Water affair flow burr data detection method and system and readable medium

Abstract

The invention relates to the technical field of water affair data processing, in particular to a water affair flow burr data detection method and system and a readable medium. The method comprises the following steps that first flow time sequence data, first pump speed time sequence data and pipe network pressure time sequence data are synchronously collected, and water affair flow related time sequence data are generated; segmenting and intercepting the water affair flow related time sequence data based on a preset sliding time window to generate a plurality of candidate burr data segments; for each candidate burr data segment, calculating a flow change rate absolute value, a first pump speed change rate absolute value, a first pressure change rate absolute value, a standard difference of adjacent sampling points and a range ratio in the data segment, and generating five types of dynamic parameters; obtaining current pump set operation state parameters; and constructing a dynamic discrimination threshold group according to the current pump set operation state parameters. According to the invention, systematic deviation caused by glitch data interference is effectively avoided.

Description

Technical Field

[0001] The present invention relates to the technical field of water service data processing, and in particular to a method, system and readable medium for detecting water service flow spike data. Background Art

[0002] Early water supply systems mainly relied on mechanical flow meters and manual inspections. The data acquisition frequency was low and lagged, making it difficult to capture instantaneous flow changes. With the progress of sensor technology, digital monitoring systems gradually replaced traditional methods, achieving real-time data acquisition and transmission. However, with the increasing complexity of water supply systems, especially the widespread application of variable frequency speed control pumps, smart water meters and remote monitoring systems, the accuracy and reliability of flow data face new challenges. In modern water supply systems, the real-time collected flow data not only needs to reflect the water usage situation at the user end, but also needs to take into account the equipment operation status and the dynamic of pipe network pressure. However, the adaptability of existing monitoring technologies under complex working conditions is still insufficient. For example, when the pump group is switching between start and stop or during variable frequency speed regulation, the transient changes in flow data are easily affected by equipment electromagnetic interference and sudden water usage behaviors at the user end, forming pulse-type spike data in the millisecond to second range. These spike data will interfere with normal flow monitoring and even cause systematic deviations in hydraulic model calculations.

[0003] Existing detection algorithms based on moving average or fixed thresholds can identify continuous anomalies under normal working conditions, but during the transient process of pump group start-stop switching, they often misjudge the flow step caused by normal regulation as a spike signal. This affects the accuracy of monitoring data. Summary of the Invention

[0004] Based on this, it is necessary for the present invention to provide a method, system and storage medium for detecting water service flow spike data to solve at least one of the above technical problems.

[0005] To achieve the above object, a method for detecting water service flow spike data includes the following steps: Step S1: Synchronously collect the first flow time series data, the first pump speed time series data and the pipe network pressure time series data to generate water service flow-related time series data; segment and intercept the water service flow-related time series data based on a preset sliding time window to generate multiple candidate spike data segments; Step S2: For each candidate spike data segment, calculate its absolute value of flow change rate, absolute value of first pump speed change rate, absolute value of first pressure change rate, standard deviation of differences between adjacent sampling points and ratio of range within the data segment to generate five types of dynamic parameters; Step S3: Obtain the current pump group operation status parameters; construct a dynamic discrimination threshold group according to the current pump group operation status parameters; Step S4: Compare the five types of dynamic parameters with the corresponding thresholds in the dynamic discrimination threshold group respectively. If at least three of the five types of dynamic parameters exceed the thresholds, mark the corresponding candidate glitch data segments as suspected glitch data segments; Step S5: Conduct multi-level verification on the suspected glitch data segments. The multi-level verification includes pump speed change leading verification, pressure change leading verification, pump unit start-stop status signal verification, and historical similar working condition verification. If any verification fails, remove the suspected marks in the suspected glitch data segments and output the candidate glitch data segments that pass the verification as the final glitch data.

[0006] The present invention can capture the glitch characteristics in the instantaneous flow rate change by synchronously collecting the time series data of flow rate, pump speed, and pipeline network pressure and segmenting and intercepting them in combination with a sliding time window, avoiding the omission or misjudgment of glitch signals caused by data segmentation. The dynamic parameter calculation method can comprehensively reflect the change characteristics of flow rate, pump speed, and pressure. Combining the construction of the dynamic discrimination threshold group, it can adjust the thresholds in real time according to the operating state of the pump unit, so as to adapt to the flow rate change characteristics under different working conditions. Through the multi-level verification mechanism, through the pump speed change leading verification, pressure change leading verification, pump unit start-stop status signal verification, and historical similar working condition verification, it further distinguishes normal regulation from glitch signals, significantly reducing the misjudgment rate. In summary, the present invention not only improves the accuracy and reliability of monitoring data, but also enhances the adaptability of the system under complex working conditions, effectively avoiding systematic deviations caused by the interference of glitch data.

[0007] Preferably, the segmenting and intercepting of the time series data related to water service flow rate based on a preset sliding time window in step S1 includes: Segment and intercept the time series data related to water service flow rate based on a preset sliding time window to generate multiple candidate glitch data segments. The length of the preset sliding time window is 0.5 seconds to 5 seconds, and the sliding step is 1 / 5 to 1 / 2 of the window length; Each candidate glitch data segment needs to meet the following conditions: The number of flow rate sampling points in the candidate glitch data segment is greater than or equal to 50; The candidate glitch data segment covers at least one fluctuation interval with a flow rate change rate exceeding ±0.1.

[0008] By optimizing the length and step size settings of the sliding time window, the present invention ensures the rationality and effectiveness of the candidate glitch data segments. The length of the sliding time window is set between 0.5 seconds and 5 seconds, which can effectively capture the glitch features in the instantaneous flow rate changes, while avoiding the omission of glitch signals caused by an overly long window or misjudgment caused by an overly short window. The sliding step size is 1 / 5 to 1 / 2 of the window length, ensuring the continuity and coverage of the data segments and avoiding the discontinuity of data segmentation. In addition, the candidate glitch data segments need to meet the condition that the number of flow rate sampling points is greater than or equal to 50, ensuring the representativeness of the data segments and avoiding misjudgment caused by insufficient data volume. The condition of covering at least one fluctuation interval with a flow rate change rate exceeding ±0.1 ensures that the candidate glitch data segments can capture obvious flow rate change features and improves the recognition accuracy of glitch data.

[0009] Preferably, step S2 includes the following steps: Step S21: Calculate the data segment duration of the candidate glitch data segment; Step S22: Extract the maximum flow rate value and the minimum flow rate value within the candidate glitch data segment, and record the quotient of the absolute value of the difference between the maximum flow rate value and the minimum flow rate value divided by the data segment duration as the absolute value of the flow rate change rate; Step S23: Obtain the calculation times identifier of the current pump speed change rate; determine the pump speed change lag time compensation coefficient according to the calculation times identifier of the current pump speed change rate; extract the maximum pump speed value and the minimum pump speed value within the candidate glitch data segment, and record the quotient of the absolute value of the difference between the maximum pump speed value and the minimum pump speed value divided by the pump speed change lag time compensation coefficient as the absolute value of the first pump speed change rate; Step S24: Obtain the calculation times identifier of the current pressure change rate; determine the pressure change lag time compensation coefficient according to the calculation times identifier of the current pressure change rate; extract the maximum pipeline network pressure value and the minimum pipeline network pressure value within the candidate glitch data segment, and record the quotient of the absolute value of the difference between the maximum pipeline network pressure value and the minimum pipeline network pressure value divided by the pressure change lag time compensation coefficient as the absolute value of the first pressure change rate; Step S25: Calculate the standard deviation of the difference sequence of adjacent two flow rate sampling points within the candidate glitch data segment to obtain the standard deviation of adjacent sampling point differences; Step S26: Calculate the ratio of the difference between the maximum flow rate and the minimum flow rate within the candidate glitch data segment to the average flow rate within the candidate glitch data segment to obtain the ratio of the range within the data segment.

[0010] By calculating the absolute value of the flow rate change rate, the absolute value of the pump speed change rate, and the absolute value of the pressure change rate, the present invention can accurately capture the instantaneous change characteristics of the flow rate, pump speed, and pressure. At the same time, a lag time compensation coefficient is introduced to adapt to the dynamic change characteristics under different working conditions. The calculation of the standard deviation of the difference between adjacent sampling points reflects the volatility of the flow rate data, further enhancing the ability to identify glitch signals. By introducing the ratio of the range, the significance of the flow rate change is highlighted, effectively distinguishing normal regulation from glitch signals.

[0011] Preferably, determining the pump speed change lag time compensation coefficient according to the calculation times identifier of the current pump speed change rate in step S23 includes: If the calculation times identifier of the current pump speed change rate is the first calculation identifier, set 0.5 seconds as the pump speed change lag time compensation coefficient; If the calculation times identifier of the current pump speed change rate is not the first calculation identifier, obtain the absolute value of the previous pump speed change rate; If the absolute value of the previous pump speed change rate is less than or equal to the preset first pump speed change rate threshold, set 0.5 seconds as the pump speed change lag time compensation coefficient; If the absolute value of the previous pump speed change rate is greater than the preset first pump speed change rate threshold and the absolute value of the previous pump speed change rate is less than or equal to the preset second pump speed change rate threshold, set 1 second as the pump speed change lag time compensation coefficient; If the absolute value of the previous pump speed change rate is greater than the preset second pump speed change rate threshold and the absolute value of the previous pump speed change rate is less than or equal to the preset third pump speed change rate threshold, set 1.5 seconds as the pump speed change lag time compensation coefficient; If the absolute value of the previous pump speed change rate is greater than the preset third pump speed change rate threshold, set 2 seconds as the pump speed change lag time compensation coefficient, where the preset first pump speed change rate threshold is less than the preset second pump speed change rate threshold, and the preset second pump speed change rate threshold is less than the preset third pump speed change rate threshold.

