Hydraulic turbine set operation system and method based on digital twinning

Abstract

The invention relates to a hydraulic turbine set operation system and method based on digital twinning, and the system comprises a hydraulic turbine set top cover, the hydraulic turbine set top cover is provided with a plurality of groups of voiceprint collection arrays, and each group of voiceprint collection array comprises a plurality of voiceprint collectors which are arranged in the vertical direction at fixed intervals. The voiceprint collector is used for collecting original voiceprint data of different positions on the top cover of the hydraulic turbine set and sending the collected original voiceprint data to the data center. After receiving the original voiceprint data, the data center preprocesses the original voiceprint data, establishes a digital twinning reference model based on a mapping relation between real-time working condition parameters and voiceprint signal features through a digital twinning technology, and constructs an expected feature vector based on a voiceprint collector position compensation vector and a mapping matrix; and adjusting the difference vector by using the correction factor. The system adopts a plurality of directions and a plurality of voiceprint collectors configured in each direction, and the voiceprint collection accuracy is high in an extreme water flow state.

Description

Technical Field

[0001] The present invention relates to the technical fields of digital twin and hydroturbine unit control, and particularly relates to an operation system and method for a hydroturbine unit based on digital twin. Background Art

[0002] During the operation of a hydroturbine unit, there are still many difficulties in collecting acoustic fingerprints under extreme water flow conditions. Among them, under extremely low water flow conditions, the overall operation state of the hydroturbine unit is relatively "stable", the vibration and acoustic wave intensity generated by the equipment are relatively weak, the collected acoustic fingerprint signal may be in a low amplitude state, the signal-to-noise ratio (SNR) is low, and it is easily masked by background environmental noise. Therefore, cavitation and other anomalies are often accompanied by high-frequency pulse signals, but at low flow rates, these signals may be difficult to excite a typical high-frequency response, resulting in insufficient energy in the characteristic frequency band and unclear spectral characteristics, which is not conducive to subsequent discrimination. Under extremely high water flow conditions, the equipment vibration and liquid impact will generate very strong noise, which may cause the signal amplitude received by the acoustic fingerprint collector to be too high, and the sensor is prone to saturation, resulting in distortion. High-speed flow will cause a large amount of turbulent noise and random noise caused by water flow collision. The spectrum may be filled with various interference components, making it difficult to separate cavitation and other specific abnormal signals from it.

[0003] In addition, when high-frequency pulse noise is generated during cavitation, its attenuation, propagation time, and phase displacement are closely related to the distance between the sensor and the cavitation source. When the distance is relatively close, the signal attenuation is small, the amplitude is high, and the details are richer. Therefore, when the distance is far, there will be a certain attenuation and phase delay. The acoustic wave generated by cavitation may have a certain directionality, and the signal amplitude, spectral distribution, and phase information captured by collectors with different installation directions will be different. For example, a collector located in the main impact direction of cavitation is more likely to capture high-energy signals, while the signals of lateral and back-facing collectors may be relatively weak.

[0004] In summary, the acoustic fingerprint collection in the prior art under extreme water flow conditions needs to be further improved. Summary of the Invention

[0005] To solve the deficiencies in the prior art, the purpose of the present invention is to solve the above-mentioned defects, and further propose an operation system and method for a hydroturbine unit based on digital twin.

[0006] The present invention adopts the following technical solutions.

[0007] In the first aspect of the present invention, an operation system for a hydroturbine unit based on digital twin is disclosed. The system includes the top cover of the hydroturbine unit, and multiple groups of acoustic fingerprint collection arrays are arranged on the top cover of the hydroturbine unit. Each group of acoustic fingerprint collection arrays includes multiple acoustic fingerprint collectors arranged in the vertical direction with a fixed spacing.

[0008] The voiceprint collector is used to collect the original voiceprint data at different positions on the top cover of the water turbine unit, and send the collected original voiceprint data to the data center;

[0009] After receiving the original voiceprint data, the data center preprocesses the original voiceprint data, and establishes a digital twin benchmark model based on the mapping relationship between real-time working condition parameters and voiceprint signal characteristics through digital twin technology, and constructs an expected feature vector based on the position compensation vector and mapping matrix of the voiceprint collector, and adjusts the difference vector using a correction factor;

[0010] Among them, the actual working condition parameters include real-time water flow rate, water pressure and water temperature, and the difference vector is calculated based on the actual feature vector after preprocessing the original voiceprint data actually collected by the voiceprint collector and the expected feature vector.

[0011] The second aspect of the present invention discloses a method for operating a water turbine unit based on digital twin, which is implemented through the digital twin-based water turbine unit operation system described in the first aspect. The method includes:

[0012] Obtain multi-directional original voiceprint data and preprocess the original voiceprint data;

[0013] Establish a real-time operation model of the water turbine unit through digital twin technology, and generate an expected spectrum model according to the structure of the water turbine unit, fluid dynamics and historical voiceprint data;

[0014] Construct an expected feature vector based on the position of the voiceprint collector based on the position compensation vector and mapping matrix of the collector, and adjust the difference vector through a correction factor;

[0015] Among them, multiple groups of voiceprint collection arrays are arranged on the top cover of the water turbine unit. Each group of voiceprint collection arrays includes multiple voiceprint collectors arranged vertically with a fixed spacing. The preprocessing includes differential preprocessing, time-frequency domain decomposition, adaptive filtering and dynamic range adjustment.

[0016] Further, the obtaining of multi-directional original voiceprint data and the preprocessing of the original voiceprint data include:

[0017] Determine the preliminary installation strategy of the voiceprint collectors on the top cover of the water turbine unit based on the preprocessed original voiceprint data combined with the digital twin benchmark model parameters. The preliminary installation strategy includes multiple groups of voiceprint collection arrays and the number, arrangement direction and arrangement spacing of the voiceprint collectors on each group of voiceprint collection arrays;

[0018] Calibrate the voiceprint collector installed according to the preliminary installation strategy, set the sampling frequency, sensitivity, and dynamic range, and adjust the amplitude of the voiceprint collector using the real-time water flow rate to obtain calibration parameters;

[0019] According to the installation position and orientation of the voiceprint collector on the top cover of the water turbine unit, combined with the calibration parameters, adopt different compensation strategies based on the real-time water flow rate to adjust the amplitude normalization of the voiceprint collector signal, and perform noise reduction and dynamic range expansion processing on the voiceprint signal to obtain the original voiceprint data after differential preprocessing;

[0020] Synchronize the time of the original voiceprint data after differential preprocessing with the real-time water flow data and other operating conditions parameters, and perform joint data fusion to form a joint data matrix;

[0021] Among them, other operating conditions parameters include water flow pressure and water flow temperature.

[0022] Further, establishing a real-time operation model of the water turbine unit through digital twin technology, and generating an expected spectrum model according to the structure of the water turbine unit, fluid dynamics, and historical voiceprint data, including:

[0023] Segment the continuous voiceprint signals of each voiceprint collector according to a fixed time window to form discrete voiceprint signal segments, and calculate the adjustment factor of each voiceprint signal segment according to the real-time water flow rate, so as to perform gain adjustment on each voiceprint signal segment based on the adjustment factor to obtain a new voiceprint signal;

[0024] Perform time-frequency domain decomposition on the new voiceprint signal of each voiceprint collector to extract the spectral characteristics of the new voiceprint signal, and perform differential noise filtering according to the position of the voiceprint collector;

[0025] Extract the time-frequency matrix through short-time Fourier transform, and combine wavelet transform to obtain the local characteristics of the new voiceprint signal.

[0026] Further, establishing a real-time operation model of the water turbine unit through digital twin technology, and generating an expected spectrum model according to the structure of the water turbine unit, fluid dynamics, and historical voiceprint data, further includes:

[0027] Extract key features based on the preprocessed video data of each voiceprint collector, perform differential fusion on the key features, and use the compensation information of the real-time water flow rate to correct the energy difference of the voiceprint signals at different positions to obtain a fused feature vector;

[0028] Construct a digital twin reference model based on the fused feature vector and real-time operating conditions parameters to establish the mapping relationship between the real-time operating conditions parameters and the voiceprint signal, and construct a joint preprocessing dataset by integrating the preprocessed original voiceprint data and model parameters;

[0029] Among them, the key features include the energy of each frequency band and the local high-frequency pulse index.

