Water conservancy project equipment fault diagnosis method and system

Abstract

The invention relates to the technical field of fault diagnosis, in particular to a hydraulic engineering equipment fault diagnosis method and system.The hydraulic engineering equipment fault diagnosis method comprises the following steps of collecting water inlet and outlet flow velocity data of a water pump to judge direction reversal, extracting rotating speed and pressure trend to collect linkage characteristics, analyzing starting and stopping behaviors to calibrate abnormal sections, and integrating multi-parameter generation structural block screening combination characteristics; arranging the structure blocks to form a continuous path, and classifying the matching mode to generate a fault trend diagnosis path set. According to the method, the direction change of the adjacent sampling points in the flow velocity data of the water inlet and outlet channel of the water pump is continuously judged, the section with transition instability in the local flow velocity structure is recognized, fault symptoms of multiple dimensions are merged into recognizable structural blocks, the overall integrity of feature recognition is improved, and the accuracy of feature recognition is improved. The fault diagnosis process is improved from single-parameter abnormal identification to composite matching of a multi-dimensional behavior mode, and the method has the capability of tracking and positioning the whole process of the operation trend of hydraulic engineering equipment.

Description

Technical Field

[0001] The present invention relates to the technical field of fault diagnosis, and in particular to a method and system for diagnosing faults of water conservancy engineering equipment. Background Art

[0002] The field of fault diagnosis technology includes methods for identifying, locating, and judging problems such as anomalies, damage, or performance degradation that occur in various types of machinery, equipment, and systems during operation. The core content of this technical field includes fault signal acquisition, diagnostic model construction, feature extraction, and analysis. The goal is to achieve timely identification and processing of potential equipment fault states and improve operational reliability and safety. This field covers a wide range of technical systems, relying on sensor monitoring data, operational information, and physical characteristic parameters, combined with technical means such as rule analysis, knowledge base reasoning, and statistical modeling to achieve diagnosis and evaluation of equipment status. Its systematic nature is reflected in multi-dimensional data fusion, diagnostic process standardization, and cross-device application capabilities. It is widely used in multiple engineering systems such as manufacturing, electricity, water conservancy, and aviation.

[0003] Among them, the fault diagnosis method for water conservancy equipment refers to the equipment used in water conservancy projects, including water pumps, gate opening and closing machines, transformers, turbines, etc., by collecting operating parameters such as vibration signals, current, voltage, water pressure, water flow velocity and other key physical quantities, and performing segmented statistics and comparative analysis within a given time interval, combining empirical threshold standards and equipment working status categories to determine whether there is abnormal operation or potential failure. The method also includes determining the trend of change based on time series analysis, detecting abnormal spectrum feature points using frequency domain analysis methods, and clarifying the diagnosis type and location through manual verification and record matching results to form an equipment operation health file. The patent subject takes physical signal measurement, quantitative analysis and state classification as the main participating contents, clearly defines the diagnostic objects, signal types and analysis methods, and constitutes a complete water conservancy equipment fault diagnosis process.

[0004] While existing technologies have proposed diagnostic pathways based on physical quantity signal analysis, they often rely on the setting of static parameter thresholds and frequency domain detection of single features, lacking the ability to identify cross-parameter linkages during equipment operation. In terms of identifying start-stop behavior, they only perform routine statistics on start-stop frequency and periodicity, lacking dynamic comparison with real-time fluid characteristics and pressure trends, making it difficult to construct a systematic explanation path for behavioral anomalies. Existing methods often identify anomalies based on single points or short-term fluctuations, failing to identify hidden fault chains within structural continuity, leading to delayed identification of trend degradation. In practical applications, minor anomalies may not be promptly classified, accumulating and evolving into failures. This is particularly true in scenarios where pump equipment operates cyclically, where signs of rhythmic anomalies can be easily missed, impacting the stability and reliability of hydraulic equipment. Summary of the Invention

[0005] The purpose of the present invention is to solve the shortcomings of the prior art and to propose a fault diagnosis method and system for water conservancy engineering equipment.

[0006] In order to achieve the above object, the present invention adopts the following technical solution, a method for diagnosing faults of water conservancy project equipment, comprising the following steps:

[0007] S1: Collect real-time flow velocity data from the pump's inlet and outlet channels, determine the direction of flow velocity change at adjacent sampling points, locate the section with direction reversal among every three consecutive sampling points, mark the time point of reversal, identify local flow velocity structure changes, and generate a sequence of flow velocity jump instability locations;

[0008] S2: Based on the time period of the node in the flow velocity jump instability position sequence, extract the pump main shaft speed curve data, identify the turning position between the extreme values ​​and divide it into the return interval, evaluate the linkage mapping relationship between the speed return and the pressure jump, and generate a linkage feature distribution map;

[0009] S3: Extracting the start and stop action records of the water pump according to the linkage time segments calibrated by the linkage feature distribution map, dividing the continuous operation time in sequence, and performing time series classification on the start and stop behaviors in adjacent operation segments to generate the rhythm fluctuation overlapping block positioning results;

[0010] S4: Call the time period identified in the rhythm fluctuation overlapping block positioning result, integrate the information into a structural block, and perform combined condition screening to see whether flow velocity reversal aggregation, speed return accumulation and start-stop concentration phenomena occur simultaneously in the structural block. If satisfied, execute the linkage mark and generate a diagnostic feature linkage group identifier.

[0011] As a further solution of the present invention, the flow rate jump instability position sequence includes the direction reversal node time point, the reversal intensive interval, and the change trend switching segment; the linkage feature distribution map includes the return structure distribution interval, the pressure trend superposition segment, and the speed and pressure coordinated evolution segment; the rhythm fluctuation overlapping block positioning result includes the start-stop frequency aggregation segment, the behavior rhythm jump block, and the restart action continuous distribution segment; the diagnostic feature linkage group block identifier includes the behavior aggregation structure block number, the multi-parameter collaborative anomaly composition group, and the linkage status feature.

[0012] As a further solution of the present invention, the steps for obtaining the flow velocity transition instability position sequence are specifically as follows:

[0013] S111: Collect real-time flow velocity data from the water pump's inlet and outlet channels, compare the flow velocity values ​​at adjacent sampling points with the corresponding time, determine whether there is a change in the sign of the value, and sequentially detect whether there is a reversal of the flow velocity direction by grouping three consecutive sampling points. Generate a flow velocity direction reversal time series by extracting the timestamp values ​​corresponding to the points where reversal occurs.

[0014] S112: calling the velocity direction reversal time series, performing a time sequence sorting operation, eliminating repeated or evenly spaced abnormal reversal time nodes, screening time segments with continuous fluctuations, and generating a local velocity change time interval set;

[0015] S113: Based on the local flow velocity change time interval set, the flow velocity change rate of the water inlet and outlet channels of the water pump in the corresponding time period, the continuous reversal frequency and the flow velocity difference before and after a single reversal are extracted, and numerical superposition, comparison and normalization judgment operations are performed on each time period to calculate the flow velocity transition degree index in the real-time interval. The index values ​​corresponding to the time periods are compared in time series, and the time periods where the transition degree has a sudden change are extracted to obtain a flow velocity transition instability position sequence.

