Real-time calculation method and system for characteristic value of dynamic data of water depth pressure, electronic equipment and storage medium

Abstract

The application provides a water depth pressure dynamic data characteristic value real-time calculation method and system, electronic equipment and storage medium, relates to the water pressure monitoring technical field, the application collects the voltage signal reflecting the water depth change of multiple channels, analyzes the correlation to determine the main fluctuation frequency of water flow; According to this, set the merging ratio, fuse multiple channel signals into target signals; Then separate the target signal into smooth trend part and mutation part, and set the mutation judgment threshold based on energy change; The threshold is used to detect the mutation time interval in the mutation part, and the maximum fluctuation amplitude and duration in the interval are extracted as the first characteristic value; At the same time, the long-term reference offset of the voltage acquisition device is predicted and corrected by using the smooth trend part, and the maximum value, minimum value and fluctuation range of the corrected signal output are taken as the second characteristic value, which can realize real-time feature extraction of dynamic water depth pressure, accurately capture mutation events and output stable extreme value range after drift correction.

Description

Technical Field

[0001] This application relates to the field of water pressure monitoring technology, and in particular to a method, system, electronic device and storage medium for real-time calculation of dynamic data characteristic values ​​of water depth and pressure. Background Technology

[0002] In applications involving high-speed fluids or complex hydrodynamic environments, real-time acquisition of accurate dynamic data on water depth or pressure is crucial. In these scenarios, the signals acquired by sensors often contain strong transient interference pulses caused by cavitation collapse, vortex shedding, or turbulent impacts. The technical requirement is to develop a method for processing such signals in real time. This method must not only robustly resist strong transient interference but also accurately separate and quantify characteristic values ​​reflecting stable fluid pulsations and key features reflecting transient events. Furthermore, it must possess the ability to correct for reference drift caused by prolonged sensor operation online, ensuring the stability and reliability of the output characteristic values ​​during long-term monitoring.

[0003] One current targeted technical solution is to combine wavelet thresholding denoising with time-domain statistical analysis. This method first uses wavelet transform to decompose the original pressure signal into multiple scales. High-frequency wavelet coefficients are suppressed or zeroed using a preset threshold to filter out noise and some transient interference, and then the signal is reconstructed. Based on the reconstructed signal, time-domain statistical analysis is used to calculate statistical quantities such as maximum, minimum, mean, and variance as feature values. Sometimes, simple peak detection is used to identify significant pulses.

[0004] However, this existing approach has significant drawbacks: under strong transient interference, wavelet thresholding denoising struggles to effectively distinguish between real fluid transient events with similar time-frequency characteristics and interference noise, leading to either excessive filtering of useful transient signals resulting in the loss of key features, or insufficient interference suppression contaminating steady-state features. Furthermore, the complete process of wavelet decomposition, thresholding, and reconstruction is computationally complex, making it difficult to meet the real-time processing requirements of high-sampling-rate data, and particularly unsuitable for deployment on resource-constrained edge devices. In addition, this method lacks a systematic separation mechanism for steady-state trends and transient mutation components in the signal, and also lacks online prediction and correction functions for sensor reference drift, causing the extracted steady-state feature values ​​to deviate over time, failing to guarantee the accuracy and reliability of feature values ​​in long-term monitoring. Summary of the Invention

[0005] The purpose of this application is to provide a method, system, electronic device and storage medium for real-time calculation of dynamic data feature values ​​of water depth and pressure, so as to solve the problem in the prior art that it is impossible to simultaneously achieve accurate identification of real fluid transient events, robust extraction of steady-state features and real-time correction of sensor drift under strong transient interference background.

[0006] To address the aforementioned technical problems, in a first aspect, this application provides a method for real-time calculation of dynamic feature values ​​of water depth pressure data, comprising:

[0007] Voltage signals from multiple channels are acquired, and these voltage signals are used to reflect changes in water depth.

[0008] By analyzing the correlation between voltage signals from multiple channels, the main frequency of water flow fluctuations can be determined.

[0009] The signal merging ratio is determined based on the main frequency, and the voltage signals of multiple channels are merged using the signal merging ratio to generate the target signal.

[0010] The target signal is separated to obtain a first signal portion reflecting the smooth trend of the target signal and a second signal portion reflecting the sudden change of the target signal, and a judgment threshold for identifying the sudden change of the target signal is set according to the energy change of the target signal;

[0011] Using the aforementioned judgment threshold, the time intervals in which abrupt changes occur are detected and marked in the second signal portion, and the voltage signal within the time interval is analyzed to determine the maximum amplitude of the fluctuation and the duration of the abrupt change within the time interval as the first characteristic value of the water depth pressure data.

[0012] By utilizing the slow-changing characteristics contained in the first signal portion, the reference offset caused by long-term use of the voltage acquisition device is predicted, and the first signal portion is corrected according to the reference offset. Based on the corrected first signal portion, the maximum value, minimum value and fluctuation range of the target signal are output as the second feature value of the water depth pressure data.

[0013] Optionally, the step of separating the target signal to obtain a first signal portion reflecting the smooth trend of the target signal and a second signal portion reflecting the abrupt change of the target signal, and setting a judgment threshold for identifying the abrupt change of the target signal based on the energy change of the target signal, includes:

[0014] The target signal is averaged to obtain a smoothed sequence, which is defined as the first signal part.

[0015] The first signal portion is subtracted from the target signal to obtain the change difference sequence, and the change difference sequence is defined as the second signal portion;

[0016] Multiple time periods are continuously selected on the target signal, and the sum of the squared values ​​of the signal values ​​in each time period is calculated. The sum of the squared values ​​is defined as the local energy value.

[0017] The highest and lowest energy levels of the local energy values ​​are statistically analyzed across all time periods, and a dynamic judgment threshold is generated based on the highest and lowest energy levels.

[0018] Optionally, the step of statistically analyzing the highest and lowest energy levels of the local energy values ​​across all time periods, and calculating a dynamic judgment threshold based on the highest and lowest energy levels, includes:

[0019] Check whether the local energy values ​​for each time period are complete and valid, exclude time periods containing missing signal points, and retain complete and valid local energy values ​​to form a valid energy set;

[0020] Find the three local energy values ​​with the largest values ​​in the effective energy set, and calculate the arithmetic mean of these three local energy values ​​as the highest energy level;

[0021] Find the three local energy values ​​with the smallest values ​​in the effective energy set, and calculate the arithmetic mean of these three local energy values ​​as the lowest energy level;

[0022] Calculate the ratio of the highest energy level to the lowest energy level, and multiply the ratio by a preset sensitivity coefficient to obtain the basic threshold value;

[0023] Calculate the dispersion value of the effective local energy value over all time periods, multiply the dispersion value by a preset fluctuation compensation factor to generate a dynamic compensation value, and modify the basic threshold value according to the dynamic compensation value to generate a dynamic judgment threshold.

[0024] Optionally, determining the dominant frequency of water flow fluctuations by analyzing the correlation between voltage signals from multiple channels includes:

[0025] The voltage signals from multiple channels are combined in pairs to form voltage signal pairs, and each voltage signal pair contains a first channel signal and a second channel signal.

[0026] For each voltage signal pair, the second channel signal is moved step by step on the time axis according to a preset step size. The matching degree value between the second channel signal and the first channel signal is calculated after each movement, and the number of movement steps corresponding to the peak value of the matching degree value is recorded as the time delay value of the voltage signal pair.

[0027] Summarize the time delay values ​​of all voltage signal pairs and establish a set of time delay relationships;

[0028] The starting positions of the voltage signals in different channels are adjusted according to the set of time delay relationships, so that the voltage signals of multiple channels are matched with each other on the time axis. The voltage signals of multiple channels after time matching are superimposed synchronously to form a comprehensive signal.

[0029] Intensity analysis is performed on the comprehensive signal to identify the highest frequency point where the intensity value exceeds a preset threshold as the main frequency of water flow fluctuation.

[0030] Optionally, the step of determining the signal combining ratio based on the main frequency and using the signal combining ratio to combine the voltage signals of multiple channels to generate the target signal includes:

[0031] At the main frequency, the signal strength value of the voltage signal of each channel is determined, and the voltage signal with the highest signal strength value among all channels is selected as the reference voltage signal;

[0032] For each channel voltage signal, the product of the signal strength value and the signal strength of the reference voltage signal at the main frequency is calculated, and the numerical result of the signal strength product is used as the correlation strength of the corresponding channel.

[0033] Divide the correlation strength of each channel by the sum of the correlation strength values ​​of all channels to obtain the signal combining ratio of the corresponding channel;

[0034] The voltage signal of each channel is multiplied by the corresponding signal merging ratio to generate a proportional adjustment signal, and all the proportional adjustment signals are arithmetically added together to obtain the target signal.