[0030] Furthermore, constructing an expected feature vector based on the acoustic fingerprint collector position based on the collector position compensation vector and the mapping matrix, and adjusting the difference vector through a correction factor, includes:

[0031] Based on the digital twin benchmark model, by combining the real-time working condition parameters and the acoustic fingerprint collector position parameters, constructing the expected feature vector of each acoustic fingerprint collector, and integrating the real-time water flow rate as a mapping parameter into the digital twin benchmark model to construct a position-dependent mapping formula;

[0032] Comparing the actual feature vector corresponding to the preprocessed acoustic fingerprint data collected by each acoustic fingerprint collector with the expected feature vector to calculate the difference vector between the actual feature vector and the expected feature vector, and correcting the difference vector according to the change of the real-time water flow rate.

[0033] Furthermore, constructing an expected feature vector based on the acoustic fingerprint collector position based on the collector position compensation vector and the mapping matrix, and adjusting the difference vector through a correction factor, also includes:

[0034] Calculating the L2 norm based on the corrected difference vector as the anomaly score of the acoustic fingerprint collector, and combining the azimuth and height of the acoustic fingerprint collector array, and using weighted average to determine whether there is cavitation anomaly in the water turbine unit;

[0035] Generating a cavitation anomaly determination report based on the anomaly score according to the preset determination rule, and feeding back the cavitation anomaly determination report to the monitoring system, and at the same time updating the digital twin benchmark model parameters based on the abnormal acoustic fingerprint data.

[0036] Furthermore, the method also includes:

[0037] Generating a basic data set based on the anomaly score and the spatial position information of each acoustic fingerprint collector, performing weighted average on the basic data set to obtain the first cavitation position, and combining auxiliary physical parameters to correct the components in each direction of the first cavitation position to obtain the second cavitation position;

[0038] Obtaining the actual cavitation position, and when the actual cavitation position is inconsistent with the second cavitation position, calculating the first deviation vector between the actual cavitation position and the second cavitation position to correct the second cavitation position to obtain the third cavitation position;

[0039] Calculate the second deviation vector between the third cavitation position and the corresponding acoustic fingerprint collector, and adjust the positions of the acoustic stripe sensors in the horizontal and vertical directions by combining the abnormal score and the current spatial position information of each acoustic fingerprint collector;

[0040] Among them, the basic data set is used to locate the cavitation position, including an abnormal score data set and a sensor position data set, and the auxiliary physical parameters include a vibration abnormality index and a temperature abnormality index in the area where the acoustic fingerprint collector is located.

[0041] The third aspect of the present invention discloses a terminal, including a processor and a storage medium; characterized in that:

[0042] The storage medium is used to store instructions;

[0043] The processor is used to operate according to the instructions to execute the steps of the method described in the second aspect.

[0044] The fourth aspect of the present invention discloses a computer-readable storage medium, on which a computer program is stored, characterized in that when the program is executed by a processor, the steps of the method described in the second aspect are implemented.

[0045] The beneficial effects of the present invention:

[0046] Multi-directional coverage and multi-level acoustic fingerprint data acquisition of the water turbine unit are achieved by using multiple directions and multiple acoustic fingerprint collectors configured in each direction, so that acoustic fingerprint collectors in different directions and at different heights can capture acoustic fingerprint signals from the water turbine unit. In addition, due to different positions of each acoustic fingerprint collector, the amplitude of the acoustic fingerprint signal received by it will vary due to different distances and directions. If a certain direction is closer to the main cavitation area, it will be more obvious in features such as high-frequency pulses and phase offsets. Due to the different relative positions between the acoustic fingerprint collector and the cavitation source, different propagation delays and phase differences will occur, and the accuracy of acoustic fingerprint acquisition under extreme water flow conditions is relatively high. Description of the Drawings

[0047] Figure 1 It is a schematic structural diagram of a water turbine unit operation system based on digital twin;

[0048] Figure 2 It is a flowchart of a water turbine unit operation method based on digital twin. Detailed Embodiments

[0049] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0050] As Figure 1 shown, in one embodiment, a digital twin-based operation system for a water turbine unit includes a top cover of the water turbine unit. Multiple groups of acoustic fingerprint acquisition arrays are arranged on the top cover of the water turbine unit. Each group of acoustic fingerprint acquisition arrays includes multiple acoustic fingerprint collectors arranged in the vertical direction with a fixed spacing.

[0051] The acoustic fingerprint collectors are used to collect the original acoustic fingerprint data at different positions on the top cover of the water turbine unit and send the collected original acoustic fingerprint data to the data center.

[0052] After receiving the original acoustic fingerprint data, the data center preprocesses the original acoustic fingerprint data, and establishes a digital twin reference model based on the mapping relationship between real-time working condition parameters and acoustic fingerprint signal characteristics through digital twin technology, and constructs an expected feature vector based on the position compensation vector and mapping matrix of the acoustic fingerprint collectors, and adjusts the difference vector using a correction factor.

[0053] It should be noted that the expression of the digital twin reference model is:

[0054] F = M * X + ∈

[0055] In the formula, F is the fused preprocessed feature vector, which contains the energy information of each frequency band; M is the model parameter matrix, and each element in the matrix represents the influence coefficient of any working condition parameter on the preprocessed feature vector; X is the working condition parameter vector, which is composed of the water flow rate, water flow pressure, and water flow temperature; ∈ is the model error vector, and each element in the vector represents the prediction residual.

[0056] A position compensation vector is set for each acoustic fingerprint collector. This vector includes the azimuth where the corresponding acoustic fingerprint collector is located (encoded numerical values, such as east = 1, south = 2), the vertical position (lower layer, middle layer, upper layer), and local environmental factors (such as structural occlusion, openness, such as the value range is 0 to 1, 1 means no occlusion, 0 means severe occlusion, and the intermediate value means partial occlusion).

[0057] The correction factor adjusts the difference vector according to the change of the current real-time water flow rate, and its expression is:

[0058] Δ_i_corrected = Δ_i / (1 + α * (Q - Q_ref))

[0059] Wherein, Δ_i_corrected is the correction factor, Δ_i is the difference vector, Q is the current real-time water flow rate, Q_ref is the reference water flow rate, and α is the water flow influence coefficient, with a range of 0.1 to 0.5.

[0060] Among them, the actual working condition parameters include the real-time water flow rate, water pressure, and water temperature. The difference vector is calculated based on the actual feature vector and the expected feature vector after preprocessing the original voiceprint data actually collected by the voiceprint collector.

[0061] Such as Figure 2 shown, in one embodiment, a method for operating a water turbine unit based on digital twin includes the following steps:

[0062] Step S110, obtain the original voiceprint data in multiple directions and preprocess the original voiceprint data.

[0063] Among them, multiple groups of voiceprint acquisition arrays are arranged on the top cover of the water turbine unit. Each group of voiceprint acquisition arrays includes multiple voiceprint collectors arranged in the vertical direction with fixed spacing. The preprocessing includes differential preprocessing, time-frequency domain decomposition, adaptive filtering, and dynamic range adjustment.

[0064] In some embodiments, for the method for operating a water turbine unit based on digital twin provided by the present invention, step S110 specifically includes the following steps:

[0065] Step S111, determine the preliminary installation strategy of the voiceprint collectors on the top cover of the water turbine unit based on the preprocessed original voiceprint data combined with the digital twin benchmark model parameters. The preliminary installation strategy includes multiple groups of voiceprint acquisition arrays and the number, arrangement direction, and arrangement spacing of the voiceprint collectors on each group of voiceprint acquisition arrays.

[0066] Step S112, calibrate the voiceprint collectors installed according to the preliminary installation strategy, set the sampling frequency, sensitivity, and dynamic range, and adjust the amplitude of the voiceprint collectors using the real-time water flow rate to obtain the correction parameters.