[0012] By dynamically adjusting the lag time compensation coefficient according to the calculation times identifier and historical value of the pump speed change rate, the present invention can effectively reflect the change characteristics of the pump group operation state, avoiding misjudgment or missed judgment caused by a fixed lag time. This can adapt to the pump speed change characteristics under different working conditions, especially during the start-stop or variable frequency speed regulation process of the pump group, and can more accurately capture the true characteristics of the flow rate change, thereby improving the reliability of glitch data detection. By hierarchically setting the lag time compensation coefficient, the adaptability of the system to complex working conditions is further enhanced, ensuring the stability and consistency of the detection results.

[0013] Preferably, determining the pressure change lag time compensation coefficient according to the calculation times identifier of the current pressure change rate in step S24 includes: If the calculation times identifier of the current pressure change rate is the first calculation identifier, set 0.2 seconds as the pressure change lag time compensation coefficient; If the calculation times identifier of the current pressure change rate is not the first calculation identifier, obtain the absolute value of the previous pressure change rate; If the absolute value of the previous pressure change rate is less than or equal to the preset first pressure change rate threshold, set 0.2 seconds as the pressure change lag time compensation coefficient; If the absolute value of the previous pressure change rate is greater than the preset first pressure change rate threshold and less than or equal to the preset second pressure change rate threshold, set 0.5 seconds as the pressure change lag time compensation coefficient; If the absolute value of the previous pressure change rate is greater than the preset second pressure change rate threshold and less than or equal to the preset third pressure change rate threshold, set 0.8 seconds as the pressure change lag time compensation coefficient; If the absolute value of the previous pressure change rate is greater than the preset third pressure change rate threshold, set 1 second as the pressure change lag time compensation coefficient, where the preset first pressure change rate threshold is less than the preset second pressure change rate threshold, and the preset second pressure change rate threshold is less than the preset third pressure change rate threshold.

[0014] The present invention can effectively reflect the dynamic characteristics of the pipeline network pressure by dynamically adjusting the lag time compensation coefficient according to the calculation times identifier of the pressure change rate and the historical value, and avoid misjudgment or missed judgment caused by a fixed lag time. This can adapt to the pressure change characteristics under different working conditions. Especially during the pipeline network pressure fluctuation or transient change process, it can more accurately capture the true characteristics of the pressure change, thereby improving the reliability of the burr data detection. By hierarchically setting the lag time compensation coefficient, the adaptability of the system to complex working conditions is further enhanced, ensuring the stability and consistency of the detection results.

[0015] Preferably, step S3 includes the following steps: Step S31: Obtain the current pump group operation state parameters; identify the dynamic operation mode of the pump group according to the current pump group operation state parameters and the preset pump group operation mode determination rule, and generate the current pump group operation mode, where the current pump group operation mode is any one of the constant pressure mode, the variable frequency voltage regulation mode, and the start-stop transition mode; Step S32: If the current pump group operation mode is the constant pressure mode, set 0.5 seconds as the pump group lag time compensation coefficient; Step S33: If the current pump group operation mode is the variable frequency voltage regulation mode, obtain the variable frequency voltage regulation rate of the pump group, and calculate the pump group lag time compensation coefficient according to the variable frequency voltage regulation rate of the pump group. The specific calculation formula is as follows: ; Among them, is the pump group lag time compensation coefficient, is the variable frequency and voltage regulation rate of the pump group; Step S34: If the current pump group operation mode is the start-stop transition mode, set 1.2 seconds as the pump group lag time compensation coefficient; Step S35: Determine the flow rate change rate threshold, pump speed change rate threshold, pressure change rate threshold, standard deviation multiple threshold, and range ratio threshold according to the pump group lag time compensation coefficient to obtain five types of dynamic parameter thresholds. Among them, the specific calculation formulas for the flow rate change rate threshold, pump speed change rate threshold, pressure change rate threshold, standard deviation multiple threshold, and range ratio threshold are as follows: I = 0.15×(1 + 0.1× ); B = 0.05×(1 + 0.2×f); Y = 3×(1 + 0.15× ); C = 1.5×(1 + 0.05×p); J = 0.2×(1 - 0.08× ); Among them, I is the flow rate change rate threshold, B is the pump speed change rate threshold, Y is the pressure change rate threshold, C is the standard deviation multiple threshold, J is the range ratio threshold, f is the peak value of the pump speed fluctuation amplitude, and p is the standard deviation of the pipe network pressure fluctuation; Step S36: Collect the internal temperature distribution thermal map of the pump group; identify the local hot spot area of the pump group winding based on the internal temperature distribution thermal map of the pump group to generate the pump group winding temperature distribution map. Among them, the pump group winding temperature distribution map includes temperature gradient distribution data and multiple regions; Step S37: According to the temperature gradient distribution data and the hot spot coordinate set, judge whether there is an area in the pump group winding temperature distribution map where the area temperature is greater than the average temperature of the adjacent area + 3°C. If so, mark the corresponding area as the pump group winding hot spot area; Step S38: Calculate the proportion of the hot spot area in the pump group winding temperature distribution map; calculate the threshold weight coefficient according to the proportion of the hot spot area. Among them, the specific calculation formula is as follows: ; Among them, is the threshold weight coefficient, and R is the proportion of the hot spot area; Step S39: Perform zoning correction on the five types of dynamic parameter thresholds according to the threshold weight coefficient to obtain the corrected five types of dynamic parameter thresholds; Step S310: Generate a dynamic discrimination threshold group based on the corrected five - type dynamic parameter thresholds. Among them, if the hot - spot area of the same pump - group winding is continuously triggered more than or equal to 3 times, then each threshold in the five - type dynamic parameter thresholds is reduced by 20%.

[0016] By identifying the operation mode of the pump group and setting the lag - time compensation coefficient accordingly, the present invention can effectively adapt to the dynamic characteristics under different working conditions and ensure the rationality of the threshold. By combining the frequency - conversion voltage - regulating rate of the pump group and the analysis of the temperature - distribution heat map, the calculation of the lag - time compensation coefficient and the threshold - weight coefficient is further optimized, enhancing the sensitivity to changes in the equipment state. By comprehensively considering factors such as the pump - speed fluctuation amplitude, the pipeline - pressure fluctuation, and the proportion of the hot - spot area, the threshold is corrected in zones, ensuring the stability and consistency of the detection results.

[0017] Preferably, step S5 includes the following steps: Step S51: Conduct a leading verification of the pump - speed change for the suspected glitch data segment to obtain a leading - verification result of the pump - speed change; Step S52: Conduct a leading verification of the pressure change for the suspected glitch data segment to obtain a leading - verification result of the pressure change; Step S53: Conduct a verification of the pump - group start - stop state signal for the suspected glitch data segment to obtain a verification result of the pump - group start - stop state signal; Step S54: Conduct a verification of historical similar working conditions for the suspected glitch data segment to obtain a verification result of historical similar working conditions; Step S55: Based on the verification results of steps S51 to S54, output the candidate glitch data segments that pass the verification as the final glitch data.

[0018] Through the leading verification of the pump - speed change and the leading verification of the pressure change, the present invention can capture the time difference and the direction consistency between the flow - rate change and the pump - speed or pressure change, effectively distinguishing normal regulation from glitch signals. The verification of the pump - group start - stop state signal further confirms whether the suspected glitch data segment is normal regulation through the analysis of the time - coincidence degree. The verification of historical similar working conditions enhances the adaptability to complex working conditions by using the similarity matching of historical data.

[0019] Preferably, the verification method for the pump - speed change time leading the flow - rate change within 1 second to 3 seconds before the suspected glitch data segment in step S51 includes: Construct a time - stamp sequence of the pump - speed change events within 1 second to 3 seconds before the suspected glitch data segment to generate second - pump - speed time - series data; Construct a time - stamp sequence of the flow - rate change events within 1 second to 3 seconds before the suspected glitch data segment to generate second - flow - rate time - series data; Identify the events with an absolute value of the change rate greater than or equal to 0.03 in the second - pump - speed time - series data, and record the event - trigger time - stamp T1; Identify events in the second flow rate time series data where the absolute value of the rate of change is greater than or equal to 0.05, and record the event trigger timestamp T2; Calculate the time difference between T1 and T2 T.

[0020] Through the construction of an accurate timestamp sequence and the calculation of time differences, the present invention effectively verifies the leading effect of pump speed changes on flow rate changes. By identifying the timestamps of pump speed change events and flow rate change events, it is possible to accurately determine whether the pump speed change precedes the flow rate change, thus providing a strong basis for distinguishing normal regulation from glitch signals. This verification method not only improves the accuracy of glitch data detection but also enhances the understanding of the system's dynamic behavior, contributing to the optimization of the pump unit's operation control strategy.