[0016] As a further solution of the present invention, the steps of obtaining the linkage feature distribution map are specifically as follows:

[0017] S211: Based on the time period of the node in the flow velocity jump instability position sequence, calling the water pump main shaft speed curve data, identifying the extreme value points in the main shaft speed curve, distinguishing and judging the turning positions between adjacent extreme value points, and screening the time period that continuously shows a reversal trend based on the change amplitude of the interval curvature to obtain the speed reversal section;

[0018] S212: Based on the time range within the speed reversal section, all pressure sensor signal data within the time period are retrieved, continuously sorted in chronological order, and a pressure change curve is recorded according to the sampling point sequence. Each continuously rising pressure trend segment is numbered and distinguished, and the corresponding time period is mapped back to the sequence identifier of the reversal section to generate a reversal section pressure rising sequence comparison table;

[0019] S213: Call the pressure rise sequence comparison table of the return section, identify the corresponding speed return number and pressure rise section number in the time window, calculate the speed return and pressure jump linkage value, sort the windows according to the linkage value, and obtain the linkage feature distribution map.

[0020] As a further solution of the present invention, the steps for obtaining the positioning result of the rhythm fluctuation overlapping blocks are specifically as follows:

[0021] S311: Extracting the start and stop action records of the water pump according to the linkage time segment calibrated by the linkage characteristic distribution map, identifying the time span between each start time and stop time, and dividing the operation into multiple operation segments in chronological order to generate an operation time division sequence;

[0022] S312: Calling the run time division sequence, classifying the start and stop behaviors in adjacent run segments, counting the start and end positions of each run segment and the intervals between adjacent segments, calculating the restart behavior difference between adjacent segments based on the degree of time overlap, the run intensity within the segment, and the interval length, analyzing the segments where the restart behaviors are concentrated based on the clustering pattern formed by the difference values ​​in the sequence, and obtaining a restart segment indicator sequence;

[0023] S313: Matching and positioning are performed based on the restart segment indicator sequence in combination with corresponding points on the time axis, identifying the intersection relationship between the segment and the time spectrum, identifying the number and distribution density of signal changes in the intersection segment, and obtaining the rhythm fluctuation overlapping block positioning result.

[0024] As a further solution of the present invention, the step of obtaining the diagnostic feature linkage block identifier is specifically as follows:

[0025] S411: calling the time period identified in the rhythm fluctuation overlapping block positioning result, retrieving the flow velocity change and direction records, performing incremental accumulation based on the flow velocity difference sequence before and after the direction reversal point and the time axis sequence, extracting the corresponding slope mutation segment as the direction change node sequence, and generating the flow velocity direction change interval value;

[0026] S412: Based on the flow velocity direction variation interval value, the speed data is compared with a set reference value, and a time period of a difference amplitude is selected to form a reversal point set. The time interval between consecutive reversal points is calculated, and the interval values ​​are interval-merged. The speed fluctuation amplitude and the number of reversals are counted to generate a speed reversal cumulative ratio.

[0027] S413: Call the flow direction change interval value and the speed reversal cumulative ratio, extract the pressure change data and start-stop records in the overlapping time period of the two, analyze the pressure increase amplitude per unit time, count the number of starts and stops in the time period, perform multiple parallel classification operations on the number of flow reversal points, speed reversal density, pressure change trend and start-stop frequency distribution, identify the common classification results of the four features, and generate a diagnostic feature linkage block identifier.

[0028] As a further solution of the present invention, the method further includes step S5:

[0029] S5: Based on the diagnostic feature linkage block identifier, the differentiated linkage block structure is arranged according to the time dimension, and the continuity and structural component fit on the time axis are calculated. When three or more groups of highly consistent structures appear continuously, the structural block is marked as a traceable failure path, and the matching patterns of reversal, return, pressure trend and start-stop characteristics in the path are classified to generate a set of equipment failure trend diagnostic paths;

[0030] The equipment failure trend diagnosis path set includes a time sequence tracking path, a structural combination pattern, and a state abnormal variation chain.

[0031] As a further solution of the present invention, the steps for obtaining the equipment failure trend diagnostic path set are specifically as follows:

[0032] S511: using the diagnostic feature linkage block identifier, arranging the linkage blocks along the time axis according to the block occurrence time sequence, extracting the structural components corresponding to the blocks, comparing the repetition of the components between consecutive blocks, determining the structural repetition degree within the time continuous interval, calculating the repetition ratio of the structural components within the interval, and forming a repetition degree sequence interval;

[0033] S512: Calling the component repetition sequence interval, identifying paths in which the number of consecutive occurrences of structural components exceeds a set number in the time sequence, marking them as traceable paths, summarizing the reversal direction, return position, pressure change trend and start / stop signal status of the blocks in the path, and generating a path structure feature combination sequence;

[0034] S513: Based on the path structure feature combination sequence, the reversal direction, return position, pressure change trend and start-stop signal status of the corresponding block in the path are extracted as similar sequences, and the change frequency and direction change degree of multiple types of sequences in the same path are compared to determine whether they belong to the same feature mode, and the path types are divided to generate a set of equipment failure trend diagnosis paths.

[0035] The hydraulic engineering equipment fault diagnosis system is used to execute the hydraulic engineering equipment fault diagnosis method described above, and the system includes:

[0036] The data sorting module collects real-time flow velocity data from the pump's inlet and outlet channels, determines the direction of flow velocity change at adjacent sampling points, locates three-point combination sections with reverse changes, and generates a sequence of flow velocity jump instability locations.

[0037] The structure positioning module calls the flow velocity transition instability position sequence, identifies the extreme value of the speed curve and delineates the return section in sections, extracts the pressure change data in chronological order in the return section, matches the return number with the pressure change trend and pairs them to generate a linkage feature distribution map;

[0038] Based on the linkage feature distribution map, the state linkage module selects the time period when the speed change and pressure jump change synchronously, extracts each continuous operation section between the start and stop of the water pump, statistically classifies the frequency of start and stop actions and the adjacent interval time, and generates the positioning result of the rhythm fluctuation overlapping block;

[0039] The behavior rhythm module calls the rhythm fluctuation overlapping block positioning result, performs time axis mapping on the marked time period, extracts the behavior segments with linkage features, compares the distribution interval and duration of the segments within the operation cycle, and generates a diagnostic feature linkage group block identifier;

[0040] The path classification module calls the linkage feature participants according to the diagnostic feature linkage group block identifier, aggregates the distribution patterns in multiple overlapping segments, marks the parameter combinations of consistency features and counts the number of recurrences, and generates a set of equipment failure trend diagnostic paths.