[0067] In some embodiments, for the method for operating a water turbine unit based on digital twin provided by the present invention, step S110 specifically further includes the following steps:

[0068] Step S113, according to the installation position and orientation of the voiceprint collectors on the top cover of the water turbine unit, combined with the correction parameters, adjust the amplitude normalization of the voiceprint collector signals using different compensation strategies based on the real-time water flow rate, and perform noise reduction and dynamic range expansion processing on the voiceprint signals to obtain the preprocessed original voiceprint data after differentiation.

[0069] Step S114: Synchronize the time and perform joint data fusion on the pre - processed original voiceprint data with the real - time water flow data and other operating condition parameters to form a joint data matrix;

[0070] Among them, other operating condition parameters include water flow pressure and water flow temperature.

[0071] Step S120: Establish a real - time operation model of the water turbine unit through digital twin technology, and generate an expected spectrum model based on the structure of the water turbine unit, fluid dynamics, and historical voiceprint data.

[0072] It should be noted that Fluid Dynamics in this example focuses on studying the laws of water flow movement and its interaction with the surrounding water turbine unit structure, following the laws of mass conservation, momentum conservation, and energy conservation. Historical voiceprint data is the voiceprint data that has been recorded and stored before the original voiceprint data currently collected by the voiceprint collector. The expected spectrum model is used to characterize the mapping relationship between real - time operating condition parameters and voiceprint signals, and its expression is:

[0073] F_expected,i = M_i * X + B_i * L_i

[0074] In the formula, F_expected,i is the expected feature vector of the i - th voiceprint collector in n frequency bands; M_i is the mapping matrix related to the operating condition parameter X, corresponding to the i - th voiceprint collector, and each of its elements represents the influence coefficient of the operating condition parameter on the feature; B_i is the position compensation matrix, composed of the position compensation vector L_i, corresponding to the i - th voiceprint collector; both X and L_i are represented as column vectors.

[0075] In some embodiments, for the method for operating a water turbine unit based on digital twin provided by the present invention, step S120 specifically includes the following steps:

[0076] Step S121: Segment the continuous voiceprint signals of each voiceprint collector according to a fixed time window to form discrete voiceprint signal segments, and calculate the adjustment factor for each voiceprint signal segment according to the real - time water flow rate, so as to perform gain adjustment on each voiceprint signal segment based on the adjustment factor to obtain new voiceprint signals.

[0077] It should be noted that the algorithm expression for calculating the adjustment factor for each voiceprint signal segment according to the real - time water flow rate is:

[0078] G = {

[0079] G_low, if Q ≤ Q_low

[0080] 1, if Q_low < Q < Q_high

[0081] G_high, Q ≥ Q_high

[0082] }

[0083] Among them, Q represents the real-time water flow rate, Q_low represents the low flow rate threshold, Q_high represents the high flow rate threshold, G_low represents the gain factor for extremely low flow rates, and G_high represents the attenuation factor for extremely high flow rates.

[0084] Combining the above expressions, the expression of the new voiceprint signal after gain adjustment is:

[0085] x_adj(n) = G * x(n)

[0086] In the formula, x_adj(n) is the new voiceprint signal after gain adjustment, x(n) is the original voiceprint signal segment of the nth segment before gain adjustment, and G is the adjustment factor.

[0087] Step S122: Perform time-frequency domain decomposition on the new voiceprint signals of each voiceprint collector to extract the spectral features of the new voiceprint signals, and perform differential noise filtering according to the positions of the voiceprint collectors.

[0088] Step S123: Extract the time-frequency matrix through short-time Fourier transform, and obtain the local features of the new voiceprint signal in combination with wavelet transform.

[0089] In some embodiments, for the method for operating a water turbine unit based on digital twin provided by the present invention, step S120 specifically further includes the following steps:

[0090] Step S124: Extract key features based on the preprocessed video data of each voiceprint collector, perform differential fusion on the key features, and correct the energy difference of the voiceprint signals at different positions using the compensation information of the real-time water flow rate to obtain a fused feature vector.

[0091] Step S125: Construct a digital twin reference model based on the fused feature vector and real-time operating conditions parameters to establish a mapping relationship between the real-time operating conditions parameters and the voiceprint signals, and construct a joint preprocessing dataset by integrating the preprocessed original voiceprint data and model parameters.

[0092] Among them, the key features include the energy of each frequency band and the local high-frequency pulse index.

[0093] It should be noted that the expression of the digital twin reference model is:

[0094] F = M * X + ∈

[0095] In the formula, F is the preprocessed feature vector after fusion, which contains the energy information of each frequency band; M is the model parameter matrix, and each element in the matrix represents the influence coefficient of any working condition parameter on the preprocessed feature vector; X is the working condition parameter vector, which is jointly composed of the water flow rate, water flow pressure, and water flow temperature; ∈ is the model error vector, and each element in the vector represents the prediction residual.

[0096] The expression of the mapping relationship between the real-time working condition parameters and the voiceprint signal is:

[0097] F_expected,i = M_i * X + B_i * L_i

[0098] In the formula, F_expected,i is the expected feature vector of the voiceprint collector i in n frequency bands; M_i is the mapping matrix related to the working condition parameter X, corresponding to the voiceprint collector i, and each of its elements represents the influence coefficient of the working condition parameter on the feature; B_i is the position compensation matrix, which is composed of the position compensation vector L_i, corresponding to the voiceprint collector i; both X and L_i are represented as column vectors.

[0099] It should be further noted that the energy of each frequency band is the signal energy intensity of each segment of the voiceprint signal after segmentation, and the local high-frequency pulse index is the preset typical cavitation signal frequency band. The joint preprocessing dataset S is expressed as:

[0100] S = {F_preprocessed, M, ∈}

[0101] In the formula, F_preprocessed represents the set of preprocessed feature vectors after joint fusion in all directions, and M and ∈ are respectively the parameters and error vectors of the digital twin reference model.

[0102] Step S130: Construct the expected feature vector based on the collector position compensation vector and the mapping matrix, and adjust the difference vector through the correction factor.

[0103] In some embodiments, for the method for operating a water turbine unit based on digital twin provided by the present invention, step S130 specifically includes the following steps:

[0104] Step S131: Based on the digital twin reference model, by combining the real-time working condition parameters and the voiceprint collector position parameters, construct the expected feature vector of each voiceprint collector, and incorporate the real-time water flow rate as a mapping parameter into the digital twin reference model to construct a position-dependent mapping formula.

[0105] Step S132: Compare the actual feature vectors corresponding to the voiceprint data actually collected by each voiceprint collector after preprocessing with the expected feature vectors to calculate the difference vector between the actual feature vector and the expected feature vector, and correct the difference vector according to the change of the real-time water flow rate.

[0106] It should be noted that each element in the actual feature vector represents the voiceprint data obtained after preprocessing the voiceprint data actually collected by each voiceprint collector. The expression for correcting the difference vector according to the change of the real-time water flow rate is:

[0107] Δ_i_corrected = Δ_i / (1 + α * (Q - Q_ref))

[0108] In the formula, Δ_i_corrected is the correction factor, Δ_i is the difference vector of the i-th voiceprint collector, Q is the current real-time water flow rate, Q_ref is the preset reference water flow rate, and α is the water flow influence coefficient, with a range of 0.1 to 0.5.

[0109] In some embodiments, for the method for operating a water turbine unit based on digital twin provided by the present invention, step S130 specifically further includes the following steps:

[0110] Step S133: Calculate the L2 norm based on the corrected difference vector as the anomaly score of the voiceprint collector, and combine the azimuth and height where the voiceprint collector array is located to determine whether there is cavitation anomaly in the water turbine unit by weighted average.

[0111] For example, for i collectors (i = 3) in the same azimuth, weighted average is used, and the expression is:

[0112] S_d = (w_1 * S_{1,\text{norm}} + w_2 * S_{2,\text{norm}} + w_3 * S_{3,\text{norm}})

[0113] In the formula, S_d represents the result after weighted average, w_i represents the weight of the i-th voiceprint collector in this azimuth, which is determined according to the installation position (upper, middle, lower layers) of the voiceprint collector. For example, the upper layer signal may be more sensitive and can be given a higher weight; S_{i,\text{norm}} represents the anomaly score after standardization of the i-th voiceprint collector.