[0021] Preferably, the present invention provides a water service flow rate glitch data detection system for performing the water service flow rate glitch data detection method described above. The water service flow rate glitch data detection system includes: A data acquisition module for synchronously collecting the first flow rate time series data, the first pump speed time series data, and the pipe network pressure time series data to generate water service flow rate related time series data; segmenting and intercepting the water service flow rate related time series data based on a preset sliding time window to generate multiple candidate glitch data segments; A parameter calculation module for calculating, for each candidate glitch data segment, the absolute value of the flow rate change rate, the absolute value of the pump speed change rate, the absolute value of the pressure change rate, the standard deviation of the differences between adjacent sampling points, and the ratio of the range within the data segment to generate five types of dynamic parameters; A threshold construction module for obtaining the current pump unit operation state parameters; constructing a dynamic discrimination threshold group according to the current pump unit operation state parameters; A data marking module for comparing the five types of dynamic parameters with the corresponding thresholds in the dynamic discrimination threshold group respectively. If at least three of the five types of dynamic parameters exceed the thresholds, the corresponding candidate glitch data segment is marked as a suspected glitch data segment; A verification and confirmation module for performing multi-level verification on the suspected glitch data segments. The multi-level verification includes pump speed change leading verification, pressure change leading verification, pump unit start-stop state signal verification, and historical similar working condition verification. If any verification fails, the suspected mark in the suspected glitch data segment is removed, and the candidate glitch data segment that passes the verification is output as the final glitch data.

[0022] In the present invention, the data acquisition module can synchronously acquire the time-series data of flow rate, pump speed, and pipe network pressure, and segment and intercept them based on a sliding time window to generate multiple candidate glitch data segments. This ensures the integrity and continuity of the data. The parameter calculation module calculates five types of dynamic parameters for each candidate glitch data segment, including the absolute value of the flow rate change rate, the absolute value of the pump speed change rate, the absolute value of the pressure change rate, the standard deviation of the difference between adjacent sampling points, and the ratio of the range within the data segment. These parameters reflect the characteristics of the flow rate data from different perspectives and provide a comprehensive quantitative basis for subsequent glitch data identification. The threshold construction module dynamically constructs a discriminant threshold group according to the operating state parameters of the pump group. By considering factors such as the operating mode of the pump group, the frequency conversion voltage regulation rate, and the temperature distribution, the threshold is dynamically adjusted to ensure that the threshold can adapt to the flow rate change characteristics under different working conditions, improving the accuracy and adaptability of detection. The data marking module can quickly and accurately identify suspected glitch data segments by comparing the dynamic parameters with the dynamic discriminant threshold group. This marking method based on multi-parameter comparison effectively reduces the possibility of false positives and false negatives. The verification and confirmation module performs multi-dimensional cross-verification on the suspected glitch data segments through various verification methods such as pump speed change leading verification, pressure change leading verification, pump group start-stop state signal verification, and historical similar working condition verification. This significantly improves the reliability and stability of glitch data detection and ensures the accuracy of the finally output glitch data. In summary, through dynamic parameter calculation, dynamic threshold construction, and a multi-level verification mechanism, the present invention can adapt to the flow rate change characteristics under different working conditions, especially in complex working conditions such as pump group start-stop switching and variable frequency speed regulation, and still maintain a high-precision glitch data detection ability.

[0023] Preferably, the present invention also provides a computer-readable medium storing a program that can be loaded and executed by a processor to perform the water flow glitch data detection method as described above. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Other features, objectives, and advantages of the present invention will become more apparent by reading the following detailed description with reference to the accompanying drawings: Figure 1 The step flow diagram of the water flow glitch data detection method according to an embodiment is shown.

[0025] Figure 2 The detailed step flow diagram of step S2 according to an embodiment is shown. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0026] The technical method of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art within the scope of the present invention without creative efforts belong to the scope of protection of the present invention.

[0027] In addition, the accompanying drawings are only schematic illustrations of the present invention and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and thus repeated descriptions thereof will be omitted. Some of the block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. The functional entities can be implemented in software form, or in one or more hardware modules or integrated circuits, or in different networks and / or processor methods and / or microcontroller methods.

[0028] It should be understood that although terms such as "first" and "second" may be used herein to describe various units, these units should not be limited by these terms. These terms are only used to distinguish one unit from another. For example, without departing from the scope of the exemplary embodiments, the first unit can be called the second unit, and similarly the second unit can be called the first unit. The term "and / or" used herein includes any and all combinations of one or more of the listed associated items.

[0029] To achieve the above object, please refer to Figures 1 to 2 , the present invention provides a method for detecting water flow spike data, including the following steps: Step S1: Synchronously collect the first flow time series data, the first pump speed time series data, and the pipe network pressure time series data to generate water flow-related time series data; segment and intercept the water flow-related time series data based on a preset sliding time window to generate a plurality of candidate spike data segments; Step S2: For each candidate spike data segment, calculate its absolute value of flow rate change rate, absolute value of first pump speed change rate, absolute value of first pressure change rate, standard deviation of differences between adjacent sampling points, and ratio of range within the data segment to generate five types of dynamic parameters; Step S3: Obtain the current pump group operating state parameters; construct a dynamic discrimination threshold group according to the current pump group operating state parameters; Step S4: Compare the five types of dynamic parameters with the corresponding thresholds in the dynamic discrimination threshold group respectively. If at least three of the five types of dynamic parameters exceed the thresholds, mark the corresponding candidate spike data segment as a suspected spike data segment; Step S5: Perform multi-level verification on the suspected spike data segment. Among them, the multi-level verification includes leading verification of pump speed change, leading verification of pressure change, verification of pump group start-stop state signal, and verification of historical similar working conditions. If any verification fails, remove the suspected mark in the suspected spike data segment, and output the candidate spike data segment that passes the verification as the final spike data.

[0030] In this embodiment, first, the Pandas library is used to read the first flow time series data, the first pump speed time series data, and the pipe network pressure time series data stored in the CSV file. These data are collected and stored in real time by sensors installed in the water supply pipe network. The time stamp format of the data is YYYY - MM - DD HH:MM:SS.mmm, and the sampling frequency is 100Hz. The rolling function of Pandas is used to set the sliding time window length to 2 seconds (i.e., 200 sampling points) and the sliding step to 0.4 seconds (i.e., 40 sampling points) to generate multiple candidate glitch data segments. The NumPy library is used to calculate five types of dynamic parameters: the absolute value of the flow rate change rate, the absolute value of the pump speed change rate, the absolute value of the pressure change rate, the standard deviation of the difference between adjacent sampling points, and the ratio of the range within the data segment. The operating state parameters of the pump group are obtained through the SCADA system, and according to the operating mode of the pump group (constant pressure mode, variable frequency voltage regulation mode, or start - stop transition mode), a preset formula is used to calculate the dynamic discrimination threshold group. The five types of dynamic parameters of each candidate glitch data segment are compared with the dynamic discrimination threshold group. If at least three types of parameters exceed the threshold, the candidate glitch data segment is marked as a suspected glitch data segment. Multistage verification is carried out: Pandas is used to extract the pump speed change rate and the flow rate change rate within 1 second to 3 seconds before the suspected glitch data segment to verify whether the pump speed change leads the flow rate change; the pressure change rate and the flow rate change rate within 0.2 seconds to 0.8 seconds before the suspected glitch data segment are extracted to verify whether the pressure change is in the same direction as the flow rate change; the trigger time of the pump group start - stop signal is obtained to verify whether the time interval of the suspected glitch data segment coincides with the start - stop signal; historical data similar to the current working condition is queried from the Elasticsearch database to verify whether the same type of fluctuations are marked as normal regulation. Finally, according to the multistage verification results, Pandas is used to mark the candidate glitch data segments that pass the verification as the final glitch data.

[0031] Especially importantly, the acquisition of the pipe network pressure time series data includes: The original pipe network pressure time series data is collected in real time by a distributed optical fiber sensor, and the original pipe network pressure time series data is amplified synchronously in multiple channels to generate pipe network pressure high - frequency vibration waveform data; The pipe network pressure high - frequency vibration waveform data is decomposed by modal decomposition to obtain the pipe network pressure vibration modal components; The spectral characteristics of the pipe network pressure vibration modal components are extracted to obtain the pipe network pressure vibration frequency distribution characteristics; Based on the pipe network pressure vibration frequency distribution characteristics, the high - frequency noise interference interval is identified, where the high - frequency noise interference interval is the characteristic frequency interval of pipeline cavitation and water hammer effect; The data segments in the original pipe network pressure time series data that belong to the high - frequency noise interference interval are filtered to generate the pipe network pressure time series data.

[0032] In this embodiment, in the urban water supply network, a distributed optical fiber sensor is used to collect the network pressure data in real time. The optical fiber sensor is laid along the network and covers key nodes (such as the outlet of the pumping station and the inlet of the user end). The sensor collects the pressure signal at a sampling frequency of 100 Hz and transmits it to the data acquisition module through the optical fiber. The acquisition module uses a multi-channel signal amplifier to amplify the signal synchronously, and the amplification factor is set to 10 times to ensure the clarity of the signal. The amplified signal is stored as the original network pressure time series data. The Hilbert-Huang Transform (HHT) toolbox of MATLAB is used to perform modal decomposition on the original network pressure time series data. The specific operations are as follows: First, import the original data into MATLAB, select Empirical Mode Decomposition (EMD) to decompose the data, and generate multiple Intrinsic Mode Functions (IMFs). By setting the stopping condition of EMD (such as the ratio of the number of extreme points to the length of the remaining signal is less than 0.5). The IMF components obtained after decomposition contain the vibration mode characteristics of different frequencies. The numpy.fft library of Python is used to perform a Fast Fourier Transform (FFT) on the modal components. The specific operations are as follows: Import the decomposed IMF components into Python, set the sampling frequency of the FFT to 100 Hz, the window length to 1024 sampling points, and the overlap rate to 50%. Calculate the spectral amplitude and phase information through the FFT, and extract the spectral features. The spectral features include the main frequency, sub-frequency, and spectral energy distribution, which are used to identify the characteristic frequencies of the network pressure vibration. Refer to the characteristic frequency ranges of water hammer effect and pipeline cavitation in ISO 4856 standard (usually 100 Hz to 5 kHz), and combine the analysis results of the spectral features to identify the high-frequency noise interference intervals. The specific operations are as follows: In the spectrogram, set the frequency range to 100 Hz to 5 kHz, and mark the intervals where the spectral energy is significantly higher than the background noise. These intervals usually correspond to the characteristic frequencies of the water hammer effect (such as pressure mutation) or pipeline cavitation (such as bubble formation and rupture). Determine the high-frequency noise interference intervals by comparing with the standard frequency range. The signal processing toolbox of MATLAB is used to perform band-stop filtering on the original network pressure time series data. The specific operations are as follows: According to the high-frequency noise interference intervals identified in step 4 (such as 100 Hz to 5 kHz), design a band-stop filter, the filter type is IIR (Infinite Impulse Response), the cut-off frequencies are set to 100 Hz and 5 kHz, and the filter order is 4th order. Process the original data through the filter to remove the data segments in the high-frequency noise interference intervals and generate the filtered network pressure time series data.