[0041] Compared with the prior art, the advantages and positive effects of the present invention are:

[0042] In the present invention, by continuously judging the direction changes of adjacent sampling points in the flow velocity data of the water inlet and outlet channels of the water pump, the sections with transition instability in the local flow velocity structure are identified, the inversion nodes are extracted and the position sequence is established in the time series dimension, providing a positioning basis for the subsequent parameter linkage, and the speed curve and pressure signal in the corresponding time period are retrieved. By identifying the turning interval and the continuous change trend, the linkage distribution map between pressure and speed is constructed, and the ability to characterize the coupling relationship between composite signals is enhanced. Combined with the time series classification and abnormal frequency identification of the start and stop behavior of the equipment, the control logic layer data and the physical signal layer data are aligned and linked, and the accuracy of abnormal behavior identification is improved. The reversal aggregation, return accumulation, pressure superposition and start-stop concentration features within a specific time period are combined and screened, and the fault signs in multiple dimensions are merged into identifiable structural blocks to improve the overall integrity of feature recognition. The failure trend path is classified according to the continuity and structural fit of the linkage structural blocks on the time axis, and the potential faults of the equipment are summarized from the cause distribution to the evolution trajectory. The flow direction reversal structure is explicitly located, and the cross-parameter feature aggregation and start-stop behavior correlation mapping are fine-grained operations. The fault diagnosis process is upgraded from single-parameter anomaly identification to composite matching of multi-dimensional behavior patterns, and has the ability to track and locate the operation trend of water conservancy project equipment throughout the entire process. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 It is a schematic diagram of the workflow of the present invention;

[0044] Figure 2 This is a flow chart for obtaining the flow rate transition instability position sequence in the present invention;

[0045] Figure 3 This is a flow chart for obtaining the linkage feature distribution map in the present invention;

[0046] Figure 4 This is a flow chart for obtaining the positioning results of the rhythm fluctuation overlapping blocks in the present invention;

[0047] Figure 5 This is a flow chart for obtaining the diagnostic feature linkage block identifier in the present invention;

[0048] Figure 6 This is a flow chart for obtaining a set of equipment failure trend diagnostic paths in the present invention. DETAILED DESCRIPTION

[0049] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0050] In the description of the present invention, it should be understood that the terms "length," "width," "up," "down," "front," "back," "left," "right," "vertical," "horizontal," "top," "bottom," "inside," "outside," and the like, indicating positions or relationships, are based on the positions or relationships shown in the accompanying drawings and are intended only to facilitate the description of the present invention and simplify the description. They do not indicate or imply that the devices or elements referred to must have a specific orientation, be constructed, or operate in a specific orientation. Therefore, they should not be construed as limiting the present invention. Furthermore, in the description of the present invention, "plurality" means two or more, unless otherwise expressly and specifically defined.

[0051] See also Figure 1 The present invention provides a technical solution, a method for diagnosing faults in water conservancy engineering equipment, comprising the following steps:

[0052] S1: Collect real-time flow velocity data from the pump's inlet and outlet channels, determine the direction of flow velocity change at adjacent sampling points, locate the section with direction reversal among every three consecutive sampling points, mark the time point of reversal, arrange the reversal time points in time sequence, identify local flow velocity structure changes, and generate a sequence of flow velocity jump instability locations;

[0053] S2: Based on the time period of the node in the velocity jump instability position sequence, the pump main shaft speed curve data is extracted, the turning points between the extreme values ​​are identified and divided into return intervals, the pressure sensor signals in the return interval are recorded in chronological order, and the number of speed return intervals and the continuous pressure rise intervals in the same time window are aggregated. The linkage mapping relationship between speed return intervals and pressure jump intervals is evaluated to generate a linkage feature distribution map.

[0054] S3: Based on the linkage time segments calibrated by the linkage feature distribution map, extract the pump start and stop action records, divide the continuous operation time between start and stop, perform time series classification on the start and stop behaviors in adjacent operation segments, identify the restart segments, align the corresponding points on the time axis, identify the abnormal interval of the fault segment, and generate the rhythm fluctuation overlapping block positioning result;

[0055] S4: Call the time period identified in the rhythm fluctuation overlapping block positioning result, retrieve the flow velocity direction change node, the number of speed return, the pressure rising trend and the start-stop frequency record, integrate the information into a structure block, and perform a combined condition screening to determine whether flow velocity reversal aggregation, speed return accumulation, pressure upward superposition and start-stop concentration occur simultaneously in the structure block. If the conditions are met, perform the execution linkage mark and generate the diagnostic feature linkage group block identifier;

[0056] S5: Based on the diagnostic feature linkage block identification, the differentiated linkage block structure is arranged according to the time dimension. The continuity and structural component fit on the time axis are calculated. When three or more groups of highly consistent structures appear continuously, the structural block is marked as a traceable failure path. The matching patterns of reversal, return, pressure trend and start-stop characteristics in the path are classified to generate a set of equipment failure trend diagnostic paths.

[0057] The velocity transition instability position sequence includes the direction reversal node time point, the reversal intensive interval, and the change trend switching segment. The linkage feature distribution map includes the return structure distribution interval, the pressure trend superposition segment, and the speed and pressure co-evolution segment. The rhythm fluctuation overlapping block positioning results include the start-stop frequency aggregation segment, the behavior rhythm jump block, and the restart action continuous distribution segment. The diagnostic feature linkage group block identification includes the behavior aggregation structure block number, the multi-parameter collaborative anomaly composition group, and the linkage state feature. The equipment failure trend diagnostic path set includes the time sequence tracking path, the structure combination pattern, and the state anomaly variation chain.

[0058] See also Figure 2 , the specific steps for obtaining the velocity transition instability position sequence are:

[0059] S111: Collect real-time flow velocity data from the water pump's inlet and outlet channels, compare the flow velocity values ​​at adjacent sampling points with the corresponding time, determine whether there is a change in the sign of the value, and sequentially detect whether there is a reversal of the flow velocity direction by grouping three consecutive sampling points. Generate a flow velocity direction reversal time series by extracting the timestamp values ​​corresponding to the points where reversal occurs.

[0060] The original data sequence collected by the flow sensor is time-aligned to ensure that each sampling point has a uniform time interval. The sampling interval is fixed to 1 second, and an array form of flow velocity changing with time is constructed. The sequence is set to v=[1.2, 1.5, 1.4, 1.0, 1.3, 1.8, 1.6]. The array is segmented using a sliding window method, and 3 consecutive sampling points are extracted from each group to form data segments such as [1.2, 1.5, 1.4], [1.5, 1.4, 1.0], [1.4, 1.0, 1.3], etc. For each group of data segments, it is determined whether there is a direction reversal in its change trend, that is, whether the middle point is at the intersection of the change directions at both ends. In the first group, 1.2 rises to 1.5 and then drops to 1.4, indicating that there is a reversal phenomenon, which establishes the direction of the data segment. The time corresponding to the center point is taken as the reversal node; this operation is repeated until the end of the data sequence, and the timestamp of each point judged to be a reversal is extracted and recorded. For example, the reversal node time is the 3rd second, the 5th second, the 7th second, etc., forming a set of timestamp sequences t=[3,5,7]. This sequence is the candidate flow velocity direction reversal time series. During the execution process, it is also necessary to eliminate the small amplitude change points caused by sensor jitter or error, and use the reversal confirmation threshold to screen, that is, to determine whether the flow velocity difference before and after the reversal is greater than 0.2m / s. If it is less than this value, it is not counted as a valid reversal point. In a set of data [1.0,1.05,1.0], although there is a reversal in the trend, the maximum flow velocity difference is only 0.05m / s, which does not exceed the judgment threshold and needs to be eliminated. The time of the valid reversal point is recorded to form the flow velocity direction reversal time series.