[0114] Step S134: Generate a cavitation anomaly determination report based on the anomaly score according to the preset determination rule, and feedback the cavitation anomaly determination report to the monitoring system. At the same time, update the parameters of the digital twin reference model based on the abnormal voiceprint data.

[0115] In some embodiments, the hydraulic turbine unit operation method based on digital twin provided by the present invention further includes the following steps:

[0116] Step S210, generating a basic data set based on the anomaly score and the spatial position information of each voiceprint collector, performing weighted averaging on the basic data set to obtain a first cavitation position, and correcting the components of the first cavitation position in each direction in combination with auxiliary physical parameters to obtain a second cavitation position.

[0117] Step S220, obtaining the actual cavitation position, and when the actual cavitation position is inconsistent with the second cavitation position, calculating the first deviation vector between the actual cavitation position and the second cavitation position to correct the second cavitation position to obtain a third cavitation position.

[0118] Step S230, calculating the second deviation vector between the third cavitation position and the corresponding voiceprint collector, and adjusting the position of each voiceprint bar sensor in the horizontal and vertical directions in combination with the abnormality score and the current spatial position information of each voiceprint collector.

[0119] It should be noted that the first cavitation position is the preliminary cavitation position obtained by weighted averaging the basic data set, the second cavitation position is the cavitation position obtained after correcting the components of the preliminary cavitation position in combination with the auxiliary physical parameters, and the third cavitation position is the cavitation position obtained after correcting the second cavitation position based on the deviation vector between the actual cavitation position and the second cavitation position.

[0120] Among them, the basic data set is used to locate the cavitation position, including the anomaly score data set and the sensor location data set. The auxiliary physical parameters include the vibration anomaly index and temperature anomaly index of the area where the voiceprint collector is located.

[0121] The above-mentioned method for operating a hydro-turbine unit based on digital twins uses multiple directions and multiple soundprint collectors configured in each direction to achieve multi-directional coverage and multi-level soundprint data collection for the hydro-turbine unit, so that soundprint collectors in different directions and at different heights can capture soundprint signals from the hydro-turbine unit. In addition, due to the different positions of each soundprint collector, the amplitude of the soundprint signal received by it will vary with the distance and direction. If a certain direction is closer to the main cavitation occurrence area, it will be more obvious in high-frequency pulses, phase shifts and other characteristics. Due to the different relative positions between the soundprint collector and the cavitation source, different propagation delays and phase differences will occur, and the accuracy of soundprint collection under extreme water flow conditions is higher.

[0122] In a specific embodiment, the hydraulic turbine unit operation method based on digital twin provided by the present invention includes steps 1 to 5:

[0123] Step 1: Multi-dimensional voiceprint data collection.

[0124] Arrange at least 4 groups of acoustic fingerprint acquisition arrays at four main orientations (east, south, west, north) of the top cover of the water turbine unit. Each group of arrays contains three acoustic fingerprint acquisition devices (SC1, SC2, SC3), which are arranged in sequence in the vertical direction, and the spacing is set to d1 (such as 0.5 - 1.0 meters).

[0125] Targeted preprocessing to reduce the risk of misjudgment: According to the variation characteristics of the signal amplitude under different water flow states, perform differential preprocessing on the original acoustic fingerprint data, so as to distinguish normal transient signals from abnormal cavitation signals during drastic changes in working conditions (such as startup, shutdown, load mutation), and reduce the misjudgment rate. Also, through adaptive filtering and dynamic range adjustment, improve the influence of multi-source noise and signal interference on the spectral characteristics, and ensure that subsequent analysis relies on high-quality data input.

[0126] Specifically, it includes steps 1.1 to 1.4:

[0127] Step 1.1, array arrangement and installation of acquisition devices.

[0128] Specifically, taking the preprocessed original acoustic fingerprint data and the parameters of the digital twin reference model as the expected signal reference, determine the installation strategy of the acoustic fingerprint acquisition devices on the top cover of the water turbine unit, that is, it is required to arrange a group of acoustic fingerprint acquisition device arrays at each of the four main orientations of the east, south, west, and north of the top cover of the water turbine unit. Each group of acoustic fingerprint acquisition device arrays contains three acoustic fingerprint acquisition devices, and these three acoustic fingerprint acquisition devices are installed from top to bottom along the vertical direction, and the installation spacing between each column of acoustic fingerprint acquisition devices is the same. In addition, considering that the acoustic fingerprint signals received by the acoustic fingerprint acquisition devices at different positions may vary under the influence of water flow, it is necessary to ensure that the installation positions of the acoustic fingerprint acquisition devices cover the areas where cavitation and other abnormalities may occur.

[0129] For example, if the vertical length of the acquisition area on the top cover of the water turbine unit is 1.2 meters, then the installation spacing between each column of acoustic fingerprint acquisition devices is 0.6 meters. In the east array, if the acoustic fingerprint acquisition device 1 in the first array is installed at the lowest position, that is, at the 0-meter position, then the acoustic fingerprint acquisition device 2 above it is installed at the 0.6-meter position, and the topmost acoustic fingerprint acquisition device 3 is installed at the 1.2-meter position.

[0130] Step 1.2, calibration of acoustic fingerprint acquisition devices and preliminary acquisition of real-time water flow data.

[0131] Specifically, calibrate each installed acoustic fingerprint acquisition device, and set the sampling frequency, sensitivity, and dynamic range. First, use the real-time water flow rate to preliminarily adjust the amplitude of the acquisition device to ensure that the measured acoustic fingerprint signals are not distorted under extremely low and extremely high water flow states, and provide accurate calibration parameters for subsequent preprocessing.

[0132] Among them, the original voiceprint signal can be expressed as:

[0133] Signal(t) = A * cos(2πft + φ)

[0134] In the formula, Signal(t) represents the original voiceprint signal at time t, A is the amplitude of the voiceprint signal, which is affected by the water flow state, f is the signal frequency, t is the time, and φ is the initial phase.

[0135] It should be noted that the real-time water flow rate is provided by the flow meter, and its influence on the signal amplitude is related to the voiceprint acquisition sensitivity compensation coefficient.

[0136] Step 1.3, differential preprocessing based on the position and orientation of the voiceprint collector.

[0137] Due to the different positions (height, orientation) of the voiceprint collectors, there will be differences in the original voiceprint signals they collect when affected by the water flow state:

[0138] The voiceprint collector located in the upper layer may receive stronger water flow impact signals, with a higher signal amplitude and more obvious high-frequency components in the spectrum.

[0139] The voiceprint collector located in the lower layer may be affected by local structures and reflections, resulting in greater signal attenuation.

[0140] Therefore, during preprocessing, different compensation strategies need to be adopted according to the installation position and orientation of the voiceprint collectors, adjust the amplitude normalization of the signals of each voiceprint collector based on the real-time water flow rate, and perform noise reduction and dynamic range expansion processing on the voiceprint signals.

[0141] For the i-th voiceprint collector, the preprocessed amplitude A_i is defined as:

[0142] A_i = k_i * Q + C_i

[0143] In the formula, k_i is the sensitivity compensation coefficient of the i-th voiceprint collector, obtained by calibration, Q is the real-time water flow rate, and C_i is the position compensation constant, which is used to reflect the influence of the voiceprint collector by structural occlusion and multipath effects, and usually takes a value range of 0 to 0.1V.

[0144] Step 1.4, data synchronization and construction of the joint data matrix.

[0145] Specifically, synchronize the time of the original voiceprint data preprocessed by each voiceprint collector with the real-time water flow rate and other working condition parameters (water flow pressure, water flow temperature) and perform joint data fusion to form a joint data matrix, providing high-precision input for subsequent digital twin modeling and anomaly detection.

[0146] Among them, the combined data matrix is composed of the preprocessed signal, water flow rate, water flow pressure, water flow temperature, and discrete sampling time collected by the voiceprint collector at any discrete time point.

[0147] Step 2, signal preprocessing and digital twin environment modeling.