[0033] Preferably, the step S1 of segmenting and intercepting the water service flow-related time series data based on a preset sliding time window includes: Segment the time series data related to water flow based on a preset sliding time window to generate multiple candidate glitch data segments. The length of the preset sliding time window is from 0.5 seconds to 5 seconds, and the sliding step is from 1 / 5 to 1 / 2 of the window length. Each candidate glitch data segment needs to meet the following conditions: The number of flow sampling points within the candidate glitch data segment is greater than or equal to 50. The candidate glitch data segment covers at least one fluctuation interval where the flow change rate exceeds ±0.1.

[0034] In this embodiment, the Pandas library of Python is used to segment the time series data related to water flow. The specific operations are as follows: First, import the time series data related to water flow into a Pandas DataFrame. Set the sliding time window length to 2 seconds (i.e., 200 sampling points), and the sliding step to 1 / 5 of the window length (i.e., 0.4 seconds, 40 sampling points). Implement the sliding window operation through the rolling function of Pandas to generate multiple candidate glitch data segments. The data of each window is stored as an independent array. Use the SciPy library of Python to calculate the flow change rate of each candidate glitch data segment. The specific operations are as follows: First, perform a first-order difference on the flow values of each data segment to calculate the flow change amount between adjacent sampling points. Then, calculate the absolute value of the change amount and check whether there is a fluctuation interval exceeding ±0.1. For example, in a certain data segment, the flow changes from 50 L / s to 55 L / s, and the change rate is 0.1 (5 / 50), which meets the condition. This data segment is marked as a valid candidate glitch data segment. If there is no such fluctuation interval in the data segment, it is excluded.

[0035] Preferably, step S2 includes the following steps: Step S21: Calculate the data segment duration of the candidate glitch data segment. Step S22: Extract the maximum flow value and the minimum flow value within the candidate glitch data segment, and record the quotient of the absolute value of the difference between the maximum flow value and the minimum flow value divided by the data segment duration as the absolute value of the flow change rate. Step S23: Obtain the calculation times identifier of the current pump speed change rate; determine the pump speed change lag time compensation coefficient according to the calculation times identifier of the current pump speed change rate; extract the maximum pump speed and the minimum pump speed within the candidate glitch data segment, and record the quotient of the absolute value of the difference between the maximum pump speed and the minimum pump speed divided by the pump speed change lag time compensation coefficient as the absolute value of the first pump speed change rate. Step S24: Obtain the calculation times identifier of the current pressure change rate; determine the pressure change lag time compensation coefficient according to the calculation times identifier of the current pressure change rate; extract the maximum pipeline network pressure and the minimum pipeline network pressure within the candidate burr data segment, and record the quotient of dividing the absolute value of the difference between the maximum pipeline network pressure and the minimum pipeline network pressure by the pressure change lag time compensation coefficient as the absolute value of the first pressure change rate; Step S25: Calculate the standard deviation of the difference sequence between two adjacent flow sampling points within the candidate burr data segment to obtain the standard deviation of the difference between adjacent sampling points; Step S26: Divide the difference between the maximum flow rate and the minimum flow rate within the candidate burr data segment by the average flow rate within the candidate burr data segment to obtain the ratio of the range within the data segment.

[0036] In this embodiment, the datetime library of Python is used to calculate the duration of the candidate glitch data segment. The specific operations are as follows: First, obtain the start and end timestamps of the candidate glitch data segment. For example, the start timestamp is 2023-10-01 12:00:00.000, and the end timestamp is 2023-10-01 12:00:02.000. Convert the string timestamp to a datetime object through the datetime.strptime function, and then calculate the time difference between the two to obtain a data segment duration of 2 seconds. Use the NumPy library of Python to extract the maximum and minimum flow values within the candidate glitch data segment. The specific operations are as follows: Import the flow value array of the candidate glitch data segment into NumPy, and use the numpy.max and numpy.min functions to find the maximum and minimum values respectively. For example, the maximum flow value is 55 L / s, the minimum flow value is 50 L / s, and the difference is 5 L / s. Divide the absolute value of this difference by the duration (2 seconds) calculated in step S21 to obtain an absolute flow change rate of 2.5 L / s². Use the Pandas library of Python to process the pump speed data. The specific operations are as follows: First, check the calculation times identifier of the current pump speed change rate. If it is the first calculation, set the pump speed change lag time compensation coefficient to 0.5 seconds. Extract the maximum and minimum pump speeds within the candidate glitch data segment. For example, the maximum value is 1200 rpm, the minimum value is 1000 rpm, and the difference is 200 rpm. Divide the absolute value of this difference by the lag time compensation coefficient (0.5 seconds) to obtain an absolute first pump speed change rate of 400 rpm / s. Use the NumPy library of Python to process the pipeline network pressure data. The specific operations are as follows: First, check the calculation times identifier of the current pressure change rate. If it is the first calculation, set the pressure change lag time compensation coefficient to 0.2 seconds. Extract the maximum and minimum pipeline network pressures within the candidate glitch data segment. For example, the maximum value is 500 kPa, the minimum value is 450 kPa, and the difference is 50 kPa. Divide the absolute value of this difference by the lag time compensation coefficient (0.2 seconds) to obtain an absolute first pressure change rate of 250 kPa / s. Use the NumPy library of Python to calculate the standard deviation of the differences between adjacent flow sampling points. The specific operations are as follows: Perform a first-order difference on the flow value array of the candidate glitch data segment to obtain a difference sequence of adjacent sampling points. For example, the difference sequence is [1, -2, 3, -1, 2...]. Use the numpy.std function to calculate the standard deviation of this sequence to obtain a standard deviation of differences between adjacent sampling points of 1.5 L / s. This standard deviation reflects the volatility of the flow data. Use the NumPy library of Python to calculate the range ratio within the data segment. The specific operations are as follows: Extract the maximum and minimum flow values within the candidate glitch data segment. For example, the maximum value is 55 L / s, the minimum value is 50 L / s, and the difference is 5 L / s.Calculate the average flow rate value within this data segment. For example, the average flow rate is 52.5 L / s. Divide the difference by the average flow rate to obtain a ratio of the range to the average flow rate of 5 / 52.5 ≈ 0.095 (about 9.5%).

[0037] Preferably, determining the pump speed change lag time compensation coefficient according to the calculation times identifier of the current pump speed change rate in step S23 includes: If the calculation times identifier of the current pump speed change rate is the first calculation identifier, set 0.5 seconds as the pump speed change lag time compensation coefficient; If the calculation times identifier of the current pump speed change rate is not the first calculation identifier, obtain the absolute value of the previous pump speed change rate; If the absolute value of the previous pump speed change rate is less than or equal to a preset first pump speed change rate threshold, set 0.5 seconds as the pump speed change lag time compensation coefficient; If the absolute value of the previous pump speed change rate is greater than the preset first pump speed change rate threshold and less than or equal to a preset second pump speed change rate threshold, set 1 second as the pump speed change lag time compensation coefficient; If the absolute value of the previous pump speed change rate is greater than the preset second pump speed change rate threshold and less than or equal to a preset third pump speed change rate threshold, set 1.5 seconds as the pump speed change lag time compensation coefficient; If the absolute value of the previous pump speed change rate is greater than the preset third pump speed change rate threshold, set 2 seconds as the pump speed change lag time compensation coefficient, where the preset first pump speed change rate threshold is less than the preset second pump speed change rate threshold, and the preset second pump speed change rate threshold is less than the preset third pump speed change rate threshold.