[0035] Optionally, the step of using the judgment threshold to detect and mark the time interval where the sudden change occurs in the second signal portion, and analyzing the voltage signal within the time interval to determine the maximum amplitude of the fluctuation and the duration of the sudden change as the first feature value of the water depth pressure data, includes:

[0036] The signal amplitude value at each time point is continuously scanned on the second signal part, and when the signal amplitude value first exceeds the judgment threshold, it is recorded as the time point of sudden change.

[0037] Continuous monitoring begins from the mutation initiation time point. When the duration of the signal amplitude value continuously falling below the judgment threshold reaches a preset stability threshold, it is recorded as the mutation termination time point.

[0038] Calculate the time difference between the mutation termination time and the mutation start time, and use this time difference as the mutation duration;

[0039] During the duration of the mutation, all signal amplitude values ​​of the second signal portion are traversed, and the signal amplitude value with the largest absolute value is recorded as the maximum fluctuation amplitude.

[0040] The maximum amplitude of the fluctuation and the duration of the sudden change are output as the first feature value of the water depth pressure data.

[0041] Optionally, the step of using the slowly changing characteristics contained in the first signal portion to predict the reference offset caused by long-term use of the voltage acquisition device, correcting the first signal portion according to the reference offset, and outputting the maximum value, minimum value, and fluctuation range of the target signal as the second feature value of the water depth pressure data based on the corrected first signal portion includes:

[0042] The first signal portion is divided into multiple consecutive time periods in chronological order. A linear trend is fitted within each time period, and the linear trends of all time periods are connected to form a baseline offset prediction line.

[0043] The predicted value at the same time point as the first signal part is intercepted on the reference offset prediction line, and the predicted value at the corresponding time point is subtracted from the value of each point of the first signal part to generate the corrected first signal part.

[0044] In the corrected first signal section, the signal values ​​are recorded at all time points. The highest signal value that appears during the scanning process is marked as the maximum value, and the lowest signal value is marked as the minimum value.

[0045] Calculate the arithmetic difference between the maximum and minimum values, and define this arithmetic difference as the fluctuation range;

[0046] Simultaneously, the maximum value, minimum value, and fluctuation range are output as the second characteristic value of the water depth pressure data.

[0047] Secondly, this application provides a real-time calculation system for dynamic water depth pressure data characteristic values, including:

[0048] The acquisition module is used to acquire voltage signals from multiple channels, and the voltage signals are used to reflect changes in water depth.

[0049] The analysis module is used to determine the main frequency of water flow fluctuations by analyzing the correlation between voltage signals from multiple channels;

[0050] The generation module is used to determine the signal merging ratio based on the main frequency, and to merge the voltage signals of multiple channels to generate a target signal using the signal merging ratio.

[0051] The identification module is used to separate the target signal, obtain a first signal portion reflecting the smooth trend of the target signal and a second signal portion reflecting the sudden change of the target signal, and set a judgment threshold for identifying the sudden change of the target signal based on the energy change of the target signal;

[0052] The output module is used to detect and mark the time interval in the second signal part where the sudden change occurs using the judgment threshold, and to analyze the voltage signal in the time interval to determine the maximum amplitude of the fluctuation and the duration of the sudden change in the time interval as the first feature value of the water depth pressure data.

[0053] The output module is also used to predict the reference offset caused by long-term use of the voltage acquisition device by utilizing the slow change characteristics contained in the first signal portion, and to correct the first signal portion according to the reference offset, and to output the maximum value, minimum value and fluctuation range of the target signal as the second feature value of the water depth pressure data based on the corrected first signal portion.

[0054] Thirdly, this application provides an electronic device, comprising:

[0055] Memory, used to store computer programs;

[0056] A processor is configured to execute the computer program to implement the steps of the real-time calculation method for dynamic data feature values ​​of water depth and pressure as described in the first aspect above.

[0057] Fourthly, this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, can implement the steps of the real-time calculation method for dynamic water depth pressure data characteristic values ​​as described in the first aspect above.

[0058] The real-time calculation method for dynamic features of water depth and pressure data provided in this application determines the dominant frequency of water flow through correlation analysis of multi-channel voltage signals and optimizes the signal merging ratio accordingly, effectively fusing multi-source information to suppress random noise. Then, it separates the smooth trend and abrupt change components of the target signal, and combines energy adaptive setting of abrupt change threshold to accurately capture and quantify the key features of real fluid transient events under strong transient interference background. At the same time, it uses the smooth trend to predict and correct long-term sensor drift, ensuring stable output of the corrected steady-state pressure extreme value range and fluctuation characteristics in a strong interference environment. Finally, it achieves high-precision real-time synchronous extraction of transient event features and drift-robust steady-state features from dynamic water depth and pressure data containing transient changes.

[0059] Furthermore, the average sequence of the target signal is calculated as the first signal component reflecting the smoothing trend, and the difference between the original signal and this average sequence is used as the second signal component reflecting abrupt changes. Simultaneously, the target signal is continuously divided into time periods, and the sum of the squared values ​​within each period is calculated. Finally, a threshold for judging abrupt changes is dynamically generated based on the highest and lowest levels of local energy values ​​across all time periods. This method utilizes mean separation to achieve rapid decoupling of signal trends and abrupt change components, ensuring real-time performance. It captures the intensity of dynamic signal fluctuations through local energy statistics and adaptively sets dynamic thresholds based on energy ranges, significantly improving sensitivity to abrupt events and resistance to random interference, avoiding the shortcomings of fixed thresholds in complex fluid environments. Attached Figure Description

[0060] To more clearly illustrate the technical solutions of the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0061] Figure 1 A flowchart illustrating a method for real-time calculation of dynamic water depth pressure feature values ​​provided in an embodiment of this application;

[0062] Figure 2 A flowchart illustrating the specific implementation of a method for real-time calculation of dynamic water depth pressure data feature values ​​provided in this application embodiment;

[0063] Figure 3 This application provides a specific implementation scenario diagram for a method for real-time calculation of dynamic water depth pressure data feature values.

[0064] Figure 4 This is a schematic diagram of the structure of a real-time calculation system for dynamic water depth pressure data feature values ​​provided in an embodiment of this application. Detailed Implementation

[0065] In the field of fluid pulsating pressure monitoring with strong transient interference, existing schemes based on wavelet threshold denoising combined with time-domain statistical analysis have significant limitations: they struggle to effectively distinguish between real fluid transient events with similar time-frequency characteristics and interference noise, leading to either over-filtering of useful transient signals and loss of key features, or insufficient interference suppression contaminating steady-state features; simultaneously, complex wavelet operations are insufficient to meet the real-time processing requirements of high sampling rate data; furthermore, the scheme lacks an online prediction and correction mechanism for long-term sensor reference drift, resulting in systematic deviations in extracted steady-state pressure extreme value ranges and other feature values ​​over time, failing to guarantee the accuracy and reliability of long-term monitoring. These shortcomings severely restrict the real-time, robust, and long-term stable accurate extraction of dynamic features of water depth pressure in critical scenarios such as high-speed fluid machinery and marine engineering.

[0066] To address the aforementioned challenges, this invention proposes a real-time calculation method for dynamic feature values ​​of water depth and pressure data. The method first determines the dominant frequency of the water flow by analyzing the correlation of multi-channel voltage signals, and then optimizes the merging ratio to generate a noise-enhanced target signal. Next, it separates the smooth trend portion and abrupt change component of the target signal, i.e., the first signal portion and the second signal portion, and adaptively sets a dynamic judgment threshold based on signal energy changes. This threshold is used to accurately identify the abrupt change time interval in the second signal portion, extracting its maximum fluctuation amplitude and duration as the first feature value, i.e., transient event feature. Simultaneously, the slow change characteristics of the first signal portion are used to predict and correct the sensor reference offset, and the maximum, minimum, and fluctuation range of the corrected signal output are used as the second feature value, i.e., steady-state feature. This method fundamentally solves the core contradictions of existing technologies: through signal separation and adaptive thresholding, it accurately captures and quantifies real transient events under strong interference, avoiding the loss of useful signals or interference contamination; through multi-channel dominant frequency fusion, it simplifies the processing flow and meets real-time requirements; through drift prediction and correction, it ensures the stability and reliability of steady-state feature values ​​in long-term monitoring, achieving high-precision synchronous extraction of transient and steady-state features.