[0148] The collected original voiceprint data is decomposed in the time-frequency domain by using the short-time Fourier transform (STFT) and wavelet transform. At the same time, adaptive noise reduction and amplitude normalization are performed in combination with the water flow state parameters (extremely low and extremely high water flows). At the same time, a real-time operation model of the water turbine unit is established through digital twin technology, and an expected spectrum model is generated based on the structure of the water turbine unit, fluid dynamics, and historical data, providing a benchmark for subsequent anomaly identification.

[0149] Different window lengths and filtering strategies are used to process the upper, middle, and lower three collectors respectively, compensating for the signal feature differences caused by the installation position and orientation, and improving the data consistency and accuracy. In addition, a mapping relationship is established between the real-time working condition parameters and the signal features. By introducing the model parameter matrix and the position compensation matrix, the digital twin model can adapt to the signal changes under different voiceprint collector positions and working conditions.

[0150] Specifically, it includes steps 2.1 to 2.4:

[0151] Step 2.1, dynamic signal segmentation and gain adjustment based on the influence of real-time water flow rate.

[0152] Specifically, when the original voiceprint signals collected by the voiceprint collectors at different positions are under different water flow rates, their amplitudes will also change significantly. To avoid the situation of insufficient amplitude and unsaturated amplitude of the voiceprint signal caused by extreme water flow states, dynamic segmentation is performed on each voiceprint collector, and gain adjustment is performed based on the real-time water flow rate.

[0153] During the voiceprint signal segmentation process, the continuous voiceprint signals collected by each voiceprint collector are segmented according to a fixed time window to form discrete signal segments. Then, the adjustment factor of each voiceprint signal segment is calculated according to the real-time water flow rate. Among them, the adjustment factor is jointly determined by the real-time water flow rate, the low water flow rate threshold, the high water flow rate threshold, the gain factor for extremely low water flows, and the attenuation factor for extremely high water flows. After the gain adjustment of each voiceprint signal segment, a new voiceprint signal can be obtained.

[0154] For example, when a certain voiceprint collector is under an extremely low water flow rate of 0.2 m 3 / s, assuming the gain factor is 4 and the amplitude mean of the original voiceprint signal segment is 0.05 V, the adjusted new voiceprint signal is about 0.2 V, ensuring that the low-amplitude signal is amplified.

[0155] Step 2.2, Time-frequency domain decomposition and differential noise filtering.

[0156] Specifically, perform time-frequency domain decomposition on the voiceprint signals of each voiceprint collector to extract its spectral features, and adopt a differential noise filtering strategy according to the position of the voiceprint collector (upper layer, middle layer, lower layer). Among them, the upper-layer voiceprint collector is impacted by direct current water, with richer high-frequency information; the lower-layer voiceprint collector may be affected by structural damping, with greater noise interference. Therefore, use the Short-Time Fourier Transform (STFT) to extract the time-frequency matrix, and then combine wavelet transform to obtain local detail features.

[0157] During the Short-Time Fourier Transform process, adjust the window length according to the installation position of the voiceprint collector, that is, use a short window length for the upper-layer voiceprint collector to improve time resolution; use a long window length for the lower-layer voiceprint collector to enhance spectral resolution. Apply the Short-Time Fourier Transform to each new voiceprint signal segment to obtain the time-frequency matrix, then use the discrete wavelet transform to extract local features, and filter out low-energy noise according to the set noise threshold (which can be determined according to the previously measured background noise energy).

[0158] Step 2.3, Feature extraction and fusion based on the position-specific characteristics of the voiceprint collector.

[0159] Specifically, different voiceprint collectors cause differences in position and orientation, and their time-frequency characteristics are different. Extract key features (such as energy in each frequency band, local high-frequency pulse index, etc.) from the preprocessed time-frequency data of each voiceprint collector, and perform differential fusion. Use the compensation information of the real-time water flow rate to further correct the energy difference between voiceprint signals at different positions, laying a foundation for constructing a unified digital twin benchmark model.

[0160] Among them, considering position compensation, a normalization vector needs to be constructed, which is obtained based on historical data statistics combined with position-specific energy bias compensation.

[0161] During the feature extraction and fusion process, for each voiceprint collector, extract the energy in different frequency bands from the time-frequency matrix and local features to form a preliminary feature vector. Combine the real-time water flow rate and the calibration parameters of each voiceprint collector to calculate the normalized feature vector. For multiple voiceprint collectors in the same orientation, use weighted average and principal component analysis methods to fuse their respective normalized feature vectors to form a unified feature vector for that orientation.

[0162] Step 2.4, Construct a digital twin benchmark model and output a joint preprocessing dataset.

[0163] Specifically, based on the azimuth feature vectors extracted and fused in Step 2.3, combined with real-time working condition parameters (water flow rate, water pressure, water temperature) and the jointly preprocessed dataset composed of the joint data matrix, a digital twin benchmark model is constructed to establish the mapping relationship between the recommended working condition parameters and the acoustic signal features. Subsequently, the preprocessed original acoustic data and model parameters are integrated to construct a jointly preprocessed dataset, which serves as the high-precision input for subsequent anomaly detection and working condition adjustment.

[0164] Among them, the working condition parameter vector is jointly composed of the real-time water flow rate, water pressure, and water temperature. The constructed digital twin benchmark model is jointly composed of the fused preprocessed feature vectors, that is, the energy information of each frequency band, the model parameter matrix, that is, the influence coefficient of the time-frequency feature element of each element on the local feature element, and the model error vector, that is, the prediction residual represented by each element. The jointly preprocessed dataset is jointly composed of the set of preprocessed feature vectors fused in all azimuths, the digital twin benchmark model parameters, and the error vector.

[0165] In the process of constructing the digital twin benchmark model and the jointly preprocessed dataset, historical data and real-time data are used. The model parameter matrix is solved by the least squares method and updated in real time. Sub-models are established for different azimuths and integrated. Finally, the jointly preprocessed dataset is integrated and output to provide a unified high-precision data input for subsequent anomaly detection and working condition optimization.

[0166] Step 3, Digital twin comparison and cavitation anomaly determination

[0167] The preprocessed original acoustic data is compared with the digital twin benchmark model in real time, and the typical cavitation signal frequency band (high-frequency pulse noise) is identified using a dynamically adjusted threshold. According to the expected characteristics under extreme water flow conditions (extremely low or extremely high water flow), the discrimination algorithm is adjusted to distinguish startup, load mutation, and cavitation impact signals to avoid misjudgment. By introducing a collector position compensation vector and a mapping matrix, an expected feature vector based on the position of each acoustic collector is constructed, enabling the model to give personalized expected values for acoustic collectors at different installation positions. The difference vector is adjusted using a correction factor to ensure the comparability of the acoustic signal deviations under extremely low and extremely high water flow conditions, thereby improving the accuracy of the anomaly score.

[0168] Specifically, it includes Steps 3.1 to 3.4:

[0169] Step 3.1, Construct an expected feature vector dependent on position.

[0170] Specifically, due to differences in installation height and orientation, the characteristics of the voiceprint signals expected to be collected by voiceprint collectors at different positions (such as frequency band energy distribution) vary. Based on the digital twin benchmark model, by combining real-time operating condition parameters with the position parameters of the voiceprint collectors, the expected feature vectors of each voiceprint collector are constructed. At the same time, the water flow rate has a direct impact on the output of the voiceprint collector. Therefore, it is necessary to incorporate the water flow rate as a mapping parameter into the model to construct a position-dependent mapping formula.

[0171] In this embodiment, a compensation vector is set for each voiceprint collector, which is jointly composed of the azimuth, vertical position, and local environmental factors (such as structural occlusion and openness) of the voiceprint collector. The expression of the position-dependent digital twin expected feature vector is:

[0172] F_expected,i=M_i*X+B_i*L_i

[0173] In the formula, F_expected,i represents the expected feature vector of voiceprint collector i in n frequency bands; M_i represents the mapping matrix related to the operating condition parameter X, corresponding to voiceprint collector i, and each element of it represents the influence coefficient of the operating condition parameter on the expected signal feature; B_i is the position compensation matrix, and each element of it represents the contribution of the position compensation factor to the expected signal feature; L_i represents the pre-set position parameter of the voiceprint collector, and both the operating condition parameter X and L_i are represented as column vectors.