[0038] In this embodiment, the calculation times identifier of the current pump speed change rate is checked. If it is the first calculation, the pump speed change lag time compensation coefficient is set to 0.5 seconds. For example, in the first data segment after the system starts, since there is no historical data, the lag time compensation coefficient is directly set to 0.5 seconds. This coefficient is used for the calculation of the pump speed change rate. If the calculation times identifier of the current pump speed change rate is not the first calculation identifier, the absolute value of the previous pump speed change rate is obtained. Assume that the preset first pump speed change rate threshold is 100 rpm / s. If the absolute value of the previous pump speed change rate is 80 rpm / s (less than the first pump speed change rate threshold), the pump speed change lag time compensation coefficient is set to 0.5 seconds. For example, in the second data segment of continuous monitoring, the previous change rate is 80 rpm / s, so the lag time compensation coefficient is still set to 0.5 seconds. If the absolute value of the previous pump speed change rate is between the first pump speed change rate threshold and the second pump speed change rate threshold, assume that the preset first pump speed change rate threshold is 100 rpm / s and the second pump speed change rate threshold is 200 rpm / s. If the absolute value of the previous pump speed change rate is 150 rpm / s (between the first pump speed change rate threshold and the second pump speed change rate threshold), the pump speed change lag time compensation coefficient is set to 1 second. For example, in the third data segment of continuous monitoring, the previous change rate is 150 rpm / s, so the lag time compensation coefficient is set to 1 second. If the absolute value of the previous pump speed change rate is between the second pump speed change rate threshold and the third pump speed change rate threshold, assume that the preset second pump speed change rate threshold is 200 rpm / s and the third threshold is 300 rpm / s. If the absolute value of the previous pump speed change rate is 250 rpm / s (between the second pump speed change rate threshold and the third pump speed change rate threshold), the pump speed change lag time compensation coefficient is set to 1.5 seconds. For example, in the fourth data segment of continuous monitoring, the previous change rate is 250 rpm / s, so the lag time compensation coefficient is set to 1.5 seconds. If the absolute value of the previous pump speed change rate is greater than the third pump speed change rate threshold, assume that the preset third pump speed change rate threshold is 300 rpm / s. If the absolute value of the previous pump speed change rate is 350 rpm / s (greater than the third pump speed change rate threshold), the pump speed change lag time compensation coefficient is set to 2 seconds. For example, in the fifth data segment of continuous monitoring, the previous change rate is 350 rpm / s, so the lag time compensation coefficient is set to 2 seconds. This higher lag time compensation coefficient can better adapt to the drastic change of the pump speed and avoid misjudgment.

[0039] Preferably, determining the pressure change lag time compensation coefficient according to the calculation times identifier of the current pressure change rate in step S24 includes: If the calculation times identifier of the current pressure change rate is the first calculation identifier, 0.2 seconds is set as the pressure change lag time compensation coefficient; If the calculation times identifier of the current pressure change rate is not the first calculation identifier, obtain the absolute value of the previous pressure change rate; If the absolute value of the previous pressure change rate is less than or equal to a preset first pressure change rate threshold, set 0.2 seconds as the pressure change lag time compensation coefficient; If the absolute value of the previous pressure change rate is greater than the preset first pressure change rate threshold and less than or equal to the preset second pressure change rate threshold, set 0.5 seconds as the pressure change lag time compensation coefficient; If the absolute value of the previous pressure change rate is greater than the preset second pressure change rate threshold and less than or equal to the preset third pressure change rate threshold, set 0.8 seconds as the pressure change lag time compensation coefficient; If the absolute value of the previous pressure change rate is greater than the preset third pressure change rate threshold, set 1 second as the pressure change lag time compensation coefficient, where the preset first pressure change rate threshold is less than the preset second pressure change rate threshold, and the preset second pressure change rate threshold is less than the preset third pressure change rate threshold.

[0040] In this embodiment, the calculation times identifier of the current pressure change rate is checked. If it is the first calculation, the pressure change lag time compensation coefficient is set to 0.2 seconds. For example, in the first data segment after the system starts, since there is no historical data, the lag time compensation coefficient is directly set to 0.2 seconds. This coefficient is used for the calculation of the pressure change rate. If the calculation times identifier of the current pressure change rate is not the first calculation identifier, the absolute value of the previous pressure change rate is obtained; assume that the preset first pressure change rate threshold is 50 kPa / s. If the absolute value of the previous pressure change rate is 30 kPa / s (less than the first pressure change rate threshold), the pressure change lag time compensation coefficient is set to 0.2 seconds. For example, in the second data segment of continuous monitoring, the previous change rate is 30 kPa / s, so the lag time compensation coefficient is still set to 0.2 seconds. Assume that the second pressure change rate threshold is 100 kPa / s. If the absolute value of the previous pressure change rate is 75 kPa / s, the pressure change lag time compensation coefficient is set to 0.5 seconds. For example, in the third data segment of continuous monitoring, the previous change rate is 75 kPa / s, so the lag time compensation coefficient is set to 0.5 seconds. Assume that the preset third pressure change rate threshold is 150 kPa / s. If the absolute value of the previous pressure change rate is 120 kPa / s, the pressure change lag time compensation coefficient is set to 0.8 seconds. For example, in the fourth data segment of continuous monitoring, the previous change rate is 120 kPa / s, so the lag time compensation coefficient is set to 0.8 seconds. If the absolute value of the previous pressure change rate is 180 kPa / s, the pressure change lag time compensation coefficient is set to 1 second. For example, in the fifth data segment of continuous monitoring, the previous change rate is 180 kPa / s, so the lag time compensation coefficient is set to 1 second.

[0041] Preferably, step S3 includes the following steps: Step S31: Obtain the current pump group operation state parameters; identify the dynamic operation mode of the pump group according to the current pump group operation state parameters and the preset pump group operation mode determination rules, and generate the current pump group operation mode, where the current pump group operation mode is any one of the constant pressure mode, the variable frequency voltage regulation mode, and the start-stop transition mode; Step S32: If the current pump group operation mode is the constant pressure mode, set 0.5 seconds as the pump group lag time compensation coefficient; Step S33: If the current pump group operation mode is the variable frequency voltage regulation mode, obtain the variable frequency voltage regulation rate of the pump group, and calculate the pump group lag time compensation coefficient according to the variable frequency voltage regulation rate of the pump group. The specific calculation formula is as follows: ; Wherein, is the pump group lag time compensation coefficient, is the variable frequency voltage regulation rate of the pump group; Step S34: If the current pump group operation mode is the start-stop transition mode, set 1.2 seconds as the pump group lag time compensation coefficient; Step S35: Determine the flow rate change rate threshold, pump speed change rate threshold, pressure change rate threshold, standard deviation multiple threshold, and range ratio threshold according to the pump group lag time compensation coefficient to obtain five types of dynamic parameter thresholds. The specific calculation formulas for the flow rate change rate threshold, pump speed change rate threshold, pressure change rate threshold, standard deviation multiple threshold, and range ratio threshold are as follows: I = 0.15×(1 + 0.1× ); B = 0.05×(1 + 0.2×f); Y = 3×(1 + 0.15× ); C = 1.5×(1 + 0.05×p); J = 0.2×(1 - 0.08× ); where I is the flow rate change rate threshold, B is the pump speed change rate threshold, Y is the pressure change rate threshold, C is the standard deviation multiple threshold, J is the range ratio threshold, f is the peak value of the pump speed fluctuation amplitude, and p is the standard deviation of the pipeline network pressure fluctuation; Step S36: Collect the internal temperature distribution heat map of the pump group; identify the local hot spot area of the pump group winding based on the internal temperature distribution heat map of the pump group to generate the pump group winding temperature distribution map, where the pump group winding temperature distribution map contains temperature gradient distribution data and multiple regions; Step S37: According to the temperature gradient distribution data and the hot spot coordinate set, determine whether there is an area in the pump group winding temperature distribution map where the temperature is greater than the average temperature of the adjacent area + 3°C. If so, mark the corresponding area as the pump group winding hot spot area; Step S38: Calculate the proportion of the hot spot area in the pump group winding temperature distribution map; calculate the threshold weight coefficient according to the proportion of the hot spot area, where the specific calculation formula is as follows: ; where is the threshold weight coefficient, and R is the proportion of the hot spot area; Step S39: Perform zonal correction on the five types of dynamic parameter thresholds according to the threshold weight coefficient to obtain the corrected five types of dynamic parameter thresholds; Step S310: Generate a dynamic discrimination threshold group based on the corrected five types of dynamic parameter thresholds. If the same pump group winding hot spot area is continuously triggered greater than or equal to 3 times, reduce each threshold in the five types of dynamic parameter thresholds by 20%.

[0042] In this embodiment, a Siemens SCADA (Supervisory Control and Data Acquisition) system can be used to obtain the operating state parameters of the pump group, including pump speed, pipeline network pressure, and flow rate. According to the preset determination rules, when the deviation between the output pressure of the pump group and the set pressure is less than ±0.5 bar, it is determined to be in the constant pressure mode; when the frequency converter of the pump group is operating in the voltage regulation state, it is determined to be in the variable frequency voltage regulation mode; when the pump group is in the transition stage of starting or stopping, it is determined to be in the start-stop transition mode. For example, if it is monitored that the output pressure of the pump group is stable at 4.0 bar (the set pressure is 4.0 bar), then the current operating mode is the constant pressure mode. When it is determined that the pump group is operating in the constant pressure mode, the pump group lag time compensation coefficient is directly set to 0.5 seconds. When it is determined that the pump group is operating in the variable frequency voltage regulation mode, the voltage regulation rate data provided by an ABB frequency converter is used. Assume that the current variable frequency voltage regulation rate is d = 2 Hz / s. According to the formula = 0.8 + 0.05×d, the calculated result is = 0.8 + 0.05×2 = 0.9 seconds. When it is determined that the pump group is in the start-stop transition mode, the pump group lag time compensation coefficient is directly set to 1.2 seconds. For example, within the first 10 seconds after the pump group starts, since the flow rate and pressure change violently, using this coefficient can better capture the characteristics of the flow rate change. According to the pump group lag time compensation coefficient = 0.9, each dynamic threshold is calculated. For example: The flow rate change rate threshold I = 0.15×(1 + 0.1×0.9) = 0.15×1.09 = 0.1635 L / s²; The pump speed change rate threshold B = 0.05×(1 + 0.2×f). Assume that the peak value of the pump speed fluctuation amplitude f = 3, then B = 0.05×1.6 = 0.08 rpm / s; The pressure change rate threshold Y = 3×(1 + 0.15×0.9) = 3×1.135 = 3.405 kPa / s; The standard deviation multiple threshold C = 1.5×(1 + 0.05×p). Assume that the standard deviation of the pipeline network pressure fluctuation p = 2, then C = 1.5×1.1 = 1.65; The range ratio threshold J = 0.2×(1 - 0.08×0.9) = 0.2×0.928 = 0.1856.