[0067] To enable those skilled in the art to better understand the present application, the present application will be further described in detail below with reference to the accompanying drawings and specific embodiments. Obviously, the described embodiments are merely some embodiments of the present application, and not all embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0068] The core of this application is to provide a method for real-time calculation of dynamic characteristic values ​​of water depth pressure data, and a flowchart of one specific implementation is shown below. Figure 1 As shown, the method includes:

[0069] S101. Acquire voltage signals from multiple channels, the voltage signals being used to reflect changes in water depth;

[0070] In the above scheme, the multi-channel voltage signal refers to the dynamic sequence of the original electrical signals output by the pressure sensor array, which includes the linear mapping characteristics between water depth pressure and voltage amplitude. Specifically, it is the voltage fluctuation corresponding to a unit water depth change, which can be used to invert the real-time pressure field distribution characteristics of the underwater environment.

[0071] In this embodiment, firstly, the raw voltage values ​​of all sensors are synchronously acquired at a fixed sampling rate. If a channel experiences an instantaneous signal anomaly, the system automatically switches to a backup sensor and triggers an alarm. Secondly, the raw signals are subjected to sliding mean filtering, which calculates the average voltage value within a window and replaces the current value. For example, 10 consecutive values ​​[1.02V, 1.05V, 0.99V...] in channel A are filtered to output 1.03V. Next, the integrity of the filtered signal is checked: if more than 5 consecutive sampling points are missing from a single channel, all channel data for that time period are discarded; otherwise, linear interpolation is used for the missing points. Finally, a high-precision timestamp is added to all valid signals to ensure strict alignment of the time axis of the multi-channel data. For example, the data at 12:00:00.000 in channel B must be associated with the data at the same time in other channels.

[0072] In practical applications, during the dynamic monitoring of a water conveyance pipeline at a certain water conservancy project, three pressure sensors located on the outside of a bend, at the bottom center, and at the top synchronously collected voltage signals. The bend sensor experienced drastic voltage fluctuations (1.8V~3.2V) due to local cavitation, while the bottom sensor maintained a stable output (2.6V±0.05V) due to steady-state water flow. The top sensor experienced a sudden anomaly due to bubble adhesion, causing the voltage at three consecutive sampling points to drop to 0.3V. The system immediately activated a backup sensor at the same location to replace the faulty node, smoothed high-frequency turbulence noise using a sliding mean filter, and performed linear interpolation compensation for abnormal periods based on adjacent sampling points. For example, at t=115, the voltage was calculated to be 2.12V using 2.1V and 2.15V at t=114 and t=116. The final output was a three-channel corrected voltage sequence with a time synchronization accuracy of 0.5ms, fully covering the transient impact and steady-state pressure change characteristics during the flood discharge process.

[0073] The overall solution of S101 described above, through redundant deployment of multiple sensors and real-time quality monitoring, ensures the reliability and synchronization of raw data in complex fluid environments, provides highly complete input for subsequent analysis, and avoids feature extraction failure due to single point of failure or noise.

[0074] S102. By analyzing the correlation between voltage signals from multiple channels, the main frequency of water flow fluctuations can be determined.

[0075] Optionally, step S102 may specifically include the following steps:

[0076] S1021. Combine the voltage signals from multiple channels in pairs to form voltage signal pairs, each voltage signal pair containing a first channel signal and a second channel signal;

[0077] S1022. For each voltage signal pair, the second channel signal is moved step by step on the time axis according to a preset step size. The matching degree value between the second channel signal and the first channel signal is calculated after each movement, and the number of movement steps corresponding to the peak value of the matching degree value is recorded as the time delay value of the voltage signal pair.

[0078] S1023. Summarize the time delay values ​​of all voltage signal pairs and establish a set of time delay relationships;

[0079] S1024. Adjust the starting position of the voltage signal in different channels according to the set of time delay relationships, so that the voltage signals of multiple channels are matched with each other on the time axis, and simultaneously superimpose the voltage signals of multiple channels after time matching to form a comprehensive signal.

[0080] S1025. Perform intensity analysis on the comprehensive signal and identify the highest frequency point where the intensity value exceeds a preset threshold as the main frequency of water flow fluctuation.

[0081] In the above scheme, a voltage signal pair refers to a dynamic combination unit of signals from any two spatially located channels in the pressure sensor array. It includes the spatiotemporal correlation characteristics of pressure disturbance wave propagation and can be used to reconstruct the propagation path of the fluid pressure field. The matching degree value is a quantitative index of the similarity between two channel signals under time-shifted conditions. It includes waveform amplitude correlation characteristics, specifically the sum of voltage products in the signal overlap area, and can be used to determine the spatial consistency of the pressure disturbance wave. The time delay value is the propagation time difference of the pressure disturbance wave between the sensor pairs. It includes the signal hysteresis characteristics under the influence of the fluid velocity field, specifically the signal travel time required for the matching degree value to reach its peak, and can be used to invert the local flow velocity distribution. The time delay relationship set refers to the topological model of the signal propagation time difference of the entire sensor network. It includes the systematic correlation characteristics of the phase difference of multi-channel signals, specifically a matrix relationship table of inter-channel delay values, and can be used to establish the mapping relationship between sensor spatial location and signal phase. The composite signal refers to the fused and enhanced signal after spatiotemporal alignment of multiple channels. It includes the cooperative response characteristics of fluid pressure waves in the sensor network, specifically manifested as a voltage sequence superimposed from the signals of each channel after delay compensation. This signal can be used to extract global water flow fluctuation characteristics submerged in noise. Intensity analysis refers to the quantitative assessment of the frequency domain energy of the composite signal, including the energy accumulation characteristics of periodic pressure fluctuations in the frequency domain. This signal is specifically manifested as the squared value of the signal energy at each frequency point, and can be used to identify the characteristic frequencies of dominant water flow fluctuations.

[0082] In this embodiment, step S1021 first determines all available sensor channels, and then uses a permutation and combination algorithm to generate pairs of voltage signal pairs. Each signal pair explicitly designates a first channel (reference signal) and a second channel (signal to be moved), ensuring that all spatial relationships are fully covered. The system automatically excludes self-pairing and duplicate pairings. For example, three sensors generate three sets of voltage signal pairs, namely AB, AC, and BC.

[0083] Secondly, in step S1022, a time-domain cross-correlation operation is performed on each voltage signal pair: the first channel signal is fixed, and the second channel signal is gradually moved by a preset step size (Δt=0.1ms). After each movement, the normalized cross-correlation coefficient of the overlapping segment of the two signals is calculated as the matching degree value, and the calculation formula is as follows: ,in Corresponding to the first channel signal, Corresponding to the second channel signal. Record the number of movement steps N corresponding to the peak of the matching degree value, and calculate the actual delay value. The calculation is completed by iterating through all signal pairs. For example, for the AB voltage signal pair, with a step size of 0.1ms, the matching degree value at step 5 is... To calculate the actual latency value, take the peak value as an example. .

[0084] Next, an N×N delay matrix is ​​created through step S1023, where N = the number of channels and the matrix elements are... Storage Channel arrive The delay value. The reverse delay is automatically derived based on physical characteristics: if... ,but Diagonal elements It is always 0. This eventually forms a complete spatiotemporal relationship mapping table.

[0085] Then, in step S1024, a reference channel is selected, usually the first channel A by default. The time axes of other channels are adjusted according to the delay matrix: the overall signal offset of channel j is adjusted. After alignment, all channel signals are strictly overlapped according to their timestamps, and point-by-point arithmetic superposition is performed: Generate a composite signal. For example, if the reference channel is A, channel B is offset by -0.5ms, corresponding to a forward shift of 0.5ms; channel C is offset by -0.8ms, corresponding to a forward shift of 0.8ms. After all channel signals are strictly aligned according to their timestamps, they are then superimposed point by point using arithmetic. This generates a composite signal.

[0086] Finally, the synthesized signal is subjected to a Fast Fourier Transform in step S1025 to calculate the frequency energy spectrum. Identify the global maximum energy value. ,set up Filter all The frequency point corresponding to the highest energy is selected as the primary frequency. For example, after the synthesized signal undergoes a Fast Fourier Transform, the output shows that the global maximum energy value is at 5Hz. =100, let α=0.3 then The frequencies exceeding the threshold were counted, with 4Hz corresponding to energy 35, 5Hz corresponding to energy 100, and 6Hz corresponding to energy 32. Finally, the 5Hz frequency with the highest energy was selected as the main frequency.

[0087] In practical applications, three pressure sensors were deployed in the dynamic monitoring of the pressure steel pipe of a hydropower station: A was located 50m upstream of the turbine in the straight pipe section, B was at the inlet of the spiral casing, and C was fixed at the bend of the tailrace pipe. The system executes the main frequency analysis process: First, three sets of voltage signal pairs (AB, AC, BC) are generated; for the AB voltage signal pair, the B channel signal is moved in 0.1s steps. When it moves to the 3rd step, the cross-correlation value reaches a peak of 0.88, corresponding to the theoretical delay of water flow velocity of 5m / s propagating 15m; the AC delay of 0.8s and the BC delay of 0.5s are measured simultaneously, and a time delay relationship set is constructed based on this; the signal is aligned with A as the reference: B is moved forward by 0.3s and C is moved forward by 0.8s. After superposition, the combined signal has an amplitude increase of 2.4 times at the impeller passing frequency of 1.67Hz; after fast Fourier transform analysis, the energy peak at 1.67Hz is 650, and the secondary peak at 3.34Hz is 210. After setting the threshold to 650×30%=195, it is confirmed as the main frequency, which perfectly matches the physical characteristics of the unit speed of 100rpm.