[0061] S112: Call the velocity direction reversal time series, perform a time sequence sorting operation, remove repeated or equally spaced reversal time nodes, filter time segments with continuous fluctuations, and generate a local velocity change time interval set;

[0062] It is necessary to perform a timing adjustment operation on the flow velocity direction reversal time series, sort the time series in ascending order, eliminate the existing disorder, and remove or merge the repeated time points or time gaps less than the set threshold (such as 1 second) in the time series. Set the original sequence to [3, 5, 5, 5.8, 7]. Since the difference between 5 seconds and 5.8 seconds is 0.8 seconds, which is less than the threshold of 1 second, it needs to be merged into 5-second time points, and the repeated 5-second points are removed to obtain [3, 5, 7]. After completing the basic cleaning operation, further judge the degree of aggregation of the reversal time points on the time axis, construct a sliding time window (such as 10 seconds), count the number of reversal points in each sliding window, and mark the segment as a local reversal active segment if the number is greater than or equal to 3. Set the time interval [20, 30] seconds to record 4 reversal points, then mark it as a valid time segment T1=[20, 30]. Extract all intervals that meet the conditions to generate multiple intervals such as T1=[20, 30]. T2 = [55, 65], etc., and the time intervals are arranged in chronological order to complete the time domain positioning of the local flow velocity fluctuation and obtain the local flow velocity change time interval set.

[0063] S113: Based on the local flow velocity change time interval set, the flow velocity change rate of the water pump inlet and outlet channels, the continuous reversal frequency, and the flow velocity difference before and after a single reversal in the corresponding time period are extracted. The numerical value superposition, comparison, and normalization judgment operations are performed for each time period using the formula:

[0064] ;

[0065] Calculate the velocity transition index within the real-time interval, compare the index values ​​corresponding to the time periods, extract the time periods where the transition degree has a sudden change, and obtain the velocity transition instability position sequence;

[0066] in, Represents the flow rate transition degree of the real-time time segment, represents the number of flow velocity reversals within the segment, Representative The difference in flow velocity before and after the reversal, Representative The reversal frequency after the reversal, Representative The rate of change of flow velocity in adjacent time slices during the reversal;

[0067] The formula is useful in that by dividing the frequency of reversals and reversal amplitude Combined with the introduction of the rate of change The gradient term is adjusted and the square root function is used to suppress local abnormal changes to ensure that the value reflects the actual fluctuation trend, as shown in Table 1;

[0068] Table 1 Local flow velocity analysis parameters:

[0069]

[0070] Extract the positions of all reversal points in each time period. During the implementation process, the flow velocity reversal point can be confirmed by the three-point comparison method, that is, to determine whether the flow velocity direction of the second point in each group of three consecutive sampling points has a reverse mutation compared with the previous and next points. Set sampling points A, B, and C. If the flow velocity from A to B is positive growth and from B to C is negative growth, then B is the reversal point. Record the corresponding time of point B and include it in the statistical range of the current time period. On this basis, count the number of reversals in each time period. In actual monitoring, the total number of velocity reversals detected within 1 minute can be used as the index basis for frequency and difference. For the T1 time period, there are 3 reversals in total according to the collection results. Secondly, the difference between the adjacent velocity sampling values ​​before and after each reversal point is calculated and recorded as , if in a reversal, the flow velocity is 1.2m / s at the previous moment and 0.4m / s at the next moment, then m / s, and record the frequency corresponding to the reversal , expressed as the number of reversals per unit time, the sampling time in T1 is set to 60 seconds, and the three reversals occur in 10s, 25s, and 50s respectively, so the frequency is about 2 times / min. The derivative is calculated by the velocity difference between two adjacent time points and the time interval. If the velocity changes from 0.9m / s to 1.4m / s between 10s and 11s, then m / s². After all parameters are obtained, substitute them into the following formula to calculate the velocity transition degree:

[0071] ;

[0072] Taking the T1 time period as an example, substitute the value 、 、 、 , set the other two inversions to be , , , , , , then substitute item by item:

[0073] ;

[0074] The results show that the velocity transition degree in the T1 time period is 1.311, reflecting the severe instability of the velocity structure in this time period. Combining the calculation results of multiple time periods and comparing them with the transition degree change benchmark value of 0.9, when the transition degree is greater than the benchmark value, the corresponding time period is recorded as the transition instability position. The time periods that exceed the transition degree benchmark are summarized to obtain the velocity transition instability position sequence.

[0075] See also Figure 3 ,The specific steps for obtaining the linkage feature distribution map are:

[0076] S211: Based on the time period of the node in the flow velocity jump instability position sequence, the water pump main shaft speed curve data is called to identify the extreme value points in the main shaft speed curve, and the turning points between adjacent extreme value points are distinguished and judged. The time period with a continuous reversal trend is selected based on the change amplitude of the interval curvature to obtain the speed reversal section;

[0077] The time series is segmented according to different time nodes, and the spindle speed data within each period is extracted. If the sampling frequency is 50Hz, 50 speed points are extracted per second. In actual working conditions, the pump operation period is set to 08:00:00 to 08:10:00, and the total amount of speed data extracted is 30,000 points. A curve trend graph is drawn for this batch of data, and the local extreme points, that is, the local maximum and minimum values, are determined. For example, the points in a certain section are [1450, 1448, 1446, 1449, 1451, 1447, 1445] , where 1451 is the maximum value and 1445 is the minimum value. According to the trend change between adjacent extreme values, the turning point 1449 is identified. The curvature of the turning interval is further identified for the local periodic segment formed between the turning points. The curve trend is calculated using the difference method. The second-order difference values ​​of the five adjacent points are set to [-2, -1, 0, 2] respectively. The average absolute change is calculated to be 1.25. If the curvature change threshold is set to 1.2, it is considered to be a significant reentry trend. The segments that continuously meet the reentry conditions are then spliced ​​and integrated, and the isolated segments with less than 3 points are eliminated to obtain the speed reentry segment.

[0078] S212: Based on the time range within the speed reversal section, all pressure sensor signal data within the time period is retrieved, sequentially sorted in chronological order, and a pressure change curve is recorded according to the sampling point sequence. Each continuously rising pressure trend segment is numbered and distinguished, and the corresponding time period is mapped back to the sequence identifier of the reversal section to generate a reversal section pressure rising sequence comparison table;

[0079] Extract the pressure data recorded by the pressure sensor and call the corresponding time data point by point according to the data sampling frequency (e.g., 10 Hz). Set a certain return segment from 08:03:10 to 08:03:20, extract 100 pressure data points in total, and record the data point sequence such as [0.35, 0.36, 0.37, 0.36, 0.38, 0.39, 0.4]. Sort the pressure value change trend and segment the continuous growth segment. Set the first three points in the sequence to show an upward trend to form an upward sequence, and the last three points to show an upward trend again to form another segment, named P1 and P2 respectively. Establish a mapping relationship between them and the corresponding return segments F1 and F2. At the same time, record the time start and end position, maximum pressure difference, and growth rate of each rising segment. Use the mapping list to form a correspondence table between the return segment and the pressure rising segment, that is, the return segment pressure rising sequence comparison table.

[0080] S213: Call the pressure rise sequence comparison table of the return section to identify the corresponding speed return number and pressure rise section number within the time window, using the formula:

[0081] ;

[0082] Calculate the linkage value between speed return and pressure jump, sort the windows according to the linkage value, and obtain the linkage feature distribution map;

[0083] in, Represents the linkage value between speed return and pressure jump, Representative The number of turns within the speed turn-back section, The number of pressure rise sections corresponding to the representative section, Indicates the duration of the segment. For the The maximum pressure slope within the window, Indicates the pressure difference before and after the speed turning point in the window. Indicates the total number of return segments in the window. is the number of speed return sections, is the number of time windows;

[0084] Extract the number of reentry segments and the corresponding number of pressure rise segments contained in the time window. If 5 seconds is used as a window unit, 3 reentry segments and 2 pressure rise segments will be generated within a certain period of time. Use the formula to calculate the linkage value.