[0043] Use a FLIR thermal imaging camera to collect the thermal map of the temperature distribution inside the pump unit. Import the thermal map into MATLAB and analyze it using the image processing toolbox. The thermal map shows the temperature gradient distribution of the pump unit windings, containing temperature data for multiple regions. For example, the temperature of a certain region is 85 °C, and the average temperature of the adjacent region is 80 °C. In MATLAB, according to the temperature gradient distribution data and the set of hot spot coordinates, check whether there is a situation where the regional temperature exceeds the average temperature of the adjacent region + 3 °C. For example, if the temperature of a certain region is 85 °C and the average temperature of the adjacent region is 80 °C, then 85 °C > 80 °C + 3 °C, and this region is marked as a hot spot area. In MATLAB, calculate the area proportion of the hot spot area. For example, if the area of the hot spot area is 12 square centimeters and the total winding area is 100 square centimeters, then the area proportion R = 12%. According to the formula = 1.2 - 0.02×12 = 1.2 - 0.24 = 0.96, the threshold weight coefficient is obtained as 0.96. In the SCADA system, according to the threshold weight coefficient 0.96, correct the dynamic threshold. For example: correct the flow rate change rate threshold I' = 0.1635×0.96 ≈ 0.157 L / s²; correct the pump speed change rate threshold B' = 0.08×0.96 ≈ 0.0768 rpm / s; correct the pressure change rate threshold Y' = 3.405×0.96 ≈ 3.27 kPa / s; correct the standard deviation multiple threshold C' = 1.65×0.96 ≈ 1.58; correct the range proportion threshold J' = 0.1856×0.96 ≈ 0.178. In the SCADA system, combine the corrected dynamic thresholds into a dynamic discrimination threshold group. If the hot spot area of a certain pump unit winding is continuously triggered 3 times, then reduce all the corrected thresholds by 20%. For example, correct the flow rate change rate threshold I' = 0.157×0.8 ≈ 0.1256 L / s². This dynamic discrimination threshold group will be used for the next burr data detection.

[0044] Preferably, step S5 includes the following steps: Step S51: Conduct a leading verification of pump speed change for the suspected burr data segment to obtain the leading verification result of pump speed change; Among them, the leading verification of pump speed change is specifically: extract the absolute value of the second pump speed change rate, the pump speed change direction, and the first flow rate change direction within 1 second to 3 seconds before the suspected burr data segment; if the absolute value of the second pump speed change rate is greater than or equal to 0.03 and the pump speed change direction is the same as the first flow rate change direction, and the pump speed change time within 1 second to 3 seconds before the suspected burr data segment leads the flow rate change time by 0.5 - 2 seconds, then determine the corresponding suspected burr data segment as normal adjustment data and remove the suspected mark; Step S52: Conduct a leading verification of pressure change for the suspected burr data segment to obtain the leading verification result of pressure change; Among them, the pressure change pilot verification is specifically as follows: extract the absolute value of the second pressure change rate, the pipeline network pressure change direction, and the second flow rate change direction within 0.2 seconds to 0.8 seconds before the suspected burr data segment; if the absolute value of the second pressure change rate is greater than or equal to 0.02 and is consistent with the pipeline network pressure change direction and the second flow rate change direction, then determine the corresponding suspected burr data segment as normal adjustment data and remove the suspected mark; Step S53: Verify the pump group start / stop status signal for the suspected burr data segment to obtain the verification result of the pump group start / stop status signal; Among them, the pump group start / stop status signal verification is specifically as follows: obtain the pump group start / stop signal trigger time. If the time interval of the suspected burr data segment coincides with the pump group start / stop signal trigger time and the coincidence duration ratio is greater than or equal to 80%, then determine the corresponding suspected burr data segment as normal adjustment data and remove the suspected mark; Step S54: Verify the historical similar working conditions for the suspected burr data segment to obtain the verification result of the historical similar working conditions; Among them, the historical similar working conditions verification is specifically as follows: obtain the current pump speed, the current pipeline network pressure change rate, and the current ambient temperature field distribution data; match the historical working condition data set similar to the current pump speed, the current pipeline network pressure change rate, and the current ambient temperature field distribution data from the preset historical working condition database according to the preset similarity matching standard. If the proportion of the same type of fluctuations marked as normal adjustment in the historical working condition data set is greater than or equal to 70%, then determine the corresponding suspected burr data segment as normal adjustment data and remove the suspected mark; Step S55: Based on the verification results of steps S51 to S54, output the candidate burr data segments that pass the verification as the final burr data.

[0045] In this embodiment, the Pandas library of Python is used to process the pump speed and flow rate data in the candidate burr data segment. The specific operations are as follows: First, extract the absolute value of the pump speed change rate, the pump speed change direction, and the flow rate change direction within 1 second to 3 seconds before the suspected burr data segment. For example, if the time interval of the suspected burr data segment is from 2023-10-01 12:00:02.000 to 2023-10-01 12:00:04.000, then extract the pump speed data within 2023-10-01 12:00:01.000 to 2023-10-01 12:00:03.000. Calculate the absolute value of the pump speed change rate. If it is greater than or equal to 0.03, and the pump speed change direction (increase or decrease) is consistent with the flow rate change direction, and the pump speed change time leads the flow rate change time by 0.5 - 2 seconds, then determine this suspected burr data segment as normal adjustment data and remove the suspected mark. Use the Pandas library of Python to process the pressure and flow rate data in the candidate burr data segment. The specific operations are as follows: Extract the absolute value of the pressure change rate, the pressure change direction, and the flow rate change direction within 0.2 seconds to 0.8 seconds before the suspected burr data segment. For example, if the time interval of the suspected burr data segment is from 2023-10-01 12:00:02.000 to 2023-10-01 12:00:04.000, then extract the pressure data within 2023-10-01 12:00:01.200 to 2023-10-01 12:00:03.200. Calculate the absolute value of the pressure change rate. If it is greater than or equal to 0.02, and the pressure change direction is consistent with the flow rate change direction, then determine this suspected burr data segment as normal adjustment data and remove the suspected mark. Obtain the trigger time of the pump group start / stop signal. For example, the trigger time of the pump group start / stop signal is from 2023-10-01 12:00:00.000 to 2023-10-01 12:00:05.000. The time interval of the suspected burr data segment is from 2023-10-01 12:00:02.000 to 2023-10-01 12:00:04.000. Calculate the time interval overlap. If the proportion of the overlapping duration is greater than or equal to 80%, then determine this suspected burr data segment as normal adjustment data and remove the suspected mark. Use the Elasticsearch database to query historical operating condition data. The specific operations are as follows: Obtain the current pump speed of 1200 rpm, the pipeline network pressure change rate of 50 kPa / s, and the environmental temperature field distribution data (such as the temperature is 25°C). According to the preset similarity matching criteria (such as the pump speed error is less than 5%, the pressure change rate error is less than 10%, and the temperature error is less than 2°C), match the similar operating condition data sets from the historical database. If more than 70% of the same type of fluctuations in the matched historical data sets are marked as normal adjustment, then determine this suspected burr data segment as normal adjustment data and remove the suspected mark.Based on the verification results of steps S51 to S54, output the candidate glitch data segments that pass the verification as the final glitch data.

[0046] Preferably, the verification method for the pump speed change time leading the flow rate change time within the first second to the third second before the suspected glitch data segment in step S51 includes: Construct a timestamp sequence of the pump speed change events within the first second to the third second before the suspected glitch data segment to generate second pump speed time series data; Construct a timestamp sequence of the flow rate change events within the first second to the third second before the suspected glitch data segment to generate second flow rate time series data; Identify events in the second pump speed time series data with an absolute value of the change rate greater than or equal to 0.03, and record the event trigger timestamp T1; Identify events in the second flow rate time series data with an absolute value of the change rate greater than or equal to 0.05, and record the event trigger timestamp T2; Calculate the time difference between T1 and T2 T.

[0047] In this embodiment, the Pandas library of Python is used to process the pump speed data from 1 second before to 3 seconds before the suspected burr data segment. The specific operations are as follows: Determine the time interval of the suspected burr data segment as 2023-10-01 12:00:04.000 to 2023-10-01 12:00:06.000. Then, extract the pump speed data from 1 second before to 3 seconds before this interval, that is, the time interval is 2023-10-01 12:00:01.000 to 2023-10-01 12:00:03.000. Use the loc method of Pandas to select the pump speed data within this time range and generate a timestamp sequence of pump speed change events. Use the Pandas library of Python to process the flow rate data from 1 second before to 3 seconds before the suspected burr data segment. The specific operations are as follows: Determine the time interval of the suspected burr data segment as 2023-10-01 12:00:04.000 to 2023-10-01 12:00:06.000. Then, extract the flow rate data from 1 second before to 3 seconds before this interval, that is, the time interval is 2023-10-01 12:00:01.000 to 2023-10-01 12:00:03.000. Use the loc method of Pandas to select the flow rate data within this time range and generate a timestamp sequence of flow rate change events. Perform a first-order difference on the generated pump speed time series data to calculate the pump speed change rate between adjacent time points. Use boolean indexing to filter out events with an absolute value of the change rate greater than or equal to 0.03, and record the trigger timestamps T1 of these events. For example, if the pump speed change rate at a certain time point is 0.04, record this timestamp. Perform a first-order difference on the generated flow rate time series data to calculate the flow rate change rate between adjacent time points. Use boolean indexing to filter out events with an absolute value of the change rate greater than or equal to 0.05, and record the trigger timestamps T2 of these events. For example, if the flow rate change rate at a certain time point is 0.06, record this timestamp. Convert the recorded timestamps T1 and T2 to datetime objects. For example, T1 is 2023-10-01 12:00:02.000 and T2 is 2023-10-01 12:00:02.500. Calculate the time difference ∆T between the two as 0.5 seconds. This time difference is used to verify whether the pump speed change leads the flow rate change.