[0088] The overall scheme of S102 described above effectively overcomes the phase deviation caused by sensor position differences through spatiotemporal correlation analysis of multi-channel voltage signals, accurately quantifies the pressure wave propagation delay using signal cross-correlation technology, significantly enhances the essential water flow wave characteristics through global alignment and synergistic superposition, and finally robustly extracts the dominant frequency from the frequency domain energy spectrum of the comprehensive signal, providing a highly reliable frequency reference for subsequent signal fusion, fundamentally solving the problem of misjudgment or missed detection of the dominant frequency in traditional single-channel spectrum analysis under strong interference environment.

[0089] S103. Determine the signal merging ratio based on the main frequency, and use the signal merging ratio to merge the voltage signals of multiple channels to generate the target signal;

[0090] Optionally, step S103 may specifically include the following steps:

[0091] S1031. At the main frequency, determine the signal strength value of the voltage signal of each channel, and select the voltage signal with the highest signal strength value among all channels as the reference voltage signal.

[0092] S1032. For each channel voltage signal, calculate the product of the signal strength value and the signal strength of the reference voltage signal at the main frequency, and use the numerical result of the signal strength product as the correlation strength of the corresponding channel.

[0093] S1033. Divide the correlation strength of each channel by the sum of the correlation strength values ​​of all channels to obtain the signal combining ratio of the corresponding channel;

[0094] S1034. Multiply the voltage signal of each channel by the corresponding signal merging ratio to generate a proportional adjustment signal, and arithmetically add all the proportional adjustment signals to obtain the target signal.

[0095] In the above scheme, the signal strength value refers to the energy quantification index of the voltage signal at a specific frequency point, including frequency-selective amplitude response characteristics. Specifically, it is represented by the squared amplitude value at that frequency in the Fast Fourier Transform spectrum, and can be used to evaluate the sensor's sensitivity to dominant water flow fluctuations. The reference voltage signal refers to the channel signal with the strongest energy response at the dominant frequency point, including the optimal observation point characteristics in the fluid pressure wave propagation path. Specifically, it is represented by the signal corresponding to the maximum signal strength value of all channels at the dominant frequency, and can serve as a benchmark anchor point for multi-source data fusion. The correlation strength refers to the synergistic quantification index between a single-channel signal and the reference signal at the dominant frequency, including frequency domain energy correlation characteristics. Specifically, it is represented by the product of the channel signal strength value and the reference signal strength value, and can be used to characterize the spatial coherence of the sensor network. The signal merging ratio refers to the normalized weighting coefficient of the channel signals in the fusion, including the distribution characteristics of the dominant frequency component contribution. Specifically, it is represented by the ratio of the correlation strength of a single channel to the sum of the correlation strengths of all channels, and can be used to achieve noise-suppressed intelligent weighting. The scaling signal refers to the version of the original voltage signal scaled according to the merging ratio. It includes channel-specific gain control characteristics, specifically the product of the original signal value and the corresponding merging ratio. It can be used to generate a preprocessed signal that highlights the dominant frequency component. The target signal refers to the fused output of the multi-channel scaling signals. It includes spatially coordinated enhanced pressure field reconstruction characteristics, specifically the arithmetic sum sequence of all scaling signals. It can serve as the core input source for subsequent transient and steady-state feature extraction.

[0096] In this embodiment, the main frequency determined in step S1025 is first obtained through step S1031. A fast Fourier transform is then performed on each channel voltage signal to extract the signal strength value at that frequency. The strength calculation formula is as follows: Compare the intensity values ​​of all channels and select the voltage signal of the channel with the highest value as the reference voltage signal. For example, if the energy values ​​of the three channels at the main frequency of 5Hz are: Channel A - 120V², Channel B - 90V², and Channel C - 60V², then Channel A is selected as the reference signal.

[0097] Secondly, based on the voltage signal strength value of the reference channel in step S1032... For each channel, calculate the correlation strength of each channel: convert the single-channel signal strength value Compared with the reference channel signal strength value The multiplication yields the correlation strength of each channel. This operation quantifies the compatibility between each channel and the reference signal on the main frequency component. For example, the signal strength value of reference channel A. The signal strength value corresponding to channel B Channel C corresponds to the signal strength value Then the correlation strength of channel B Then the correlation strength of channel C Channel A's self-correlation strength .

[0098] Next, the sum of the correlation strengths of all channels is calculated in step S1033. Then, divide the single-channel correlation strength by the sum to obtain the normalized merging ratio. The calculation formula is as follows: ,in Represents the normalized merging ratio, The correlation strength corresponding to each channel, This is the sum of the correlation strengths of all channels. Continuing the previous example, Channel A ratio Channel B ratio Channel C ratio The total verification result is 1.

[0099] Finally, the original signals of each channel are processed in step S1034. Multiply by the corresponding merging ratio A proportional adjustment signal is generated, calculated using the following formula: Proportional Adjustment Signal Then, all adjustment signals are added point by point to generate the final target signal. The calculation formula is as follows: Target signal For example, the original value at a certain moment: , After adjustment: , , Target signal .

[0100] In practical applications, during the dynamic monitoring of the spillway gate of a tidal power station, three pressure sensors were installed at the gate's leading edge (channel A), 5m downstream (channel B), and the seabed base (channel C). The system detected the tidal cycle corresponding to the dominant water flow frequency of 1.25Hz. First, the signal strength values ​​of each channel were calculated at the dominant frequency of 1.25Hz: channel A had an energy of 500V², channel B 650V², and channel C 300V². Channel B was selected as the reference voltage signal. Next, the correlation strength was calculated: the correlation strength of channel A... =500×650=325,000, Channel B's self-correlation strength =650×650=422,500, Channel C correlation strength =300 × 650 = 195,000; then calculate the merging ratio: total correlation strength. =325,000 + 422,500 + 195,000 = 942,500, Channel A ratio Channel B ratio Channel C ratio Finally, at t=12.8s, the original voltage values ​​for channel A were 3.5V, channel B was 3.2V, and channel C was 2.8V. After proportional adjustment, they were approximately 1.208V (3.5×0.345), 1.434V (3.2×0.448), and 0.580V (2.8×0.207). The superposition generated the target signal value of 3.222V, which significantly enhanced the tidal frequency characteristics and weakened the random noise of the waves.

[0101] The overall scheme of S103 described above is based on intelligent allocation of channel weights according to the dominant water flow fluctuation frequency. Through correlation-weighted signal fusion, it significantly enhances the target frequency components and suppresses uncorrelated noise, generating a high-fidelity pressure characteristic signal, providing a high-quality input source for subsequent transient and steady-state separation.

[0102] S104. Separate the target signal to obtain a first signal portion reflecting the smooth trend of the target signal and a second signal portion reflecting the sudden change of the target signal, and set a judgment threshold for identifying the sudden change of the target signal based on the energy change of the target signal;

[0103] Optionally, step S104 may specifically include the following steps:

[0104] S1041. The target signal is averaged to obtain a smoothed sequence, and the smoothed sequence is defined as the first signal part;

[0105] S1042. Subtract the first signal portion from the target signal to obtain a change difference sequence, and define the change difference sequence as the second signal portion;

[0106] S1043. Select multiple time periods consecutively on the target signal, calculate the cumulative sum of the squared values ​​of the signal values ​​in each time period, and define the cumulative sum of the squared values ​​as the local energy value.

[0107] S1044. Statistically calculate the highest and lowest energy levels of the local energy values ​​within all time periods, and generate a dynamic judgment threshold based on the highest and lowest energy levels.

[0108] Specifically, step S1044 includes the following processes: checking whether the local energy values ​​in each time period are complete and valid, excluding time periods containing missing signal points, and retaining complete and valid local energy values ​​to form an effective energy set; finding the three local energy values ​​with the largest values ​​in the effective energy set, and calculating the arithmetic mean of these three local energy values ​​as the highest energy level; finding the three local energy values ​​with the smallest values ​​in the effective energy set, and calculating the arithmetic mean of these three local energy values ​​as the lowest energy level; calculating the ratio of the highest energy level to the lowest energy level, and multiplying the ratio by a preset sensitivity coefficient to obtain a basic threshold value; calculating the dispersion value of the effective local energy values ​​in all time periods, multiplying the dispersion value by a preset fluctuation compensation factor to generate a dynamic compensation value, and correcting the basic threshold value according to the dynamic compensation value to generate a dynamic judgment threshold.