[0085] In the formula, the first term represents the average difference between the number of reentry segments and the number and duration of pressure in all reentry segments. The second term is the joint normalized value of the pressure change slope and pressure difference based on the number of reentry segments in each time window.

[0086] The formula is used to calculate the linkage value between speed reversal and pressure jump. It consists of two parts. The first part is to average the linkage strength of the speed reversal points and calculate the number of reversals at each reversal point. and pressure jump characteristic value The difference is calculated and the absolute value is taken to reflect the consistency between the speed change and the pressure rise. The second part is to count the impact of the pressure mutation in each time window on the linkage degree. With the maximum pressure slope Multiply them, normalize by the number of return segments plus 1, and finally sum them. The entire formula comprehensively considers the number of speed returns, pressure rise amplitude, duration, and instantaneous mutation degree, quantitatively reflecting the dynamic linkage relationship between the two. The larger the value, the stronger the linkage, which is suitable for feature extraction and fault identification analysis.

[0087] The parameters are explained as follows:

[0088] Indicates the The number of turns in each section is obtained by detecting the extreme speed sequence. For example, if the fluctuation sequence in the section is [1480, 1478, 1481, 1476], the number of turns is 2.

[0089] Indicates the number of pressure rising sections, divided by continuous monotonically increasing sections. For example, if there are 2 monotonically increasing sections in this section, the number is 2.

[0090] Indicates duration, records the start and end time of the segment, and calculates the time length. For example, from 08:03:12 to 08:03:22 is 10 seconds;

[0091] Indicates the maximum value of the pressure slope, which is calculated by dividing the pressure difference between adjacent points in each segment by the time difference. For example, if the point value changes from 0.4 to 0.48 in 2 seconds, it is 0.04MPa / s;

[0092] Indicates the speed turning point corresponding to the front and rear pressure difference, for example, from 0.38 to 0.46, the difference is 0.08;

[0093] Indicates the number of return segments in the window;

[0094] Indicates the number of all return sections, which is 4 in this example;

[0095] Indicates the number of all time windows, set to 4;

[0096] The actual values ​​are shown in the table:

[0097] As shown in Table 2, the values ​​of each parameter are as follows:

[0098] Table 2 Linkage analysis example parameters:

[0099]

[0100] Substitute the parameters into the formula:

[0101] ;

[0102] ;

[0103] The first item expands:

[0104] ;

[0105] The second expansion:

[0106] ;

[0107] Substitute into the formula for calculation:

[0108] ;

[0109] The results show that the linkage value between the return and the pressure jump in the current time window is 2.0703. The linkage value between the return and the pressure jump is a comprehensive indicator to measure the correlation strength between the speed change and the pressure fluctuation. The larger the value, the more significant the linkage between the two. It can be used to identify abnormal events (such as surge, mutation, etc.) and obtain significant characteristic time windows through sorting for further analysis or modeling. This value can be used to further generate a linkage feature distribution map. The benefit of the formula is that by normalizing the difference between the square root relationship of the number of returns and the number of pressures, and introducing the joint weight score form of the pressure slope and the difference, the coordinated trend of speed fluctuation and pressure fluctuation in the dynamic process is comprehensively evaluated, and a numerical basis for mapping a clear linkage relationship is established.

[0110] See also Figure 4 ,The specific steps for obtaining the positioning results of the rhythm fluctuation overlapping blocks are:

[0111] S311: Extracting the pump start and stop action records based on the linkage time segments calibrated by the linkage feature distribution map, identifying the time span between each start and stop time, and dividing the operation into multiple operation segments in chronological order to generate an operation time division sequence;

[0112] The start and stop time data of the water pump in the record are read one by one and mapped to a unified time series coordinate system. The start time and the corresponding stop time are extracted to form a time segment set. In a specific embodiment, the first start time is set to 09:30:00 and the corresponding stop time is 09:32:00. The running time of this segment is 120 seconds. The start-stop pairs are extracted in sequence to obtain multiple running segments. Then, they are sorted in the order of the start time. The segment sequences obtained are set to [09:30-09:32], [09:35-09:37], and [09:40-09:42]. The duration of the running segments is uniformly converted to seconds, that is, the running segment lengths are 120 seconds, 120 seconds, and 120 seconds. The time intervals between consecutive running segments are further counted. For example, the intervals between the above segments are 180 seconds and 180 seconds respectively. Combined with the time characteristics of each running segment, a running length division sequence is generated.

[0113] S312: Call the running time division sequence to classify the start and stop behaviors in adjacent running segments, count the start position and end position of each running segment and the interval between adjacent segments, and use the formula based on the degree of time overlap, the running intensity within the segment and the interval length:

[0114] ;

[0115] Calculate the restart behavior difference between adjacent segments. Based on the clustering pattern of the difference values ​​in the sequence, analyze the segments where the restart behavior is concentrated to obtain the restart segment indicator sequence.

[0116] in, Represents the restart behavior difference value, Representative Segment end time, Representative Segment start time, 、 Respectively Paragraph and Section The running time of the segment, 、 Respectively Paragraph and Section The operating intensity of the segment, For the Number of accelerations during segment operation, For the The total number of actions in the segment;

[0117] The formula is used to calculate the restart behavior difference between adjacent running segments , which measures the degree of difference in the start-stop characteristics when the equipment switches between operation phases. The formula consists of three parts:

[0118] The first term is the time difference normalization term, , represents the time interval between the end of segment o and the start of segment b, normalized to eliminate the effect of running time; the second term is the average value of the operating intensity of adjacent segments, reflecting the impact of load fluctuation on restart behavior; the third item The acceleration behavior impact represents the weighted sum of each acceleration degree and behavior frequency within the segment, reflecting the cumulative impact of operating habits on the current segment behavior.

[0119] Overall, the difference The smaller the value, the more consistent the restart characteristics of adjacent segments are, which is conducive to cluster analysis of typical operating modes and calibration of restart segments. This indicator is often used for device behavior identification, energy-saving analysis, or operation and maintenance strategy optimization.

[0120] Table 3 Restart behavior analysis parameters:

[0121]

[0122] Read the start and end time, running duration, running intensity and other information of each running segment, and compare the time intervals between adjacent segments and the changes in running status. In specific operations, if the first The end time of the segment is 260 seconds. The start time of the segment is 300 seconds, the interval between adjacent segments is 40 seconds, the corresponding running time is 120 seconds and 100 seconds, and the corresponding running intensity is set to 0.7 and 0.6. The denominator is normalized by the time difference;

[0123] Will 、 、 、 、 、 、 Substituting in:

[0124] ;

[0125] ;

[0126] This result shows that the Paragraph and Section There are moderate to high differences in restart behavior characteristics between segments. The restart behavior difference value is used to measure the degree of difference in start-stop characteristics between adjacent running segments, taking into account the influence of the time interval between segments, running intensity and acceleration behavior. The smaller the difference value, the smoother the running switch and the more consistent the restart behavior. Further, the difference between multiple groups of segments is used to measure the difference in start-stop characteristics between adjacent running segments. Collect and determine whether they are concentrated in a certain time window. If multiple If the values ​​are all higher than 0.45, it can be determined that there are intensive start-stop behaviors in the segment, and the restart segment indicator sequence can be obtained. The benefit of the formula is that by introducing the standardized ratio of the operating segment time difference, the operating intensity superposition term and the correction factor of the acceleration density, the restart behavior differences of operating segments of different natures can be judged under a unified standard.