[0174] In the process of constructing the position-dependent expected feature vector, for each voiceprint collector, according to the historical data and preprocessed features obtained above, the mapping matrix and the position compensation matrix are obtained by least squares fitting, and the expected feature vector is calculated according to the real-time collected operating condition parameters and the pre-set position parameters of the voiceprint collector.

[0175] Step 3.2, Difference vector calculation and water flow state correction.

[0176] Specifically, the preprocessed feature vector actually collected by each voiceprint collector is compared with the corresponding expected feature vector to calculate the difference vector between the two. At the same time, according to the real-time change of the water flow rate, the calculated difference vector is corrected to ensure the comparability of the voiceprint signal deviation under low water flow state and high water flow state. In addition, in order to correct the influence of water flow, a correction factor can be defined to adjust the difference vector according to the change of the water flow rate.

[0177] Among them, the difference vector is the difference between the preprocessed feature vector actually collected by the voiceprint collector and the corresponding expected feature vector. The correction factor is jointly composed of the current real-time water flow rate, the reference water flow rate, and the water flow influence coefficient, and the water flow influence coefficient is determined by historical data.

[0178] In this embodiment, for each voiceprint collector, it is necessary to calculate the original difference vector by a calculator, calculate the correction factor according to the real-time water flow rate and the reference water flow rate, adjust the original difference vector to the corrected difference vector, and use the corrected difference vector as the basic index for the abnormality determination of each voiceprint collector.

[0179] Step 3.3, abnormal score calculation and voiceprint collector abnormality determination.

[0180] Specifically, based on the corrected difference vectors of each voiceprint collector obtained above, the L2 norm is used to calculate the abnormal score, and comprehensive determination is carried out in combination with the azimuth and height information of the voiceprint collector array. Among them, it is not only necessary to calculate the abnormal score of a single voiceprint collector, but also necessary to determine whether there is cavitation abnormality as a whole by using the weighted average method according to its different azimuths.

[0181] Among them, the abnormal score is jointly determined by the corrected difference vector of the voiceprint collector in the feature dimension and the number of feature dimensions.

[0182] In the process of abnormal score calculation and comprehensive abnormality determination of the voiceprint collector, calculate the abnormal score of each voiceprint collector, standardize each abnormal score based on the preset determination threshold determined according to historical data to obtain the standardized abnormal score. Then, perform weighted averaging on the standardized abnormal scores of each group of voiceprint collectors according to the azimuth information of the voiceprint collector to obtain the comprehensive abnormal score of this azimuth. Finally, according to the preset determination rule, if the comprehensive abnormal score exceeds the set threshold of 1, it is determined that there is an abnormality in this azimuth. If more than 50% of the azimuths are abnormal, it is globally determined as cavitation abnormality.

[0183] Step 3.4, abnormal determination feedback and adaptive update of the digital twin benchmark model.

[0184] Specifically, after the abnormal score calculation is completed, a cavitation abnormality determination report is generated according to the preset determination rule and fed back to the monitoring system. At the same time, the digital twin benchmark model is updated based on the abnormal data therein to achieve adaptive learning, so as to ensure that the model can continuously correct the deviation caused by the changes in the position of the voiceprint collector and the water flow state.

[0185] In this embodiment, for each voiceprint collector, the update amount is calculated according to the deviation between its actual measurement result and the model prediction result to update the model parameters corresponding to all voiceprint collectors. The content of the generated cavitation abnormality determination report includes the abnormal scores, determination conclusions, real-time working condition parameters, and recommended measures (such as adjusting the load and water flow speed) of each voiceprint collector and azimuth, and the updated model parameters are saved for use in the next training cycle.

[0186] Step 4, further determine the location of cavitation.

[0187] Specifically, it includes steps 4.1 to 4.4:

[0188] Step 4.1, collect voiceprint anomaly data and sensor position information.

[0189] Based on the anomaly scores of the denoised voiceprint data and combined with the spatial position information of each voiceprint collector, generate a basic data set for positioning, including an anomaly score set and a sensor position set.

[0190] Specifically, first, construct an anomaly score set based on the comprehensive anomaly scores calculated in step 3.3. This anomaly score set is jointly composed of the anomaly score values of each voiceprint collector. Secondly, according to the spatial positions of each voiceprint collector in the digital twin reference model, construct a corresponding sensor position set. This position set is jointly composed of the position vectors corresponding to each voiceprint collector. The position vector of each voiceprint collector is composed of the X-axis, Y-axis, and Z-axis coordinates of this voiceprint collector.

[0191] In this example, it is necessary to ensure that the anomaly scores of all voiceprint collectors have undergone deep adaptive noise reduction processing, and all position information included in the position set has been corrected by on-site measurement and converted to a unified coordinate system.

[0192] Step 4.2, preliminarily calculate the anomaly centroid position.

[0193] Specifically, based on the anomaly score set and the sensor position set, calculate the centroid position of the abnormal voiceprint by weighted average as the preliminarily estimated cavitation occurrence position. First, define the preliminary anomaly centroid position through the weighted average formula, and then expand the preliminary anomaly centroid position to obtain the corresponding components, that is, the means of the anomaly centroid position on the X-axis, Y-axis, and Z-axis.

[0194] Step 4.3, introduce auxiliary physical parameters for correction.

[0195] Specifically, combine auxiliary physical parameters such as vibration and temperature to correct the preliminary anomaly centroid position obtained in step 4.2 to obtain a further refined cavitation position, and then use the comprehensive anomaly index to correct the contribution weights of each voiceprint collector to enhance the accuracy of cavitation positioning.

[0196] First, for each voiceprint collector, in addition to the voiceprint anomaly score, it is also necessary to obtain a vibration anomaly index for reflecting the vibration anomaly situation in the area where the voiceprint collector is located and a temperature anomaly index for reflecting the temperature anomaly situation in the area where the voiceprint collector is located. The above voiceprint anomaly score, vibration anomaly index, and temperature anomaly index are all monitored by vibration sensors and infrared thermography cameras installed at the corresponding positions and are converted into standard dimensionless indexes after preprocessing.

[0197] Secondly, in the process of defining the comprehensive anomaly index, the weight coefficients corresponding to the voiceprint anomaly score, vibration anomaly index, and temperature anomaly index are introduced, and the sum of the weight coefficients is 1. Therefore, the comprehensive anomaly index is the sum of the products of the voiceprint anomaly score, vibration anomaly index, and temperature anomaly index and their respective corresponding weight coefficients, which is used to reflect the comprehensive detection anomaly situation.

[0198] Finally, the above comprehensive anomaly index is used to weight the anomaly positions of each voiceprint collector to obtain the corrected cavitation position.

[0199] Step 4.4, on-site verification and feedback correction.

[0200] Specifically, the corrected cavitation position is verified using the actual on-site monitoring data, and the cavitation position is further corrected based on the on-site feedback to obtain the finally determined cavitation occurrence position.

[0201] First, the real-time data collected through on-site inspections, fixed monitoring cameras, and high-precision sensors is used to verify whether the cavitation anomaly area displayed on the digital twin platform is consistent with the actual situation. Then, a correction deviation vector is calculated by defining the reference position obtained from on-site detection, that is, the difference to be corrected between the actual cavitation position and the cavitation position on the digital twin platform. The correction deviation vector is added to the cavitation position obtained in Step 4.3 to further correct it and obtain the final cavitation occurrence position.

[0202] The present invention corrects the calculation result through on-site feedback data, similar to the differential GPS technology, which fine-tunes the measurement result through known reference points, and can ensure that the finally determined cavitation position is consistent with the actual situation.

[0203] Step 5, based on the finally determined cavitation position, readjust the positions of the 12 sensors.

[0204] The position where cavitation occurs is closely related to the equipment design and installation status, so it is usually relatively fixed in the short term. However, with the long-term operation, wear, maintenance of the equipment, and changes in environmental conditions, the cavitation area may experience slight displacement and expansion. Therefore, it is necessary to readjust the positions of the voiceprint collectors so as to pay closer attention to the cavitation state and obtain more accurate results.