[0109] In the above scheme, the smoothed sequence refers to the low-frequency trend component of the target signal after moving average processing, which includes the long-term evolution characteristics of fluid pressure. Specifically, it is represented by the continuous output of the average voltage within a fixed time window, and can be used to extract the steady-state baseline of water depth pressure changes. The variation difference sequence refers to the instantaneous residual between the target signal and the smoothed sequence, which includes the transient disturbance characteristics of the pressure field. Specifically, it is represented by the algebraic difference between the original signal value and the smoothed value at the corresponding moment, and can be used to capture abrupt events such as turbulent impact and cavitation collapse. The local energy value refers to the intensity quantification index of signal fluctuations within a fixed time period, which includes the energy accumulation characteristics of short-term pressure disturbances. Specifically, it is represented by the cumulative sum of the squared voltage values ​​of all sampling points within the time period, and can be used to assess the transient activity level of the fluid environment. The highest / lowest energy level refers to the energy extreme value characterization quantity after robust statistical processing, which includes the energy boundary characteristics of non-stationary fluid background. Specifically, it is represented by the arithmetic mean of the three largest / smallest local energy values ​​in the effective energy set, and can be used to eliminate the interference of outliers on threshold calculation. Dynamic judgment threshold refers to the threshold for identifying sudden changes in the fluid environment. It includes a decision boundary that integrates energy statistical characteristics and fluctuation characteristics. Specifically, it is a linear combination of basic threshold value and discrete compensation value, which can be used to accurately distinguish between real transient events and background noise.

[0110] In this embodiment, the target signal is first processed by moving average in step S1041: a fixed-width sliding window is set, and the window is gradually moved from the starting point of the signal to calculate the signal mean value corresponding to the center point of each window. Boundary points are filled by mirroring; for example, when t=1, the value of t=2 is copied to fill t=0. The smoothed signal is defined as the first signal part. For example, in flood discharge monitoring, the target signal contains high-frequency turbulent noise with a fluctuation range of ±0.8V. After a 100-point moving average, a smoothed sequence is output with a fluctuation of <±0.05V. This sequence is defined as the first signal part reflecting the slow change in water level.

[0111] Next, in step S1042, the target signal and the smoothed sequence are aligned point by point and algebraic subtraction is performed. Abrupt components in the original signal are retained as significant peaks or troughs, while steady-state components are suppressed to near zero, resulting in an interpolated variation sequence, which is defined as the second signal component. Continuing the previous example: at the instant the floodgate opens, the target signal exhibits a 3.2V pulse (t=125s), at which point the smoothed value is only 2.4V. The difference sequence records a sudden change of +0.8V at this point, and this sequence is defined as the second signal component reflecting the transient event.

[0112] Next, in step S1043, the target signal is continuously divided into fixed-length time periods. The voltage values ​​of all sampling points within each complete time period are squared and summed to generate a local energy value sequence. For example, in the monitoring of lock discharge, sampling is performed at a sampling frequency of 100Hz within a 5-second time period. The voltage value fluctuates between [0.8V, 1.5V] within 5 seconds. The local energy value is calculated as 0.8² + 0.9² + ... + 1.5² = 635V²s, which quantifies the intensity of water flow pulsation during that time period.

[0113] Finally, as Figure 2 As shown, step S1044 first checks the data integrity of local energy values ​​for all time periods, excluding time periods containing missing sampling points. If a period is missing ≥5 points, it is discarded, and complete data is retained to form a valid energy set. Next, the three local energy values ​​with the largest values ​​in the valid set are selected, and their arithmetic mean is calculated as the highest energy level. At the same time, the three smallest values ​​are taken and their mean is calculated as the lowest energy level. Then, the basic threshold value is calculated using the following formula: Then, the standard deviation σ of the effective energy value is calculated and multiplied by a preset fluctuation compensation factor to generate a dynamic compensation value. Finally, a dynamic judgment threshold is generated by the linear superposition of the basic threshold value and the dynamic compensation value. For example, the effective energy set detected by cavitation monitoring in the turbine tailrace pipe is: [82,85,790,88,805]. The three largest local energy values ​​[790,805,88] are taken to calculate the highest energy level as 561; the three smallest local energy values ​​[82,85,88] are taken to calculate the lowest energy level as 85; the sensitivity coefficient is set to 0.8, and the calculated value is... The effective value standard deviation σ≈316.3, the compensation factor is 0.03, and the calculated dynamic compensation value is 316.3×0.03≈9.49; then the final dynamic judgment threshold is 5.28+9.49=14.77.

[0114] In practical applications, during pressure monitoring of the tailrace tube of a mixed-flow turbine, the target signal exhibits steady-state vortex band oscillations and cavitation burst pulses, with a normal range of 0.8-1.2V and sudden pulse bursts of 3-5V peaks. First, a 200-point moving average is applied to the target signal over a 2-second window to generate a smoothed sequence, i.e., the first signal portion, successfully extracting the 0.5Hz vortex band pressure oscillation baseline. Next, the smoothed sequence is subtracted from the original signal to obtain the difference sequence, i.e., the second signal portion. Then, the signal is divided into 0.5-second time intervals. Calculate the local energy values. The energy during a normal vortex band period is approximately 85 V²s, such as period 15: 1.1² × 50 ≈ 60.5. However, the energy during period t = 32.1s is as high as 1125 V²s (3.12² × 50 + background fluctuation ≈ 1125). Finally, generate a dynamic threshold: after removing two missing signal periods, take the maximum three values ​​[1125, 1108, 1050] of the effective energy set, with an average of 1094.3 V²s, and the minimum three values ​​[82, 85, 88], with an average of 85 V²s. The standard deviation σ is calculated to be 298.7, the dynamic compensation value is 298.7 × 0.05 ≈ 14.94, and the final generated threshold is 11.58 + 14.94 = 26.52. The sensitivity coefficient is 0.9 and the compensation factor is 0.05 during the calculation process.

[0115] The overall scheme of S104 described above uses the moving average technique to robustly separate the steady-state pressure baseline and transient pulse components from the target signal. Based on the energy statistics of the time period, it constructs an anti-disturbance extreme value model and an environmental adaptive compensation mechanism, and finally generates a dynamic judgment threshold. This achieves a highly robust abrupt event identification benchmark in non-stationary fluid environments, effectively overcoming the problem that fixed thresholds are insensitive to turbulent background fluctuations or have a lag in response to real events, laying the foundation for subsequent accurate quantification of transient characteristics.

[0116] S105. Using the judgment threshold, detect and mark the time interval in the second signal part where the sudden change occurs, and analyze the voltage signal in the time interval to determine the maximum amplitude of the fluctuation and the duration of the sudden change in the time interval as the first feature value of the water depth pressure data.

[0117] Optionally, step S105 may specifically include the following steps:

[0118] S1051. Continuously scan the signal amplitude value at each time point on the second signal part, and record the time point when the signal amplitude value first exceeds the judgment threshold as the start time point of the sudden change.

[0119] S1052. Continuous monitoring is performed starting from the mutation start time point. When the duration of the signal amplitude value being continuously lower than the judgment threshold reaches a preset stability threshold, it is recorded as the mutation termination time point.

[0120] S1053. Calculate the time difference between the mutation termination time and the mutation start time, and use the time difference as the mutation duration.

[0121] S1054. During the duration of the mutation, traverse all signal amplitude values ​​of the second signal portion and record the signal amplitude value with the largest absolute value as the maximum fluctuation amplitude.

[0122] S1055. Output the maximum amplitude of the fluctuation and the duration of the sudden change as the first feature value of the water depth pressure data.

[0123] In the above scheme, the signal amplitude value refers to the instantaneous voltage quantization value of the second signal part, including the polarity-independent intensity characteristics of the voltage pulse, and can be used to characterize the instantaneous intensity of transient pressure disturbances. The stability threshold refers to the sustained low-level time standard for determining the end of a transient event, including the inertial judgment characteristics against signal jitter, specifically manifested as the product of a preset number of consecutive low-level sampling points and the sampling interval, which can be used to prevent event missegmentation caused by noise fluctuations. The maximum fluctuation amplitude refers to the peak voltage intensity within the transient event time interval, including the maximum energy release characteristics of the pressure shock event, specifically manifested as the maximum absolute value of all signal amplitudes within the mutation duration, which can be used to assess the intensity level of fluid transient shocks. The mutation duration refers to the complete time span from the start to the end of the transient event, including the time-domain scale characteristics of pressure disturbance energy release, specifically manifested as the algebraic difference between the mutation termination time point and the start time point, which can be used to quantify the duration of fluid instability.