[0127] S313: Matching and locating the restart segment indicator sequence in combination with the corresponding points on the time axis, identifying the intersection relationship between the segment and the time spectrum, identifying the number and distribution density of signal changes within the intersection segment, and obtaining the positioning result of the rhythm fluctuation overlapping block;

[0128] The time axis index aligned with each segment is called, and the intersection judgment is performed by comparing the time index with the event peak point position in the linkage map to identify whether the current segment coincides with the key abnormal period. If there is an overlap, the sensor fluctuation data in the segment is extracted. In practical applications, if the current restart segment index range is from 800s to 1000s, and the abnormal peaks in the linkage map appear densely in the interval from 900s to 950s, there is an effective overlap. The number of signal point changes in the overlapping interval is extracted, and 34 signal mutations are set in the segment, with an average duration of 2.5s for each mutation. Each fluctuation mutation is classified and aggregated, and the proportion and average interval of each type of fluctuation mutation in the segment are recorded. A secondary differential analysis is performed on the mutation event sequence to identify the rhythmic change pattern and obtain the positioning result of the rhythmic fluctuation overlapping block.

[0129] See also Figure 5 ,The steps for obtaining the diagnostic feature linkage block identifier are as follows:

[0130] S411: Call the time period identified in the rhythm fluctuation overlapping block positioning result, retrieve the flow velocity change and direction records, perform incremental accumulation based on the flow velocity difference sequence before and after the direction reversal point and the time axis sequence, extract the corresponding slope mutation segment as the direction change node sequence, and generate the flow velocity direction change interval value;

[0131] Analyze the start and end time, set this time period as the data extraction range, and obtain the flow rate change record and direction change mark sequence within the corresponding time period. The flow rate data can be derived from the data stream collected by the flow rate sensor at a fixed sampling frequency, and the direction record is a derived parameter, which represents the instantaneous change state of the medium flow direction. In this time period, the direction reversal point needs to be identified point by point, which is manifested as the direction mark changing from positive to negative or from negative to positive. The flow rate data within a certain time range before and after the reversal point is extracted, and a difference sequence of continuous flow rate values ​​is established to form a change trend curve before and after the direction change. On this basis, the specific time period of the mutation trend in the difference curve is analyzed, and the slope change area is compared with the time index of the direction reversal point. The local time period with a prominent slope change amplitude is extracted, and this type of time period is integrated, classified, and deduplicated, and uniformly marked as a direction change node interval to generate a flow rate direction change interval value.

[0132] S412: Based on the flow velocity direction variation interval value, the speed data is compared with the set reference value, and the time period of the difference amplitude is selected to form a reversal point set. The time interval between consecutive reversal points is calculated, and the interval values ​​are interval-merged. The speed fluctuation amplitude and the number of reversals are counted to generate a speed reversal cumulative ratio;

[0133] The speed data is retrieved for further processing. At the initial stage of the operation, the speed value corresponding to each time point needs to be compared with the reference set value to determine whether the numerical difference is within a stable range. If the difference within a certain period remains stable at continuous sampling points, it can be considered that there are signs of speed reversal in this period. The time period set that meets this condition is uniformly constructed as a reversal point set. The time intervals between the reversal points are further sorted and classified, and the time difference between two adjacent reversal points is calculated. The time difference is then analyzed for distribution to determine whether there are concentrated interval clusters. The speed fluctuation range within each reversal interval needs to be further determined, and the number of reversal actions is counted to identify time segments where multiple reversals occur in a relatively short period of time and the overall change amplitude is obvious. The intervals are sorted and summarized to form a new indicator set to obtain the cumulative ratio of speed reversal within the time period.

[0134] S413: Calling the flow velocity direction change interval value and the speed reversal cumulative ratio, extracting the pressure change data and start-stop records within the overlapping time period, analyzing the pressure increase amplitude per unit time, counting the number of starts and stops within the time period, performing multiple parallel classification operations on the number of flow velocity reversal points, speed reversal density, pressure change trend, and start-stop frequency distribution, identifying the common classification results of the four features, and generating a diagnostic feature linkage group identifier;

[0135] To call the flow velocity direction change interval value and the cumulative ratio of speed reversal, the time periods covered by the two must be compared and processed. The time intersection must be selected as the core time period for subsequent processing. Within this intersection period, pressure monitoring data and start-stop state sequence data must be synchronously called. The pressure monitoring data is derived from the real-time recording of the pressure sensor, while the start-stop state sequence is based on the record of the equipment start-stop signal changes. The continuous pressure change trend within this time period is extracted, and the pressure increase within the time period is integrated with the number of equipment starts and stops per minute. After initially constructing the pressure change curve and start-stop record table, the number of flow velocity reversals, the frequency of speed reversals, the shape of the pressure upward curve, and the number of start-stop signals are jointly analyzed by time slice. Multiple feature vector combinations are constructed and classified. The combination items are independently marked when the four types of features are concentrated in the same time period. A set of feature structure combinations with diagnostic significance is established. Such combinations not only reflect the coordinated changes of flow velocity reversals and speed reversals, but also reflect the multi-point resonance relationship between pressure changes and start-stop behaviors. Such combination structures are archived and output as diagnostic feature linkage block identifiers.

[0136] See also Figure 6 ,The steps for obtaining the equipment failure trend diagnosis path set are as follows:

[0137] S511: Using the diagnostic feature linkage block identifier, the linkage blocks are arranged along the time axis according to the block occurrence time sequence, the structural components corresponding to the blocks are extracted, the repetition of the components between consecutive blocks is compared, the structural repetition degree within the time continuous interval is determined, and the repetition ratio of the structural components within the interval is calculated to form a repetition degree sequence interval;

[0138] Based on operating data collected during equipment operation, such as bearing temperature, drive load, pressure fluctuation, and current amplitude, a multi-dimensional condition monitoring log is established. Combined with the pre-set anomaly identification mechanism for diagnosis, monitoring data with short-term behavioral deviations are grouped into linkage blocks. Timestamps are extracted and sorted according to the trigger time of each block, and a timeline arrangement structure is constructed. The structural components involved in each block are extracted one by one. For example, pressure can be divided into the main control valve body, feedback spring, and buffer. After the component items are extracted, the structural components of two adjacent blocks on the timeline are cross-compared. The overlap and cross-occurrence of structural types are counted, and whether they appear continuously in time is determined. When the time interval of continuous occurrence is within the set threshold and the degree of structural repetition exceeds a certain ratio, the structural combination segment is recorded as a continuous interval. Combined with the records of multiple subsequent continuous intervals, intervals with high component item repetition are grouped together and summarized into a component repetition sequence interval. During the maintenance period of a certain unit, if blocks containing the main control valve body and feedback spring combination appear continuously, forming a continuous structure on the timeline and repeating frequently, a repetition sequence interval is formed.

