency rate of 97.7%.

[0167] For shutdown events with abnormal characteristics in the unit's operating conditions, the causes are analyzed from two aspects: data characteristics and on-site investigation. The main causes include data missing, data anomalies, unit commissioning, and unit experiments.

[0168]

[0169] Table 7: Reliability event identification results;

[0170] Based on user reporting habits, the reliability system typically only reports full-output operation events when no other events occur in the current month. Events spanning multiple months are further segmented. The results show that the reliability event results identified based on the above steps have a high degree of consistency with the reported data.

Claims

1. A method for identifying reliability events in a hydroelectric power plant based on production monitoring data, characterized by, Comprise: S1, constructing a data set for reliability event identification, the data set comprising equipment monitoring data, hydroelectric generating set maintenance records and historical reliability events reported in a reliability system; S2, data preprocessing and data labeling are performed on the equipment monitoring data in the data set to obtain labeled data; S3, generating a hydroelectric generating set start-stop record according to the labeled data; S4, judging the shutdown category based on the hydroelectric generating set start-stop record; S5, generating a hydroelectric generating set reliability event list based on event switching mode according to the hydroelectric generating set start-stop record, the shutdown category judgment result and the hydroelectric generating set maintenance record; Specifically comprising: S501, based on historical data and unit operating conditions, three common event switching modes are summarized, including: The first mode: running, standby, running mode; The second mode: running, standby, planned shutdown, standby, running mode; The third mode: running, unplanned shutdown, planned shutdown, standby, running mode; S502, generating a hydroelectric generating set reliability event list according to the hydroelectric generating set start-stop record, the shutdown category judgment result and the hydroelectric generating set maintenance record by the following steps: a. For all "start to stop" time periods, it is judged as a running event; b. For "stop to start" time period and normal shutdown type, if there is no maintenance record in the time period, it is judged as a standby event; If there is a maintenance record in the time period, the reliability events in the time period are standby, planned shutdown and standby in turn; c. For "stop to start" time period and general abnormal shutdown type, find the maintenance record in the time period, the reliability events in the time period are unplanned shutdown, planned shutdown and standby in turn; If no maintenance record is found, the maintenance record is supplemented into the data set; d. For "stop to start" time period and data missing abnormal shutdown type, if there is a maintenance record in the time period, the reliability events in the time period are missing unplanned shutdown, planned shutdown and standby in turn; If there is no maintenance record in the time period, the reliability event in the time period is missing unplanned shutdown; Wherein, "start to stop" time period represents the time period between start time and stop time; "stop to start" time period represents the time period between stop time and start time; S6, comparing the generated hydroelectric generating set reliability event list with the historical reporting records in the reliability system, performing result checking and rule optimization.

2. The method for identifying reliability events of a hydroelectric power plant based on production monitoring data according to claim 1, characterized in that, In step S1, the equipment monitoring data comprises hydroelectric generating set power data and equipment state data; The power data includes active power and reactive power; one power measuring point is set for each hydroelectric generating set, and the power data is second-level data during the operation of the hydroelectric generating set; The device state is a state of a hydroelectric generating set, including eight device states of a shutdown state, an idling state, an unloaded state, a grid-connected state, a transition state, a water system operation state, a maintenance state, and an unknown state; each type of state is provided with a device state measuring point, and each hydroelectric generating set is provided with eight device state measuring points.

3. The method of claim 2, wherein the method further comprises: In step S1, the hydroelectric generating set maintenance record includes a maintenance application form, and the maintenance application form includes a device name, a maintenance level, a work content, a ladder adjustment order time, and a completion report time.

4. The method of claim 3, wherein the method further comprises: In step S1, the historical reliability events filled in the reliability system include: The reliability event type includes an operation event and a shutdown event; The operation event is composed of full-power operation and reduced-power operation; the shutdown event is composed of standby, planned shutdown, unplanned shutdown, and shutdown event; the reliability event is filled in according to the opening and shutdown record in the power production management system and the maintenance application form.

5. A method of identifying reliability events in a hydroelectric power plant based on production monitoring data according to claim 4, characterized in that, In step S2, the device monitoring data in the data set are preprocessed and marked to obtain the marked data, including: The data in the data set are processed into minute-level data, and the minute node of the start or stop operation is confirmed through marking means; Set minimum operating threshold for hydro-generator units Based on minute-level data, the active power of the hydro-generator unit is considered to be greater than this threshold. The data is recorded as the operating status and marked as S. The active power of the hydro-generator unit in the operating status is less than this threshold. The data record is in a shutdown state and is marked as R.

6. The method of claim 5, wherein the method further comprises: In step S3, the start-stop record of the hydroelectric generating set is generated according to the marked data, including: The start time and the stop time of the opening in the period are generated according to the data marking in step S2, and the generation rule is: The last record time node of N continuous data marked as R is the shutdown termination time of the hydroelectric generating set; The first record time node of N continuous data marked as S is the start time of the opening of the hydroelectric generating set; The marked data are corrected.

7. A method of identifying reliability events in a hydroelectric power plant based on production monitoring data according to claim 6, characterized in that, In step S4, the shutdown category determination based on the start-stop record of the hydroelectric generating set includes: Based on the shutdown termination time identified in step S3, the original second-level data characteristics of the active power of the hydroelectric generating set are used to determine whether the shutdown process belongs to normal shutdown or abnormal shutdown, wherein the original data are the data with a second-level frequency collected in step S1; the specific determination method is: S301, set a reference variation range and interval duration threshold ; S302, with the shutdown termination time as a reference time point, a time period T is formed t minutes before the reference time point, and the time period T is cyclically calculated active power data and the last previous time node active power data variation range and interval duration , the variation range and interval duration are determined: If the change range exceeds the reference change range , and the interval duration is not less than the interval duration threshold , it is determined that the current shutdown is a data missing abnormal shutdown, and the calculation formula is as follows: ； If the change range exceeds the reference change range , and the interval duration is less than the interval duration threshold , the current shutdown is determined as a general abnormal shutdown, and the calculation formula is as follows: ； If the variation range does not exceed the reference variation range , the current shutdown is determined as a normal shutdown, and the calculation formula is as follows: ； S303, setting a loop stop condition: When the following conditions are met for the first time, stop the loop, and reassign the time node of the data to the shutdown start time of the hydroelectric generating set; ； wherein is the active power at the current time node is the active power at the current time node is the minimum threshold value of the operation of the hydro-generator set set in step S2; is the set reference variation range; Stop the loop after all T period data are circulated; If no data is searched in the T period, it is directly determined as data missing abnormal shutdown.

8. The method of claim 7, wherein the method further comprises: The reference change amplitude is the difference between the active power data of the adjacent two time nodes in the normal shutdown process; the data of the normal shutdown process are determined by the planned shutdown historical data in the shutdown event in the data set.

9. The method of claim 6, wherein the method further comprises: The marked data are corrected, including the real-time data review of the data marked as R based on the grid-connected state, and the review method is: Taking the time node corresponding to the data marked as R as the starting point, the first grid-connected state record in the past or future M time is searched, if the grid-connected state is grid-connected, and the digital expression is 1, the marked R is corrected to S, otherwise it remains unchanged; When the above review method is executed, the data marked as R need to meet the following conditions: ； wherein, is the active power at the current time node is the active power at the current time node is the minimum threshold value of the operation of the hydroelectric generating unit determined in step S2; is the minimum threshold value of the operation of the hydroelectric generating unit determined in step S2; is the grid-connected state of the unit at the time instant; is the minimum time instant satisfying the following condition: ； For the case that no data is searched in M time, the original identification is maintained.

Images