[0124] In this embodiment, the second signal portion, i.e., the mutation component sequence, is first time-series scanned in step S1051: starting from the beginning of the time axis, the absolute value of the signal amplitude is compared with the dynamic judgment threshold point by point. When the first exceeding the limit point is detected, that moment is immediately recorded as the mutation start time point. For example, in the monitoring of cavitation in a water turbine, a mutation is detected at t=32.15s. When the voltage exceeds the threshold of 2.8V, the system records the start time of the mutation. And then proceed to the next stage.

[0125] Secondly, based on the mutation initiation time point in step S1052... Continue scanning: When the detected signal amplitude is below the threshold, a stabilization timer is started. The preset stabilization threshold of 0.1 seconds corresponds to 10 consecutive sampling points. If the subsequent 10 consecutive points meet the threshold, the timer will continue to stabilize. If the value is less than or equal to the threshold, the current time is recorded as the termination point of the mutation. Continuing the previous example: the value first falls below the threshold at t=32.25s, and then for 10 consecutive points from t=32.26s to t=32.35s, the value is less than or equal to the threshold of 2.8V. Finally, the termination point of the mutation is recorded. If the limit is exceeded during the period, the timer will be reset immediately.

[0126] Next, based on the mutation initiation time point obtained in steps S1051 and S1052, step S1053 is used. With the mutation termination time point Calculate the time difference between start and end: The time unit is accurate to milliseconds, and the difference between the start and end times is used as the duration of the mutation. For example, the mutation start time point. mutation termination time point Then the difference between the start and end times Seconds. This value reflects the total duration of a transient event from its occurrence to its dissipation, such as the duration of a cavitation bubble collapse process.

[0127] Then, through step S1054, the mutation duration [ , Traverse all signal points within the inner loop: calculate the value at each sampling point. The absolute value of the signal is recorded, and the maximum value is taken as the maximum amplitude of the fluctuation. For example, the duration of the mutation [32.150s, 32.350s] contains 21 sampling points (including endpoints), and the signal value sequence is [+3.5V, +4.1V, -3.8V, +2.9V, -1.2V...], corresponding to absolute values ​​of [3.5, 4.1, 3.8, 2.9, 1.2...]. The maximum value is obtained by scanning. It appears at t=32.18s.

[0128] Finally, the maximum fluctuation amplitude is determined through step S1055. Duration of mutation Combined into ordered data pairs ( , As the first feature value output of the water depth pressure data, this feature value accurately quantifies the impact intensity and duration of cavitation events.

[0129] In practical applications, during the monitoring of Karman vortex shedding from turbine runner blades, when the axial-flow propeller turbine is operating under partial load, the second signal section detects periodic pressure pulsations: the scan begins at t=45.00s, and by t=45.20s, the signal amplitude reaches +1.1V, exceeding the dynamic threshold of 0.9V. 45.20s is marked as the start of the abrupt change. Although the voltage momentarily dropped to 0.85V at t=45.50s during continuous monitoring, falling below the threshold, it did not reach the preset 0.2-second stabilization threshold, so monitoring continued until t=45.25s when it exceeded the limit again. Finally, at t=47.70s, the signal was detected to be continuously below the threshold for 0.21 seconds, with 21 consecutive points less than or equal to the dynamic threshold of 0.9V, confirming 47.70s as the point where the sudden change terminated. ; Calculate duration Within the interval [45.20s, 47.70s], 250 sampling points were traversed to capture three vortex oscillation peaks: +1.8V at t=46.05s, +1.7V at t=46.55s, and +1.6V at t=47.15s. The maximum absolute value of 1.8V was taken as the maximum amplitude of the fluctuation, and the first characteristic value (1.8V, 2.50s) was output to accurately characterize the intensity and duration of the Karman vortex street.

[0130] The overall solution of S105 described above accurately locks the starting point of transient events through a dynamic threshold triggering mechanism and reliably determines the termination point by combining a continuous monitoring strategy with stable threshold constraints. It fully captures the event time boundary in the context of strong noise. Based on this boundary, it accurately calculates the duration and extracts the absolute peak value within the interval, outputting a quantitative combination of the maximum amplitude of fluctuation and the duration of the sudden change. This enables robust extraction of key features of fluid transient impact events and provides core indicators for equipment condition diagnosis that are not affected by background fluctuations.

[0131] S106. Utilizing the slow change characteristics contained in the first signal portion, predict the reference offset caused by long-term use of the voltage acquisition device, correct the first signal portion according to the reference offset, and output the maximum value, minimum value and fluctuation range of the target signal as the second feature value of the water depth pressure data based on the corrected first signal portion.

[0132] Optionally, step S106 may specifically include the following steps:

[0133] S1061. Divide the first signal portion into multiple consecutive time periods in chronological order, fit a linear trend in each time period, and connect the linear trends of all time periods to form a baseline offset prediction line.

[0134] S1062. Extract the predicted value at the same time point as the first signal part from the reference offset prediction line, and subtract the predicted value at the corresponding time point from the value of each point of the first signal part to generate the corrected first signal part.

[0135] S1063. On the corrected first signal section, scan all time points to record signal values, mark the highest signal value that appears during the scanning process as the maximum value, and mark the lowest signal value as the minimum value.

[0136] S1064. Calculate the arithmetic difference between the maximum and minimum values, and define the arithmetic difference as the fluctuation range;

[0137] S1065. Simultaneously output the three sets of values: maximum value, minimum value, and fluctuation range, as the second characteristic value of the water depth pressure data.

[0138] In the above scheme, the reference offset prediction line refers to the global drift trajectory formed by connecting segmented straight-line trends. It includes the continuous evolution characteristics of long-term sensor drift, specifically manifested as a broken line where the fitted line segments at each time period meet at their endpoints. It can be used to predict the reference voltage deviation at any given time. The predicted value refers to the voltage estimate at the corresponding time on the reference offset prediction line, including the time-domain mapping characteristics of the drift component, and can be used as a reference for signal correction. The corrected first signal part refers to the steady-state voltage sequence after eliminating the reference offset, including the fidelity characteristics of the actual water depth and pressure changes, and can be used to extract steady-state features without drift interference. The fluctuation range refers to the absolute value of the extreme difference of the corrected signal, including the amplitude characteristics of steady-state fluctuations in water depth and pressure. Specifically, it is manifested as the algebraic difference between the maximum and minimum values, and can be used to quantify the actual water level change amplitude within the monitoring period.

[0139] In this embodiment of the application, the first signal portion (i.e., the smoothed sequence) is first divided into continuous time periods of fixed duration, such as every 24 hours, through step S1061. The least squares method is used to fit a linear trend for the data points in each time period, as shown in the following formula: The slope and intercept The baseline offset prediction line is calculated by combining the time sequence and voltage values ​​of all sampling points within a given time period. The endpoints of the fitting lines for adjacent time periods are connected to form a continuous baseline offset prediction line. For example, in marine monitoring, the fitting line for the first day is V=0.0012t+2.5, and the fitting line for the second day is V=0.0015t+2.3. The two lines are connected at t=86400s (24h), and the endpoint value of the first segment, 106.1V, is used as the starting value of the second segment, 131.9V.

[0140] Secondly, in step S1062, predicted values ​​that completely correspond to the timestamps of each sampling point in the first signal portion are extracted from the reference offset prediction line. ; smooth the original value of the first signal portion Subtracting the predicted value generates the corrected first signal portion, calculated using the following formula: Corrected first signal portion For example, at t=129600s (36h): the original value 3.8V minus the predicted value 196.7V, the output correction value -192.9V retains the sign to reflect the actual relative change in water level.

[0141] Next, step S1063 involves sequentially scanning the entire time range of the corrected first signal portion: initializing the maximum value. negative infinity, minimum value As positive infinity; update point by point: when Updated regularly ,when Updated regularly For example, in tidal monitoring, data is captured at t=high tide. =1.2V, captured at t=low tide. =-0.8V.

[0142] Then, the maximum value obtained by the scan is processed in step S1064. and minimum value The formula for performing algebraic subtraction is as follows: The calculation result is defined as the fluctuation range. Note that sign operations should be preserved, for example... =1.2V, At -0.8V, This fluctuation range quantifies the actual water level change span after eliminating drift.

[0143] Finally, the maximum value is obtained through step S1065. Minimum value and fluctuation range The combination is a triplet data, in the following format: ( , , The second characteristic value of the water depth pressure data is output as (1.2V, -0.8V, 2.0V). For example, the output result (1.2V, -0.8V, 2.0V) corresponds to the highest water level, lowest water level, and total fluctuation range during the monitoring period, respectively.