[0139] S512: Calling the composition repetition sequence interval, identifying paths where the number of consecutive occurrences of structural components exceeds the set number in the time sequence, marking them as traceable paths, summarizing the reversal direction, return position, pressure change trend and start / stop signal status of the blocks in the path, and generating a path structure feature combination sequence;

[0140] Identify whether the structural components appear repeatedly in multiple time periods, number the structural segments, and establish a path mapping table with the unique identifier of the components in the block. For the numbered segments on the time axis, determine one by one whether the structural components appear repeatedly in multiple segments. If a combination, such as elastic bearing + sealing component + guide block, remains unchanged or changes very little in three or more consecutive segments, then mark its corresponding path as a traceable path. On this basis, extract the operating parameters of the blocks in the path, such as axial reversal direction, return point sequence number, pressure rise or fall trend, and each group The start-stop signal status recorded by the block is set. During the several starting processes of a hydraulic actuator, its path block display is reversed to forward-reverse-forward, the return point returns from position 4 to position 2, the pressure gradually decreases from 5.6MPa to 3.9MPa, and the start-stop status is continuously enabled to closed. This is recorded as the path characteristic behavior set, and the characteristic behavior sequence is constructed into a path structure feature combination sequence according to the path order. This sequence is used for subsequent trend behavior identification and path typing to ensure that each type of traceable path has comparable behavior feature index data to generate a path structure feature combination sequence.

[0141] S513: Based on the path structure feature combination sequence, the reversal direction, return position, pressure change trend, and start / stop signal status of the corresponding block in the path are extracted as similar sequences. The change frequency and direction change degree of multiple sequences in the same path are compared to determine whether they belong to the same characteristic pattern. The path types are then divided to generate a set of equipment failure trend diagnosis paths.

[0142] The feature items extracted from each block in the path are further classified to form a reversal direction sequence, a return position sequence, a pressure change trend sequence and a start-stop state sequence. In the path of a certain motor, the reversal direction forms an alternating pattern such as positive-negative-positive-negative, the return position is 9-7-6, the pressure trend is an upward state, and the start-stop state is start-start-stop. By comparing the sequence features, multiple sequence features under the same type of path are merged to identify their common indicators in terms of change frequency, direction switching number, trend fluctuation range, etc., and determine whether they belong to the same The change pattern is identified, and the paths are classified into the same type based on the identification results. If the pressure trend in a certain type of path is generally a continuous rise, the reversal direction is maintained in periodic alternation, and the start-stop state is mainly three starts and one stop, it can be determined as a typical periodic stress rise path. The trend extension length of each type of path is further counted. For example, if the total time length of the trend is continuously maintained for 48 minutes in the monitoring log, and the return position changes within 3 positions, then the path is marked as a type of trend structure combination and distinguished from the remaining paths to form a set of equipment failure trend diagnosis paths.

[0143] The hydraulic engineering equipment fault diagnosis system is used to execute the hydraulic engineering equipment fault diagnosis method described above, and the system includes:

[0144] The data sorting module collects real-time flow velocity data from the pump's inlet and outlet channels, determines the direction of flow velocity change at adjacent sampling points, locates three-point combination sections with reverse changes, and generates a sequence of flow velocity jump instability locations.

[0145] The structural positioning module uses the velocity transition instability position sequence to identify the extreme value of the speed curve and segmentally delineate the return section. In the return section, the pressure change data is extracted in chronological order. The return number is matched with the pressure change trend and paired to generate a linkage feature distribution map.

[0146] Based on the linkage feature distribution map, the state linkage module screens the time periods when the speed changes and the pressure jumps change synchronously, extracts each continuous operation section between the start and stop of the water pump, statistically classifies the frequency of start and stop actions and the adjacent interval time, and generates the positioning results of the rhythm fluctuation overlapping blocks;

[0147] The behavior rhythm module calls the rhythm fluctuation overlapping block positioning results, performs time axis mapping on the marked time period, extracts the behavior segments with linkage features, compares the distribution intervals and duration of the segments within the operation cycle, and generates diagnostic feature linkage block identifiers;

[0148] The path classification module calls the linkage feature participants according to the diagnostic feature linkage group block identifier, classifies the distribution patterns in multiple overlapping segments, marks the parameter combinations of consistent features and counts the number of recurrences, and generates a set of equipment failure trend diagnostic paths.

[0149] The above are merely preferred embodiments of the present invention and do not limit the present invention in any other form. Any technician familiar with the profession may use the technical content disclosed above to change or modify it into an equivalent embodiment with equivalent changes and apply it to other fields. However, any simple modification, equivalent change and modification made to the above embodiment based on the technical essence of the present invention without departing from the content of the technical solution of the present invention shall still fall within the scope of protection of the technical solution of the present invention.

Claims

1. A method for diagnosing faults in water conservancy engineering equipment, characterized in that: The following steps are involved: S1: Collect real-time flow velocity data from the pump's inlet and outlet channels, determine the direction of flow velocity change at adjacent sampling points, locate the section with direction reversal among every three consecutive sampling points, mark the time point of reversal, identify local flow velocity structure changes, and generate a sequence of flow velocity jump instability locations; S2: Based on the time period of the node in the flow velocity jump instability position sequence, extract the pump main shaft speed curve data, identify the turning position between the extreme values ​​and divide it into the return interval, evaluate the linkage mapping relationship between the speed return and the pressure jump, and generate a linkage feature distribution map; S3: Extracting the start and stop action records of the water pump according to the linkage time segments calibrated by the linkage feature distribution map, dividing the continuous operation time in sequence, and performing time series classification on the start and stop behaviors in adjacent operation segments to generate the rhythm fluctuation overlapping block positioning results; S4: Call the time period identified in the rhythm fluctuation overlapping block positioning result, integrate the information into a structural block, and perform combined condition screening to see whether flow velocity reversal aggregation, speed return accumulation and start-stop concentration phenomena occur simultaneously in the structural block. If satisfied, execute the linkage mark and generate a diagnostic feature linkage group identifier.

2. The method for diagnosing faults in water conservancy engineering equipment according to claim 1, characterized in that: The flow rate jump instability position sequence includes the direction reversal node time point, the reversal intensive interval, and the change trend switching segment. The linkage feature distribution map includes the return structure distribution interval, the pressure trend superposition segment, and the speed and pressure coordinated evolution segment. The rhythm fluctuation overlapping block positioning result includes the start-stop frequency aggregation segment, the behavior rhythm jump block, and the restart action continuous distribution segment. The diagnostic feature linkage group block identifier includes the behavior aggregation structure block number, the multi-parameter collaborative anomaly composition group, and the linkage status feature.