[0048] Especially importantly, step S5 further includes: Based on the finally verified burr data segment, construct a burr feature vector library, and the feature vectors in the burr feature vector library include the flow rate change rate gradient, the pump speed-pressure phase difference, and the range ratio fluctuation coefficient; Dynamically update the burr feature vector library through an incremental federated learning model to generate a global burr discrimination model; Collect the five types of dynamic parameters of the current operation status parameters of the pump group and the candidate glitch data segments in real time, input them into the global glitch discrimination model, and output the glitch confidence score; If the glitch confidence score is greater than or equal to 0.95, directly mark the corresponding candidate glitch data segment as the final glitch data; If 0.85 ≤ glitch confidence score < 0.95, synchronously collect the pump group vibration spectrum and electromagnetic interference intensity data. If the energy increase in the 50Hz - 200Hz frequency band of the pump group vibration spectrum is greater than or equal to 20dB or the electromagnetic interference intensity is greater than or equal to 10mV, confirm the corresponding candidate glitch data segment as the final glitch data; If the glitch confidence score is less than 0.85, remove the glitch mark and record it as a false alarm sample for the incremental training of the federated learning model.

[0049] In this embodiment, the Pandas library and NumPy library of Python are used to process the finally verified burr data segments. The specific operations are as follows: Extract the flow rate change rate gradient, pump speed-pressure phase difference, and range ratio fluctuation coefficient of each burr data segment. For example, the flow rate change rate gradient is 2.5 L / s², the pump speed-pressure phase difference is 0.3 seconds, and the range ratio fluctuation coefficient is 0.1. Store these feature vectors in a Pandas DataFrame to construct a burr feature vector library. Use TensorFlow Federated (TFF) to implement an incremental federated learning model. The specific operations are as follows: Divide the data in the burr feature vector library into multiple client datasets, and each client uses a local model for training. Aggregate the local model updates through the federated averaging function of TFF to generate a global burr discrimination model. For example, the local model of each client uses the Adam optimizer, with a learning rate of 0.001 and a batch size of 32. Use the Pandas library to collect the current operating state parameters of the pump group and five types of dynamic parameters of the candidate burr data segments in real time. The specific operations are as follows: Input the collected parameters (such as the absolute value of the flow rate change rate, the absolute value of the pump speed change rate, etc.) into the global burr discrimination model. The model outputs a burr confidence score. For example, if the burr confidence score is 0.96, directly mark the corresponding candidate burr data segment as the final burr data. Use the Pandas library and NumPy library to process the burr confidence score. The specific operations are as follows: If the burr confidence score is greater than or equal to 0.95, directly mark the corresponding candidate burr data segment as the final burr data. For example, the confidence score of a certain candidate burr data segment is 0.96, and the system automatically marks it as the final burr data and records it in the database. Use a FLIR thermal imaging camera to collect the vibration spectrum of the pump group, and use a Keysight oscilloscope to collect the electromagnetic interference intensity data. The specific operations are as follows: If the burr confidence score is between 0.85 and 0.95, synchronously collect the vibration spectrum of the pump group and the electromagnetic interference intensity data. For example, if the energy increase in the 50 Hz - 200 Hz frequency band of the pump group vibration spectrum is 25 dB, or the electromagnetic interference intensity is 12 mV, then confirm the corresponding candidate burr data segment as the final burr data. Use TensorFlow Federated (TFF) to process the burr data with a low confidence score. The specific operations are as follows: If the burr confidence score is less than 0.85, remove the burr mark and record this data as a false alarm sample. For example, the confidence score of a certain candidate burr data segment is 0.80, and the system marks it as a false alarm sample and adds this sample to the training set of the federated learning model for subsequent incremental training.

[0050] Preferably, the present invention provides a water flow burr data detection system for performing the water flow burr data detection method described above. The water flow burr data detection system includes: A data acquisition module, configured to synchronously acquire first flow time-series data, first pump speed time-series data, and pipe network pressure time-series data, generate water flow-related time-series data; segment and intercept the water flow-related time-series data based on a preset sliding time window to generate multiple candidate glitch data segments; A parameter calculation module, configured to calculate, for each candidate glitch data segment, the absolute value of the flow rate change rate, the absolute value of the pump speed change rate, the absolute value of the pressure change rate, the standard deviation of the differences between adjacent sampling points, and the ratio of the range within the data segment, to generate five types of dynamic parameters; A threshold construction module, configured to obtain the current pump group operation state parameters; construct a dynamic discrimination threshold group according to the current pump group operation state parameters; A data marking module, configured to compare the five types of dynamic parameters with the corresponding thresholds in the dynamic discrimination threshold group respectively. If at least three of the five types of dynamic parameters exceed the thresholds, mark the corresponding candidate glitch data segment as a suspected glitch data segment; A verification and confirmation module, configured to perform multi-level verification on the suspected glitch data segments. The multi-level verification includes leading verification of pump speed change, leading verification of pressure change, verification of pump group start-stop state signals, and verification of historical similar working conditions. If any verification fails, remove the suspected marks in the suspected glitch data segments and output the candidate glitch data segments that pass the verification as the final glitch data.

[0051] Preferably, the present invention further provides a computer-readable medium storing a program capable of being loaded and executed by a processor to perform the water flow glitch data detection method as described above.

[0052] Therefore, from any perspective, the embodiments should be regarded as exemplary and non-restrictive. The scope of the present invention is defined by the appended claims rather than the above description. Therefore, all changes falling within the meaning and scope of the equivalent elements of the application document are intended to be encompassed within the present invention.

[0053] The above are only specific embodiments of the present invention, enabling those skilled in the art to understand or implement the present invention. Various modifications to these embodiments will be obvious to those skilled in the art. The general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the broadest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for detecting burr data of water flow, characterized in that: The following steps are involved: Step S1: synchronously collect first flow time series data, first pump speed time series data and pipe network pressure time series data to generate water service flow related time series data; Based on a preset sliding time window, the water flow related time series data is segmented and intercepted to generate multiple candidate burr data segments; Step S2: for each candidate burr data segment, calculate the absolute value of the flow rate change rate, the absolute value of the first pump speed change rate, the absolute value of the first pressure change rate, the standard deviation of the difference between adjacent sampling points and the proportion of the range in the data segment, and generate five types of dynamic parameters; Step S3: obtaining the current pump group operating state parameters; constructing a dynamic discrimination threshold group according to the current pump group operating state parameters; Step S4: comparing the five types of dynamic parameters with the corresponding thresholds in the dynamic discrimination threshold group respectively, if at least three types of parameters among the five types of dynamic parameters exceed the thresholds, marking the corresponding candidate burr data segment as a suspected burr data segment; Step S5: Perform multi-level verification on the suspected burr data segment, wherein the multi-level verification includes pump speed change pilot verification, pressure change pilot verification, pump group start and stop status signal verification and historical similar working condition verification. If any verification fails, the suspected mark in the suspected burr data segment is removed, and the candidate burr data segment that passes the verification is output as the final burr data.

2. The water flow burr data detection method according to claim 1, characterized in that: In step S1, segmenting and intercepting the water flow-related time series data based on the preset sliding time window includes: Based on a preset sliding time window, the water flow related time series data is segmented and intercepted to generate multiple candidate burr data segments, wherein the length of the preset sliding time window is 0.5 seconds to 5 seconds, and the sliding step is 1 / 5 to 1 / 2 of the window length; Each candidate glitch data segment must meet the following conditions: The number of flow sampling points in the candidate burr data segment is greater than or equal to 50; The candidate burr data segment covers at least one fluctuation interval in which the flow rate change rate exceeds ±0.

1.

3. The water flow burr data detection method according to claim 1, characterized in that: Step S2 includes the following steps: Step S21: Calculate the data segment duration of the candidate burr data segment; Step S22: extracting the maximum flow value and the minimum flow value in the candidate burr data segment, and recording the absolute value of the difference between the maximum flow value and the minimum flow value divided by the data segment duration as the absolute value of the flow change rate; Step S23: obtaining the calculation times identifier of the current pump speed change rate; determining the pump speed change lag time compensation coefficient according to the calculation times identifier of the current pump speed change rate; extracting the maximum pump speed and the minimum pump speed in the candidate burr data segment, and recording the absolute value of the difference between the maximum pump speed and the minimum pump speed divided by the pump speed change lag time compensation coefficient as the first pump speed change rate absolute value; Step S24: obtaining the calculation times identifier of the current pressure change rate; determining the pressure change lag time compensation coefficient according to the calculation times identifier of the current pressure change rate; extracting the maximum value and the minimum value of the pipeline network pressure in the candidate burr data segment, and dividing the absolute value of the difference between the maximum value and the minimum value of the pipeline network pressure by the pressure change lag time compensation coefficient as the first pressure change rate absolute value; Step S25: Calculate the standard deviation of the difference sequence of two adjacent flow sampling points in the candidate burr data segment to obtain the standard deviation of the difference between adjacent sampling points; Step S26: Divide the difference between the maximum flow rate and the minimum flow rate in the candidate burr data segment by the ratio of the average flow rate in the candidate burr data segment to obtain the range ratio in the data segment.