[0144] In practical applications, during annual water level monitoring in gravity dam reservoirs, the daily water level baseline of the first signal, after moving average calculation, is divided into 12 monthly time periods. The drift trend for each month is fitted using the least squares method; for example, in month 3: V = 0.004t + 1.2, reflecting sensor temperature drift, and the 12 months are connected to form a baseline offset prediction line. At t = 90 days and 15:00, the original smoothed value... =3.5V minus the predicted value =0.004×7,776,000+1.2=31,105.2V, generating a correction value of -31,101.7V; Full-year scan correction signal: highest value during the high-water season. =2.8V, corresponding to t=212 days 08:00, the lowest value during the dry season. =-1.5V, corresponding to t=328 days 04:00; calculate the fluctuation range. =2.8-(-1.5)=4.3V; the final output is the second characteristic value (2.8V,-1.5V,4.3V), which accurately restores the true annual water level variation characteristics, with a difference of 12.3 meters between wet and dry seasons. The sensor coefficient of 2.8V / meter is verified to eliminate the influence of sensor zero drift, which accumulates to 35V throughout the year.

[0145] The overall scheme of S106 described above accurately predicts the long-term reference drift trend of the voltage acquisition device through piecewise linear modeling, eliminates sensor system errors through algebraic correction, extracts the true extreme values ​​and fluctuation ranges based on the corrected steady-state signal, and outputs water depth pressure steady-state characteristic values ​​that are resistant to drift interference. This fundamentally solves the problem of monitoring data distortion caused by equipment aging in traditional methods, and ensures that the measurement accuracy of water level extreme values ​​and fluctuation ranges remains at the sub-millimeter level during continuous observations over months or even years, providing a long-term reliable data benchmark for the safety assessment of large-scale water conservancy facilities.

[0146] The following is a complete example for steps 101-106, such as Figure 3 As shown, pressure sensors are deployed at three locations: the turbine top cover, the draft tube elbow, and the spiral casing inlet. Voltage signals are synchronously acquired at a sampling rate of 10kHz. Due to the dynamic load of the runner, the top cover sensor outputs high-frequency pulsations (fluctuating between 1.8-3.2V), the draft tube sensor captures 0.8Hz vortex band oscillations (2.5V±0.3V), and the spiral casing sensor experiences sudden zero-point drift due to silt wear, with the voltage continuously decreasing from t=12h onwards. The system automatically activates the backup sensor to replace the spiral casing node, and after sliding mean filtering and timestamp alignment, outputs a three-channel effective voltage sequence.

[0147] Next, sensor pairs were generated: top cover-tailpipe, top cover-volute, and tailpipe-volute. Cross-correlation analysis revealed the pressure wave propagation delay: 0.15s from top cover to tailpipe, corresponding to a water flow velocity of 12m / s over a distance of 1.8m; the volute signal was delayed by 0.08s due to positional differences. Using the top cover as a reference, the signals were aligned and superimposed to generate a composite signal. Fast Fourier Transform analysis showed significant peaks at a frequency of 12.5Hz (corresponding to 850V²) and a blade passage frequency of 62.5Hz (corresponding to 420V²), determining the dominant water flow fluctuation frequency to be 12.5Hz.

[0148] Next, the channel energy was calculated at the main frequency of 12.5Hz: the highest energy of the top cover sensor, 780V², was set as the reference. The calculated correlation strength of the tailrace pipe was 650×780=507,000, and the correlation strength of the volute was 700×780=546,000; the merging ratio was: top cover 0.38, tailrace pipe 0.32, and volute 0.30. The original values ​​at t=25.3s were 2.8V for the top cover, 2.5V for the tailrace pipe, and 2.6V for the volute. After weighted fusion, the target signal was obtained, calculated as: target signal = 2.8×0.38 + 2.5×0.32 + 2.6×0.30 = 2.65V.

[0149] Then, a 500-point moving average is used to extract the first signal component from the target signal; the difference is used to generate the second signal component. Local energy is calculated by dividing the signal into 0.1s intervals: the normal vortex band period is approximately 120V²s, and the cavitation burst period at t=183.2s reaches 2150V²s. The maximum three values ​​[2150, 1980, 1850] with an average of 1993V²s are taken as the effective set, and the minimum three values ​​[115, 118, 122] with an average of 118.3V²s are taken as the effective set, generating a dynamic threshold. The volatility compensation factor is 0.05, and the sensitivity coefficient is 0.8.

[0150] When t=183.21s, the second signal exceeds the threshold of 14.2, marking the starting point; monitoring continues until t=183.35s, satisfying a continuous 0.1s low level, marking the ending point; duration 0.14s; peak value within the interval. =6.3V, corresponding to the cavitation collapse point at t=183.25s; output the first eigenvalue (6.3V, 0.14s).

[0151] Finally, the first signal portion is segmented by month, and the drift lines for each month are fitted. For example, in the third month: V = 0.004t + 1.8. At noon on day 60: the original smoothed value 3.2V minus the predicted value 0.004 × 5,184,000 + 1.8 = 20,738V yields the correction value -20,734.8V. Full quarterly scan: maximum value during the high-water season. =4.1V, minimum value during the dry season =-0.9V; fluctuation range Output the second characteristic value (4.1V, -0.9V, 5.0V) to eliminate the sensor zero drift of 22V accumulated.

[0152] Figure 4 This is a schematic diagram of a specific implementation of a real-time calculation system for dynamic water depth pressure data features provided in this application, with reference to... Figure 4 The system may include:

[0153] Acquisition module 41 is used to acquire voltage signals from multiple channels, the voltage signals being used to reflect changes in water depth;

[0154] Analysis module 42 is used to determine the main frequency of water flow fluctuations by analyzing the correlation between voltage signals from multiple channels;

[0155] The generation module 43 is used to determine the signal merging ratio according to the main frequency, and to merge the voltage signals of multiple channels to generate the target signal using the signal merging ratio.

[0156] The identification module 44 is used to separate the target signal, obtain a first signal portion reflecting the smooth trend of the target signal and a second signal portion reflecting the sudden change of the target signal, and set a judgment threshold for identifying the sudden change of the target signal based on the energy change of the target signal;

[0157] The output module 45 is used to detect and mark the time interval in the second signal part by using the judgment threshold, and to analyze the voltage signal in the time interval to determine the maximum amplitude of the fluctuation and the duration of the sudden change in the time interval as the first feature value of the water depth pressure data.

[0158] The output module is also used to predict the reference offset caused by long-term use of the voltage acquisition device by utilizing the slow change characteristics contained in the first signal portion, and to correct the first signal portion according to the reference offset, and to output the maximum value, minimum value and fluctuation range of the target signal as the second feature value of the water depth pressure data based on the corrected first signal portion.

[0159] The real-time calculation system for dynamic water depth pressure data feature values ​​in this application is used to implement the aforementioned real-time calculation method for dynamic water depth pressure data feature values. Therefore, the specific implementation of the real-time calculation system for dynamic water depth pressure data feature values ​​can be found in the embodiment section of the real-time calculation method for dynamic water depth pressure data feature values ​​above. The specific implementation can be referred to the description of the corresponding embodiments, and will not be repeated here.

[0160] This application also provides an electronic device, comprising: a memory for storing a computer program; and a processor for executing the computer program to implement the steps of the real-time calculation method for dynamic data characteristic values ​​of water depth and pressure described above.

[0161] This application also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of any of the above-described methods for real-time calculation of dynamic water depth pressure data feature values.

[0162] In one exemplary embodiment, the aforementioned computer-readable storage medium may include, but is not limited to, various media capable of storing computer programs, such as USB flash drives, read-only memory, random access memory, portable hard drives, magnetic disks, or optical disks.

[0163] Embodiments of the present invention also provide a computer program product, which includes a computer program that, when executed by a processor, implements the steps in any of the embodiments of the real-time calculation method for dynamic water depth pressure data feature values.

[0164] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0165] The foregoing has provided a detailed description of the real-time calculation method, system, electronic device, and storage medium for dynamic water depth pressure data characteristic values ​​provided in this application. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the embodiments above are merely for the purpose of helping to understand the method and its core ideas. It should be noted that those skilled in the art can make various improvements and modifications to this application without departing from its principles, and these improvements and modifications also fall within the protection scope of this application.