3. The method for diagnosing faults in water conservancy engineering equipment according to claim 1, characterized in that: The steps for obtaining the flow velocity transition instability position sequence are specifically as follows: S111: Collect real-time flow velocity data from the water pump's inlet and outlet channels, compare the flow velocity values ​​at adjacent sampling points with the corresponding time, determine whether there is a change in the sign of the value, and sequentially detect whether there is a reversal of the flow velocity direction by grouping three consecutive sampling points. Generate a flow velocity direction reversal time series by extracting the timestamp values ​​corresponding to the points where reversal occurs. S112: calling the velocity direction reversal time series, performing a time sequence sorting operation, eliminating repeated or evenly spaced abnormal reversal time nodes, screening time segments with continuous fluctuations, and generating a local velocity change time interval set; S113: Based on the local flow velocity change time interval set, the flow velocity change rate of the water inlet and outlet channels of the water pump in the corresponding time period, the continuous reversal frequency and the flow velocity difference before and after a single reversal are extracted, and numerical superposition, comparison and normalization judgment operations are performed on each time period to calculate the flow velocity transition degree index in the real-time interval. The index values ​​corresponding to the time periods are compared in time series, and the time periods where the transition degree has a sudden change are extracted to obtain a flow velocity transition instability position sequence.

4. The method for diagnosing faults in water conservancy engineering equipment according to claim 3, characterized in that: The steps for obtaining the linkage feature distribution map are specifically as follows: S211: Based on the time period of the node in the flow velocity jump instability position sequence, calling the water pump main shaft speed curve data, identifying the extreme value points in the main shaft speed curve, distinguishing and judging the turning positions between adjacent extreme value points, and screening the time period that continuously shows a reversal trend based on the change amplitude of the interval curvature to obtain the speed reversal section; S212: Based on the time range within the speed reversal section, all pressure sensor signal data within the time period are retrieved, continuously sorted in chronological order, and a pressure change curve is recorded according to the sampling point sequence. Each continuously rising pressure trend segment is numbered and distinguished, and the corresponding time period is mapped back to the sequence identifier of the reversal section to generate a reversal section pressure rising sequence comparison table; S213: Call the pressure rise sequence comparison table of the return section, identify the corresponding speed return number and pressure rise section number in the time window, calculate the speed return and pressure jump linkage value, sort the windows according to the linkage value, and obtain the linkage feature distribution map.

5. The method for diagnosing faults in water conservancy engineering equipment according to claim 4, characterized in that: The steps for obtaining the positioning result of the rhythm fluctuation overlapping block are specifically as follows: S311: Extracting the start and stop action records of the water pump according to the linkage time segment calibrated by the linkage characteristic distribution map, identifying the time span between each start time and stop time, and dividing the operation into multiple operation segments in chronological order to generate an operation time division sequence; S312: Calling the run time division sequence, classifying the start and stop behaviors in adjacent run segments, counting the start and end positions of each run segment and the intervals between adjacent segments, calculating the restart behavior difference between adjacent segments based on the degree of time overlap, the run intensity within the segment, and the interval length, analyzing the segments where the restart behaviors are concentrated based on the clustering pattern formed by the difference values ​​in the sequence, and obtaining a restart segment indicator sequence; S313: Matching and positioning are performed based on the restart segment indicator sequence in combination with corresponding points on the time axis, identifying the intersection relationship between the segment and the time spectrum, identifying the number and distribution density of signal changes in the intersection segment, and obtaining the rhythm fluctuation overlapping block positioning result.

6. The method for diagnosing faults in water conservancy engineering equipment according to claim 5, characterized in that: The steps for obtaining the diagnostic feature linkage block identifier are as follows: S411: calling the time period identified in the rhythm fluctuation overlapping block positioning result, retrieving the flow velocity change and direction records, performing incremental accumulation based on the flow velocity difference sequence before and after the direction reversal point and the time axis sequence, extracting the corresponding slope mutation segment as the direction change node sequence, and generating the flow velocity direction change interval value; S412: Based on the flow velocity direction variation interval value, the speed data is compared with a set reference value, and a time period of a difference amplitude is selected to form a reversal point set. The time interval between consecutive reversal points is calculated, and the interval values ​​are interval-merged. The speed fluctuation amplitude and the number of reversals are counted to generate a speed reversal cumulative ratio. S413: Call the flow direction change interval value and the speed reversal cumulative ratio, extract the pressure change data and start-stop records in the overlapping time period of the two, analyze the pressure increase amplitude per unit time, count the number of starts and stops in the time period, perform multiple parallel classification operations on the number of flow reversal points, speed reversal density, pressure change trend and start-stop frequency distribution, identify the common classification results of the four features, and generate a diagnostic feature linkage block identifier.

7. The method for diagnosing faults in water conservancy engineering equipment according to claim 1, characterized in that: The method further comprises step S5: S5: Based on the diagnostic feature linkage block identifier, the differentiated linkage block structure is arranged according to the time dimension, and the continuity and structural component fit on the time axis are calculated. When three or more groups of highly consistent structures appear continuously, the structural block is marked as a traceable failure path, and the matching patterns of reversal, return, pressure trend and start-stop characteristics in the path are classified to generate a set of equipment failure trend diagnostic paths; The equipment failure trend diagnosis path set includes a time sequence tracking path, a structural combination pattern, and a state abnormal variation chain.

8. The method for diagnosing faults in water conservancy engineering equipment according to claim 7, characterized in that: The steps for obtaining the equipment failure trend diagnosis path set are specifically as follows: S511: using the diagnostic feature linkage block identifier, arranging the linkage blocks along the time axis according to the block occurrence time sequence, extracting the structural components corresponding to the blocks, comparing the repetition of the components between consecutive blocks, determining the structural repetition degree within the time continuous interval, calculating the repetition ratio of the structural components within the interval, and forming a repetition degree sequence interval; S512: Calling the component repetition sequence interval, identifying paths in which the number of consecutive occurrences of structural components exceeds a set number in the time sequence, marking them as traceable paths, summarizing the reversal direction, return position, pressure change trend and start / stop signal status of the blocks in the path, and generating a path structure feature combination sequence; S513: Based on the path structure feature combination sequence, the reversal direction, return position, pressure change trend and start-stop signal status of the corresponding block in the path are extracted as similar sequences, and the change frequency and direction change degree of multiple types of sequences in the same path are compared to determine whether they belong to the same feature mode, and the path types are divided to generate a set of equipment failure trend diagnosis paths.

9. A water conservancy project equipment fault diagnosis system, characterized in that: The system is used to implement the water conservancy project equipment fault diagnosis method according to any one of claims 1 to 8, and the system includes: The data sorting module collects real-time flow velocity data from the pump's inlet and outlet channels, determines the direction of flow velocity change at adjacent sampling points, locates three-point combination sections with reverse changes, and generates a sequence of flow velocity jump instability locations. The structure positioning module calls the flow velocity transition instability position sequence, identifies the extreme value of the speed curve and delineates the return section in sections, extracts the pressure change data in chronological order in the return section, matches the return number with the pressure change trend and pairs them to generate a linkage feature distribution map; Based on the linkage feature distribution map, the state linkage module selects the time period when the speed change and pressure jump change synchronously, extracts each continuous operation section between the start and stop of the water pump, statistically classifies the frequency of start and stop actions and the adjacent interval time, and generates the positioning result of the rhythm fluctuation overlapping block; The behavior rhythm module calls the rhythm fluctuation overlapping block positioning result, performs time axis mapping on the marked time period, extracts the behavior segments with linkage features, compares the distribution interval and duration of the segments within the operation cycle, and generates a diagnostic feature linkage group block identifier; The path classification module calls the linkage feature participants according to the diagnostic feature linkage group block identifier, aggregates the distribution patterns in multiple overlapping segments, marks the parameter combinations of consistency features and counts the number of recurrences, and generates a set of equipment failure trend diagnostic paths.

Images