4. The water flow burr data detection method according to claim 3 is characterized in that: In step S23, the pump speed change lag time compensation coefficient is determined according to the calculation number mark of the current pump speed change rate, including: If the calculation times of the current pump speed change rate is marked as the first calculation, 0.5 seconds is set as the pump speed change lag time compensation coefficient; If the calculation times mark of the current pump speed change rate is not the first calculation mark, the absolute value of the previous pump speed change rate is obtained; If the absolute value of the previous pump speed change rate is less than or equal to the preset first pump speed change rate threshold, 0.5 seconds is set as the pump speed change lag time compensation coefficient; If the absolute value of the previous pump speed change rate is greater than the preset first pump speed change rate threshold and the absolute value of the previous pump speed change rate is less than or equal to the preset second pump speed change rate threshold, 1 second is set as the pump speed change lag time compensation coefficient; If the absolute value of the previous pump speed change rate is greater than the preset second pump speed change rate threshold and the absolute value of the previous pump speed change rate is less than or equal to the preset third pump speed change rate threshold, 1.5 seconds is set as the pump speed change lag time compensation coefficient; If the absolute value of the previous pump speed change rate is greater than the preset third pump speed change rate threshold, 2 seconds is set as the pump speed change lag time compensation coefficient, wherein the preset first pump speed change rate threshold is smaller than the preset second pump speed change rate threshold, and the preset second pump speed change rate threshold is smaller than the preset third pump speed change rate threshold.

5. The water flow burr data detection method according to claim 3 is characterized in that: In step S24, determining the pressure change lag time compensation coefficient according to the calculation number mark of the current pressure change rate includes: If the calculation times of the current pressure change rate is marked as the first calculation, 0.2 seconds is set as the pressure change lag time compensation coefficient; If the calculation times mark of the current pressure change rate is not the first calculation mark, the absolute value of the previous pressure change rate is obtained; If the absolute value of the previous pressure change rate is less than or equal to the preset first pressure change rate threshold, 0.2 seconds is set as the pressure change lag time compensation coefficient; If the absolute value of the previous pressure change rate is greater than the preset first pressure change rate threshold and the absolute value of the previous pressure change rate is less than or equal to the preset second pressure change rate threshold, 0.5 seconds is set as the pressure change lag time compensation coefficient; If the absolute value of the previous pressure change rate is greater than the preset second pressure change rate threshold and the absolute value of the previous pressure change rate is less than or equal to the preset third pressure change rate threshold, 0.8 seconds is set as the pressure change lag time compensation coefficient; If the absolute value of the previous pressure change rate is greater than the preset third pressure change rate threshold, 1 second is set as the pressure change lag time compensation coefficient, wherein the preset first pressure change rate threshold is smaller than the preset second pressure change rate threshold, and the preset second pressure change rate threshold is smaller than the preset third pressure change rate threshold.

6. The method for detecting burr data of water flow according to claim 1, characterized in that: Step S3 includes the following steps: Step S31: obtaining the current pump group operating state parameters; identifying the pump group operating mode dynamics according to the current pump group operating state parameters and the preset pump group operating mode determination rules, and generating the current pump group operating mode, wherein the current pump group operating mode is any one of the constant pressure mode, the variable frequency voltage regulation mode and the start-stop transition mode; Step S32: If the current pump group operation mode is the constant pressure mode, 0.5 seconds is set as the pump group lag time compensation coefficient; Step S33: If the current pump group operation mode is the variable frequency pressure regulation mode, the variable frequency pressure regulation rate of the pump group is obtained, and the pump group lag time compensation coefficient is calculated according to the variable frequency pressure regulation rate of the pump group, wherein the specific calculation formula is as follows: ； in, is the pump group delay time compensation coefficient, The variable frequency pressure regulation rate of the pump group; Step S34: if the current pump group operation mode is the start-stop transition mode, 1.2 seconds is set as the pump group lag time compensation coefficient; Step S35: Determine the flow rate change rate threshold, pump speed change rate threshold, pressure change rate threshold, standard deviation multiple threshold and range ratio threshold according to the pump group lag time compensation coefficient, and obtain five types of dynamic parameter thresholds, among which the specific calculation formulas of the flow rate change rate threshold, pump speed change rate threshold, pressure change rate threshold, standard deviation multiple threshold and range ratio threshold are as follows: I=0.15×(1+0.1× )； B = 0.05 × (1 + 0.2 × f); Y=3×(1+0.15× )； C = 1.5 × (1 + 0.05 × p); J=0.2×(1−0.08× )； Among them, I is the flow rate change rate threshold, B is the pump speed change rate threshold, Y is the pressure change rate threshold, C is the standard deviation multiple threshold, J is the range ratio threshold, f is the pump speed fluctuation amplitude peak, and p is the standard deviation of the pipeline network pressure fluctuation; Step S36: collecting a heat map of the internal temperature distribution of the pump group; identifying a local hot spot area of ​​the pump group winding based on the heat map of the internal temperature distribution of the pump group, and generating a temperature distribution map of the pump group winding, wherein the temperature distribution map of the pump group winding includes temperature gradient distribution data and multiple areas; Step S37: judging whether there is a region in the pump winding temperature distribution diagram whose temperature is greater than the mean temperature of the adjacent regions +3°C according to the temperature gradient distribution data and the hot spot coordinate set; if so, marking the corresponding region as the pump winding hot spot region; Step S38: Calculate the area ratio of the hot spot area based on the temperature distribution diagram of the pump winding; calculate the threshold weight coefficient according to the area ratio of the hot spot area, wherein the specific calculation formula is as follows: ； in, is the threshold weight coefficient, R is the area ratio of the hotspot area; Step S39: performing partition correction on the five types of dynamic parameter thresholds according to the threshold weight coefficients to obtain the corrected five types of dynamic parameter thresholds; Step S310: Generate a dynamic discrimination threshold group based on five types of dynamic parameter thresholds, wherein if the hot spot area of ​​the winding of the same pump group is triggered continuously for more than or equal to 3 times, each threshold of the five types of dynamic parameter thresholds will be reduced by 20%.

7. The water flow burr data detection method according to claim 1, characterized in that: Step S5 includes the following steps: Step S51: performing a pump speed change pilot verification on the suspected burr data segment to obtain a pump speed change pilot verification result; Step S52: performing pressure change pilot verification on the suspected burr data segment to obtain a pressure change pilot verification result; Step S53: verifying the pump group start / stop status signal for the suspected burr data segment to obtain a pump group start / stop status signal verification result; Step S54: verifying the suspected burr data segment under historical similar working conditions to obtain a historical similar working condition verification result; Step S55: Based on the verification results of steps S51 to S54, the candidate glitch data segments that have passed the verification are output as the final glitch data.

8. The method for detecting burr data of water flow according to claim 7, characterized in that: The method for verifying that the pump speed change time leads the flow rate change time within 1 second to 3 seconds before the suspected burr data segment in step S51 includes: Construct a timestamp sequence of pump speed change events within 1 second to 3 seconds before the suspected burr data segment to generate second pump speed time series data; Construct a timestamp sequence of traffic change events from 1 second before to 3 seconds before the suspected burr data segment to generate second traffic time series data; Identify an event whose absolute value of the rate of change is greater than or equal to 0.03 in the second pump speed time series data, and record the event triggering timestamp T1; Identify an event whose absolute value of the rate of change is greater than or equal to 0.05 in the second traffic time series data, and record the event triggering timestamp T2; Calculate the time difference between T1 and T2 T.

9. A water flow burr data detection system, characterized in that: Used to execute the water flow burr data detection method as claimed in claim 1, the water flow burr data detection system comprises: A data acquisition module is used to synchronously acquire the first flow time series data, the first pump speed time series data and the pipe network pressure time series data to generate water service flow related time series data; segmentally intercept the water service flow related time series data based on a preset sliding time window to generate multiple candidate burr data segments; The parameter calculation module is used to calculate the absolute value of the flow rate change rate, the absolute value of the pump speed change rate, the absolute value of the pressure change rate, the standard deviation of the difference between adjacent sampling points and the proportion of the range in the data segment for each candidate burr data segment, and generate five types of dynamic parameters; A threshold building module is used to obtain the current pump group operating state parameters; and to build a dynamic discrimination threshold group according to the current pump group operating state parameters; A data marking module is used to compare the five types of dynamic parameters with the corresponding thresholds in the dynamic discrimination threshold group respectively, and if at least three types of parameters among the five types of dynamic parameters exceed the thresholds, the corresponding candidate burr data segment is marked as a suspected burr data segment; The verification and confirmation module is used to perform multi-level verification on the suspected burr data segment, where the multi-level verification includes pump speed change pilot verification, pressure change pilot verification, pump group start and stop status signal verification and historical similar working condition verification. If any verification fails, the suspected mark in the suspected burr data segment is removed, and the candidate burr data segment that passes the verification is output as the final burr data.

10. A computer-readable medium, characterized in that A program is stored which can be loaded by a processor and execute the method for detecting burr data of water flow according to any one of claims 1 to 8.

Images