Claims

1. A method for real-time calculation of dynamic feature values ​​of water depth pressure data, characterized in that, include: Voltage signals from multiple channels are acquired, and these voltage signals are used to reflect changes in water depth. By analyzing the correlation between voltage signals from multiple channels, the main frequency of water flow fluctuations can be determined. The signal merging ratio is determined based on the main frequency, and the voltage signals of multiple channels are merged using the signal merging ratio to generate the target signal. The target signal is separated to obtain a first signal portion reflecting the smooth trend of the target signal and a second signal portion reflecting the sudden change of the target signal, and a judgment threshold for identifying the sudden change of the target signal is set according to the energy change of the target signal; Using the aforementioned judgment threshold, the time intervals in which abrupt changes occur are detected and marked in the second signal portion, and the voltage signal within the time interval is analyzed to determine the maximum amplitude of the fluctuation and the duration of the abrupt change within the time interval as the first characteristic value of the water depth pressure data. By utilizing the slow-changing characteristics contained in the first signal portion, the reference offset caused by long-term use of the voltage acquisition device is predicted, and the first signal portion is corrected according to the reference offset. Based on the corrected first signal portion, the maximum value, minimum value and fluctuation range of the target signal are output as the second feature value of the water depth pressure data. The step of determining the signal combining ratio based on the main frequency and using the signal combining ratio to combine the voltage signals of multiple channels to generate the target signal includes: At the main frequency, the signal strength value of the voltage signal of each channel is determined, and the voltage signal with the highest signal strength value among all channels is selected as the reference voltage signal; For each channel voltage signal, the product of the signal strength value and the signal strength of the reference voltage signal at the main frequency is calculated, and the numerical result of the signal strength product is used as the correlation strength of the corresponding channel. Divide the correlation strength of each channel by the sum of the correlation strength values ​​of all channels to obtain the signal combining ratio of the corresponding channel; The voltage signal of each channel is multiplied by the corresponding signal merging ratio to generate a proportional adjustment signal, and all the proportional adjustment signals are arithmetically added together to obtain the target signal.

2. The method according to claim 1, characterized in that, The process of separating the target signal to obtain a first signal portion reflecting the smooth trend of the target signal and a second signal portion reflecting the abrupt change of the target signal, and setting a judgment threshold for identifying the abrupt change of the target signal based on the energy change of the target signal, includes: The target signal is averaged to obtain a smoothed sequence, which is defined as the first signal part. The first signal portion is subtracted from the target signal to obtain the change difference sequence, and the change difference sequence is defined as the second signal portion; Multiple time periods are continuously selected on the target signal, and the sum of the squared values ​​of the signal values ​​in each time period is calculated. The sum of the squared values ​​is defined as the local energy value. The highest and lowest energy levels of the local energy values ​​are statistically analyzed across all time periods, and a dynamic judgment threshold is generated based on the highest and lowest energy levels.

3. The method according to claim 2, characterized in that, The highest and lowest energy levels of the local energy values ​​are statistically analyzed across all time periods. A dynamic judgment threshold is then calculated based on these highest and lowest energy levels, including: Check whether the local energy values ​​for each time period are complete and valid, exclude time periods containing missing signal points, and retain complete and valid local energy values ​​to form a valid energy set; Find the three local energy values ​​with the largest values ​​in the effective energy set, and calculate the arithmetic mean of these three local energy values ​​as the highest energy level; Find the three local energy values ​​with the smallest values ​​in the effective energy set, and calculate the arithmetic mean of these three local energy values ​​as the lowest energy level; Calculate the ratio of the highest energy level to the lowest energy level, and multiply the ratio by a preset sensitivity coefficient to obtain the basic threshold value; Calculate the dispersion value of the effective local energy value over all time periods, multiply the dispersion value by a preset fluctuation compensation factor to generate a dynamic compensation value, and modify the basic threshold value according to the dynamic compensation value to generate a dynamic judgment threshold.

4. The method according to claim 1, characterized in that, The method of determining the main frequency of water flow fluctuations by analyzing the correlation between voltage signals from multiple channels includes: The voltage signals from multiple channels are combined in pairs to form voltage signal pairs, and each voltage signal pair contains a first channel signal and a second channel signal. For each voltage signal pair, the second channel signal is moved step by step on the time axis according to a preset step size. The matching degree value between the second channel signal and the first channel signal is calculated after each movement, and the number of movement steps corresponding to the peak value of the matching degree value is recorded as the time delay value of the voltage signal pair. Summarize the time delay values ​​of all voltage signal pairs and establish a set of time delay relationships; The starting positions of the voltage signals in different channels are adjusted according to the set of time delay relationships, so that the voltage signals of multiple channels are matched with each other on the time axis. The voltage signals of multiple channels after time matching are superimposed synchronously to form a comprehensive signal. Intensity analysis is performed on the comprehensive signal to identify the highest frequency point where the intensity value exceeds a preset threshold as the main frequency of water flow fluctuation.

5. The method according to claim 1, characterized in that, The step of using the judgment threshold to detect and mark the time intervals in the second signal portion where abrupt changes occur, and analyzing the voltage signal within the time interval to determine the maximum amplitude of the fluctuation and the duration of the abrupt change as the first feature value of the water depth pressure data, includes: The signal amplitude value at each time point is continuously scanned on the second signal part, and when the signal amplitude value first exceeds the judgment threshold, it is recorded as the time point of sudden change. Continuous monitoring begins from the mutation initiation time point. When the duration of the signal amplitude value continuously falling below the judgment threshold reaches a preset stability threshold, it is recorded as the mutation termination time point. Calculate the time difference between the mutation termination time and the mutation start time, and use this time difference as the mutation duration; During the duration of the mutation, all signal amplitude values ​​of the second signal portion are traversed, and the signal amplitude value with the largest absolute value is recorded as the maximum fluctuation amplitude. The maximum amplitude of the fluctuation and the duration of the sudden change are output as the first feature value of the water depth pressure data.

6. The method according to claim 1, characterized in that, The method of using the slowly changing characteristics contained in the first signal portion to predict the reference offset caused by long-term use of the voltage acquisition device, correcting the first signal portion according to the reference offset, and outputting the maximum value, minimum value, and fluctuation range of the target signal as the second feature value of the water depth pressure data based on the corrected first signal portion includes: The first signal portion is divided into multiple consecutive time periods in chronological order. A linear trend is fitted within each time period, and the linear trends of all time periods are connected to form a baseline offset prediction line. The predicted value at the same time point as the first signal part is intercepted on the reference offset prediction line, and the predicted value at the corresponding time point is subtracted from the value of each point of the first signal part to generate the corrected first signal part. In the corrected first signal section, the signal values ​​are recorded at all time points. The highest signal value that appears during the scanning process is marked as the maximum value, and the lowest signal value is marked as the minimum value. Calculate the arithmetic difference between the maximum and minimum values, and define this arithmetic difference as the fluctuation range; Simultaneously, the maximum value, minimum value, and fluctuation range are output as the second characteristic value of the water depth pressure data.

7. A real-time calculation system for dynamic data characteristic values ​​of water depth and pressure, characterized in that, include: The acquisition module is used to acquire voltage signals from multiple channels, and the voltage signals are used to reflect changes in water depth. The analysis module is used to determine the main frequency of water flow fluctuations by analyzing the correlation between voltage signals from multiple channels; The generation module is used to determine the signal merging ratio based on the main frequency, and to merge the voltage signals of multiple channels to generate a target signal using the signal merging ratio. The identification module is used to separate the target signal, obtain a first signal portion reflecting the smooth trend of the target signal and a second signal portion reflecting the sudden change of the target signal, and set a judgment threshold for identifying the sudden change of the target signal based on the energy change of the target signal; The output module is used to detect and mark the time interval in the second signal part where the sudden change occurs using the judgment threshold, and to analyze the voltage signal in the time interval to determine the maximum amplitude of the fluctuation and the duration of the sudden change in the time interval as the first feature value of the water depth pressure data. The output module is also used to predict the reference offset caused by long-term use of the voltage acquisition device by utilizing the slow change characteristics contained in the first signal portion, and to correct the first signal portion according to the reference offset, and to output the maximum value, minimum value and fluctuation range of the target signal as the second feature value of the water depth pressure data based on the corrected first signal portion. The step of determining the signal combining ratio based on the main frequency and using the signal combining ratio to combine the voltage signals of multiple channels to generate the target signal includes: At the main frequency, the signal strength value of the voltage signal of each channel is determined, and the voltage signal with the highest signal strength value among all channels is selected as the reference voltage signal; For each channel voltage signal, the product of the signal strength value and the signal strength of the reference voltage signal at the main frequency is calculated, and the numerical result of the signal strength product is used as the correlation strength of the corresponding channel. Divide the correlation strength of each channel by the sum of the correlation strength values ​​of all channels to obtain the signal combining ratio of the corresponding channel; The voltage signal of each channel is multiplied by the corresponding signal merging ratio to generate a proportional adjustment signal, and all the proportional adjustment signals are arithmetically added together to obtain the target signal.

8. An electronic device, characterized in that, include: Memory, used to store computer programs; A processor is configured to execute the computer program to implement the steps of the real-time calculation method for dynamic data feature values ​​of water depth pressure as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, enables the real-time calculation method for dynamic data characteristic values ​​of water depth and pressure as described in any one of claims 1 to 6.

Images