[0205] Specifically, it includes Steps 5.1 to 5.4:

[0206] Step 5.1, data preprocessing and voiceprint anomaly distribution statistics.

[0207] Based on the voiceprint anomaly scores obtained in Step 3 and combined with the finally determined cavitation position in Step 4, the relative distribution of each voiceprint collector and its sensor position is statistically analyzed to provide a basis for calculating the distance between voiceprint collectors in the follow-up.

[0208] Specifically, first, according to the above abnormal score set, sensor position set, and the finally determined cavitation position, calculate the deviation vector of each acoustic fingerprint collector relative to the finally determined cavitation position. Then, use the acoustic fingerprint abnormal score as the weight to calculate its weighted standard deviation in each position and direction (X-axis, Y-axis, Z-axis), which is used as an index to measure the width of the abnormal data distribution. Among them, in the process of calculating the weighted standard deviation of the acoustic fingerprint abnormal score in each position and direction, it is necessary to first calculate its weighted average deviation in each position and direction, and then calculate the weighted standard deviation based on its weighted average deviation. The weighted standard deviations in each position and direction (X-axis, Y-axis, Z-axis) respectively reflect the distribution and diffusion degree of the acoustic fingerprint abnormality in each direction, and the more dispersed the acoustic fingerprint abnormal data distribution is, the higher the demand for acoustic fingerprint monitoring in that direction.

[0209] Step 5.2, determine the calculation of the vertical spacing within the acoustic fingerprint collector group.

[0210] Based on the standard deviation of the acoustic fingerprint abnormality distribution in the vertical direction obtained in Step 5.1 and the on-site monitoring requirements, determine the optimal vertical spacing within the acoustic fingerprint collector group.

[0211] Specifically, first, obtain the interval between adjacent acoustic fingerprint collectors in the same group in the vertical direction, and construct a vertical spacing adjustment formula by combining the vertical diffusion standard deviation of the acoustic fingerprint abnormality distribution and the on-site monitoring accuracy requirements, that is, the adjusted vertical spacing is the product of the vertical adjustment coefficient (obtained according to historical data and on-site preset adjustment) and the standard deviation of the abnormal distribution in the vertical direction. If the acoustic fingerprint abnormality is more dispersed in the vertical direction (i.e., the standard deviation is larger), a larger vertical spacing is required to ensure that each layer of acoustic fingerprint collectors covers different heights. Conversely, reduce the vertical spacing of the acoustic fingerprint collectors to improve the local acoustic fingerprint monitoring accuracy.

[0212] Step 5.3, determine the horizontal spacing between acoustic fingerprint collector groups.

[0213] Based on the standard deviation of the abnormal distribution of the acoustic fingerprint collectors in the horizontal direction obtained in Step 5.1, and combined with the characteristics of the cavitation area, determine the optimal horizontal spacing between each acoustic fingerprint collector group (at the same height).

[0214] Specifically, first determine the original horizontal spacing between different acoustic fingerprint collector groups in the horizontal direction, and then consider the abnormal distribution of the acoustic fingerprint in the horizontal direction, and define a comprehensive horizontal diffusion index by combining the weighted standard deviations in the X-axis and Y-axis directions respectively. Therefore, the adjustment formula for the horizontal spacing is the product of the comprehensive horizontal diffusion index and the horizontal adjustment coefficient (obtained according to the on-site situation preset).

[0215] Step 5.4, comprehensive scheme verification and final parameter output.

[0216] Integrate the optimal vertical spacing and the optimal horizontal spacing obtained in Steps 5.2 and 5.3 respectively to obtain the total calculation result, and then verify whether the total calculation result meets the on-site uniform coverage requirement, and finally output the finally optimized horizontal adjustment spacing and vertical adjustment spacing.

[0217] Specifically, compare the horizontal adjustment spacing and the vertical adjustment spacing with the finally determined cavitation position and the standard deviation of abnormal acoustic patterns to check whether each group of acoustic pattern collectors can cover the key area under the theoretical layout. For example, check whether the horizontal distance between the centers of each group of acoustic pattern collectors is satisfied theoretically, that is, calculate whether the distances from the acoustic pattern collectors at the same height to their respective center positions are half of the distance between them.

[0218] In this example, the spacing adjustment scheme of the newly arranged acoustic pattern collectors can also be simulated by using the horizontal adjustment spacing and the vertical adjustment spacing obtained by the above calculation to ensure that each group of acoustic pattern collectors is evenly distributed in the vertical and horizontal directions, and the positions of the acoustic pattern collectors meet the design requirements. When simulating, the method of calculating the new positions (i.e., Step 5.3) can be adopted, and it is checked whether the distances from the calculated new positions to the cavitation center are within the preset range. After verification, the finally optimized horizontal adjustment spacing and vertical adjustment spacing can be determined as the finally optimized parameters and recorded as the basis for the subsequent reinstallation of the acoustic pattern collectors and the system update.

[0219] In the description of the present invention, the terms "first" and "second" are only used for descriptive purposes and cannot be construed as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, the features defined with "first" and "second" may explicitly or implicitly include one or more of such features. The meaning of "a plurality" is two or more unless otherwise specifically defined.

[0220] In the present invention, unless otherwise clearly specified and limited, the terms such as "installation", "connection", "connection", "fixation" and the like should be understood in a broad sense. For example, it may be a fixed connection, a detachable connection, or integrated; it may be a mechanical connection or an electrical connection; it may be directly connected or indirectly connected through an intermediate medium, and it may be the internal communication of two components or the interaction relationship between two components. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific situations.

[0221] In the present invention, unless otherwise clearly defined and limited, the first feature being "on" or "under" the second feature may mean that the first and second features are in direct contact, or the first and second features are indirectly in contact through an intermediate medium. Moreover, the first feature being "above", "over" and "on top of" the second feature may mean that the first feature is directly above or obliquely above the second feature, or simply means that the horizontal height of the first feature is higher than that of the second feature. The first feature being "under", "below" and "beneath" the second feature may mean that the first feature is directly below or obliquely below the second feature, or simply means that the horizontal height of the first feature is less than that of the second feature.

[0222] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples", etc. means that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not have to be directed to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, without contradiction, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of different embodiments or examples.

[0223] Any process or method description shown in a flowchart or described in other ways herein may be understood to represent a module, segment, or portion of code including one or more executable instructions for implementing a specific logical function or process, and the scope of the preferred embodiments of the present invention includes additional implementations, where the functions may be executed in a manner that is not shown or discussed in the order, including in a substantially simultaneous manner according to the functions involved or in a reverse order, which should be understood by those skilled in the art to which the embodiments of the present invention pertain.

[0224] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a definable sequence list of executable instructions for implementing logical functions, which can be specifically implemented in any computer-readable medium for use by an instruction execution system, apparatus, or device (such as a computer-based system, a system including a processor, or other systems that can fetch and execute instructions from the instruction execution system, apparatus, or device), or used in combination with these instruction execution systems, apparatuses, or devices. For the purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of computer-readable media include the following: an electrical connection portion with one or more wirings (electronic device), a portable computer diskette (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). Additionally, the computer-readable medium can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other media, followed by editing, interpretation, or other suitable processing as necessary, and then stored in a computer memory.

[0225] It should be understood that various parts of the present invention can be implemented by hardware, software, firmware, or a combination thereof. In the above-described embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, any one or a combination of the following techniques well known in the art can be used: discrete logic circuits having logic gate circuits for implementing logical functions on data signals, application specific integrated circuits having appropriate combinational logic gate circuits, programmable gate arrays (PGAs), field programmable gate arrays (FPGAs), etc.

[0226] Those of ordinary skill in the art of this technology can understand that all or part of the steps carried by the method of the above embodiments can be completed by a program instructing relevant hardware, and the program can be stored in a computer-readable storage medium. When the program is executed, it includes one or a combination of the steps of the method embodiments.

[0227] In addition, each functional unit in various embodiments of the present invention may be integrated into one processing module, may exist separately as individual units physically, or two or more units may be integrated into one module. The above-mentioned integrated module may be implemented in the form of hardware, or may be implemented in the form of a software functional module. When the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it may also be stored in a computer-readable storage medium.

[0228] Although the embodiments of the present invention have been shown and described above, it can be understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those of ordinary skill in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of the present invention.

Claims

1. A water turbine unit operation system based on digital twin, characterized in that, The system includes the top cover of the water turbine unit, and multiple groups of voiceprint acquisition arrays are arranged on the top cover of the water turbine unit. Each group of voiceprint acquisition arrays includes multiple voiceprint collectors arranged in the vertical direction and having a fixed spacing; The voiceprint collectors are used to collect the original voiceprint data at different positions on the top cover of the water turbine unit and send the collected original voiceprint data to the data center; After receiving the original voiceprint data, the data center preprocesses the original voiceprint data, and establishes a digital twin reference model based on the mapping relationship between real-time operating condition parameters and voiceprint signal characteristics through digital twin technology, and constructs an expected feature vector based on the position compensation vector and mapping matrix of the voiceprint collector, and adjusts the difference vector using a correction factor; Among them, the actual operating condition parameters include real-time water flow rate, water flow pressure, and water flow temperature, and the difference vector is calculated based on the actual feature vector after preprocessing the original voiceprint data actually collected by the voiceprint collector and the expected feature vector.

2. A method for operating a water turbine unit based on digital twin, characterized in that, Implemented by the digital twin-based water turbine unit operation system according to claim 1, the method includes: Obtain multi-directional original voiceprint data and preprocess the original voiceprint data; Establish a real-time operation model of the water turbine unit through digital twin technology, and generate an expected spectrum model according to the structure of the water turbine unit, fluid dynamics, and historical voiceprint data; Construct an expected feature vector based on the position of the voiceprint collector based on the position compensation vector and mapping matrix of the collector, and adjust the difference vector using a correction factor; Among them, multiple groups of voiceprint acquisition arrays are arranged on the top cover of the water turbine unit. Each group of voiceprint acquisition arrays includes multiple voiceprint collectors arranged in the vertical direction and having a fixed spacing. The preprocessing includes differential preprocessing, time-frequency domain decomposition, adaptive filtering, and dynamic range adjustment.

3. The method for operating a water turbine unit based on digital twin according to claim 2, wherein, The obtaining of multi-directional original voiceprint data and the preprocessing of the original voiceprint data include: Determine the preliminary installation strategy of the voiceprint collectors on the top cover of the water turbine unit based on the preprocessed original voiceprint data and the digital twin reference model parameters. The preliminary installation strategy includes multiple groups of voiceprint acquisition arrays and the number, arrangement direction, and arrangement spacing of the voiceprint collectors on each group of voiceprint acquisition arrays; Calibrate the voiceprint collectors installed according to the preliminary installation strategy, and set the sampling frequency, sensitivity, and dynamic range. Adjust the amplitude of the voiceprint collectors using the real-time water flow rate to obtain correction parameters; According to the installation position and orientation of the voiceprint collectors on the top cover of the water turbine unit, combined with the correction parameters, adjust the amplitude normalization of the voiceprint collector signals using different compensation strategies based on the real-time water flow rate, and perform noise reduction and dynamic range expansion processing on the voiceprint signals to obtain the original voiceprint data after differential preprocessing; Synchronize the time of the original voiceprint data after differential preprocessing with the real-time water flow data and other operating condition parameters and perform joint data fusion to form a joint data matrix; Among them, other operating condition parameters include water flow pressure and water flow temperature.

4. The operation method of the water turbine unit based on digital twin according to claim 3, characterized in that Establishing a real-time operation model of a hydraulic turbine unit through digital twin technology, and generating an expected spectrum model based on the structure of the hydraulic turbine unit, fluid dynamics, and historical acoustic fingerprint data, including: Segmenting the continuous acoustic fingerprint signals of each acoustic fingerprint collector according to a fixed time window to form discrete acoustic fingerprint signal segments, and calculating an adjustment factor for each acoustic fingerprint signal segment according to the real-time water flow rate, so as to perform gain adjustment on each acoustic fingerprint signal segment based on the adjustment factor to obtain a new acoustic fingerprint signal; Performing time-frequency domain decomposition on the new acoustic fingerprint signals of each acoustic fingerprint collector to extract the spectral characteristics of the new acoustic fingerprint signals, and performing differential noise filtering according to the positions of the acoustic fingerprint collectors; Extracting a time-frequency matrix through short-time Fourier transform, and obtaining the local characteristics of the new acoustic fingerprint signals in combination with wavelet transform.

5. The operating method of the water turbine unit based on digital twin according to claim 4, characterized in that, Establishing a real-time operation model of a hydraulic turbine unit through digital twin technology, and generating an expected spectrum model based on the structure of the hydraulic turbine unit, fluid dynamics, and historical acoustic fingerprint data, further including: Extracting key features based on the preprocessed video data of each acoustic fingerprint collector, performing differential fusion on the key features, and correcting the energy difference of the acoustic fingerprint signals at different positions by using the compensation information of the real-time water flow rate to obtain a fused feature vector; Constructing a digital twin reference model based on the fused feature vector and real-time operating conditions parameters to establish the mapping relationship between the real-time operating conditions parameters and the acoustic fingerprint signals, and constructing a joint preprocessed data set by integrating the preprocessed original acoustic fingerprint data and model parameters; Wherein, the key features include the energy of each frequency band and the local high-frequency pulse index.

6. The method for operating a water turbine unit based on digital twin according to claim 5, wherein Constructing an expected feature vector based on the acoustic fingerprint collector position by using the collector position compensation vector and the mapping matrix, and adjusting the difference vector by a correction factor, including: Based on the digital twin reference model, by combining the real-time operating conditions parameters and the acoustic fingerprint collector position parameters, constructing an expected feature vector for each acoustic fingerprint collector, and integrating the real-time water flow rate as a mapping parameter into the digital twin reference model to construct a position-dependent mapping formula; Comparing the actual feature vector corresponding to the preprocessed acoustic fingerprint data actually collected by each acoustic fingerprint collector with the expected feature vector to calculate the difference vector between the actual feature vector and the expected feature vector, and correcting the difference vector according to the change of the real-time water flow rate.

7. The operation method of the water turbine unit based on digital twin according to claim 6, characterized in that, Constructing an expected feature vector based on the acoustic fingerprint collector position by using the collector position compensation vector and the mapping matrix, and adjusting the difference vector by a correction factor, further including: Calculating the L2 norm based on the corrected difference vector as the anomaly score of the acoustic fingerprint collector, and combining the azimuth and height of the acoustic fingerprint collector array to determine whether there is cavitation anomaly in the hydraulic turbine unit by using weighted average; Generating a cavitation anomaly determination report based on the anomaly score according to a preset determination rule, and feeding back the cavitation anomaly determination report to the monitoring system, and at the same time updating the digital twin reference model parameters based on the abnormal acoustic fingerprint data.

8. The operation method of the water turbine unit based on digital twin according to claim 7, characterized in that, The method further includes: Generate a basic data set based on the anomaly score and the spatial location information of each voiceprint collector, perform weighted averaging on the basic data set to obtain the first cavitation position, and correct the components in each direction of the first cavitation position by combining auxiliary physical parameters to obtain the second cavitation position; Obtain the actual cavitation position, and when the actual cavitation position is inconsistent with the second cavitation position, calculate the first deviation vector between the actual cavitation position and the second cavitation position to correct the second cavitation position to obtain the third cavitation position; Calculate the second deviation vector between the third cavitation position and the corresponding voiceprint collector, and adjust the positions of each voiceprint bar sensor in the horizontal and vertical directions by combining the anomaly score and the current spatial location information of each voiceprint collector; Among them, the basic data set is used to locate the cavitation position, including an anomaly score data set and a sensor position data set, and the auxiliary physical parameters include a vibration anomaly index and a temperature anomaly index in the area where the voiceprint collector is located.

9. A terminal, comprising a processor and a storage medium; characterized in that: The storage medium is used to store instructions; The processor is used to operate according to the instructions to execute the steps of the method according to any one of claims 2-8.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, the steps of the method according to any one of claims 2-8 are implemented.

Images